Add general introduction and quick installation.#43
Conversation
Code Review ReportProject: BANK OF AI Documentation ( PR OverviewBranch Information
Commit History
Review SummaryVerdict
Findings at a Glance
SummaryThis PR introduces a substantial new "BANK OF AI" documentation section ( The primary concern is a systematic removal of "restart your AI client" instructions across multiple files ( Four Minor issues cover an example password visible in documentation output examples, unverified internal cross-reference links in new pages, a mildly misleading "100% offline" claim, and a subtle security-tone inconsistency in the wallet backup warnings. Three Suggestions cover adding a troubleshooting section for the new conversational install path, clarifying i18n mirror maintenance, and expanding the "What's Next" navigation for the BANK-OF-AI section. Change Summary1. New: BANK OF AI Documentation Section
Purpose: Introduce BANK OF AI as a top-level product concept with its own navigation section, giving users a single entry point before diving into individual components. 2. Updated: MCP Server Installation Guides (TRON & SUN)
Purpose: Standardise installation UX across all MCP guides; make conversational install the recommended default. 3. Updated: SKILLS Installation Guide
Purpose: Mirror the conversational-install pattern from MCP guides into the SKILLS guide for consistency. 4. Updated: Agent-Wallet Quick Start
Purpose: Allow users who already have BANK OF AI skills installed to create a wallet entirely through chat, without touching the terminal. 5. Infrastructure / Config
Purpose: Minor version bump reflecting new docs content; i18n files kept in sync with English source. Detailed FindingsMajor[MJ-01] Restart Guidance Removed Without Explanation — Users Will Silently Fail
Description
Code (removed from -After the command completes, restart your MCP client.
...
-## Restart and Test
-
-After saving the configuration, **completely quit and restart** your AI client.
-After restarting, the client will automatically connect to TRON MCP Server.
-
-Then enter your first query in the chat:
+## Test It Out
+
+Enter your first query in the chat:Code (removed from If you encounter issues, first confirm:
1. Node.js version >= 20.0.0 (run `node --version` to check)
2. Network can reach `tron-mcp-server.bankofai.io`
-3. AI client has been fully quit and restarted (not just refreshed)Recommendation Either:
Example for If you encounter issues, first confirm:
1. Node.js version >= 20.0.0 (run `node --version` to check)
2. Network can reach `tron-mcp-server.bankofai.io`
3. If you used Method 2 or 3 (CLI install): fully quit and restart your AI client
(config changes are only picked up at startup, not via refresh)Minor[MN-01] Example Output Shows a Plausible-Looking Real Password
Description
Code 🔐 Wallet password
$mU80m9fLYB5Recommendation Replace with an obviously fake placeholder that no user would mistake for a valid password: 🔐 Wallet password
<your-auto-generated-password-appears-here>Or use a clearly synthetic format: $ExAmPlE-PaSsWoRd-D0-n0t-Use![MN-02] New Pages Contain Unverified Internal Cross-Reference Links
Description
Links at risk (not validated in diff): [LLM Service Introduction](../llmservice/introduction.md)
[Skills Introduction](../McpServer-Skills/SKILLS/Intro.md)
[Agent-Wallet Introduction](../Agent-Wallet/Intro.md)
[x402 Protocol Introduction](../x402/index.md)
[8004 Protocol Introduction](../8004/general.md)
[MCP Server Introduction](../McpServer-Skills/MCP/Intro.md)
[SUN CLI Introduction](../McpServer-Skills/Tools/SUNCli/Intro.md)
[TRON MCP — Local Private Deployment](../McpServer-Skills/MCP/TRONMCPServer/LocalPrivatizedDeployment.md)
[SUN MCP — Local Private Deployment](../McpServer-Skills/MCP/SUNMCPServer/LocalPrivatizedDeployment.md)Recommendation Run Docusaurus's built-in broken-link checker before merge: yarn build 2>&1 | grep -i "broken\|404\|not found"Or run [MN-03] "100% Offline" Signing Claim Is Misleading
Description
Code - **Local Signing**: all signing happens entirely on your machine — **100% offline** —
your private key never leaves your computerRecommendation Qualify the claim to describe what is actually offline — the signing step, not the broadcast: - **Local Signing**: all transaction signing happens entirely on your machine — your
private key never leaves your computer. Only the signed payload is transmitted to the network.[MN-04] Inconsistent Security Tone in Password Backup Warnings
Description
Code :::caution Back up your wallet password!
Even though the password is auto-saved to `~/.agent-wallet/runtime_secrets.json`,
**you still must copy it separately** into a password manager ...
:::
:::tip Done — no env-var configuration needed
... You don't need to manually export `AGENT_WALLET_PASSWORD`.
:::Recommendation Reorder the tip to appear first (installation convenience), then the caution (backup urgency), so the caution is the last thing users read before moving on: :::tip Done — no env-var configuration needed
Conversational setup auto-saves the password to `~/.agent-wallet/runtime_secrets.json`,
which your AI Agent reads automatically. You don't need to manually export `AGENT_WALLET_PASSWORD`.
:::
:::caution Back up your wallet password!
The auto-saved file is convenient, but it is not a backup.
If the file is deleted, your disk fails, or you reinstall your OS, **you permanently lose wallet access**.
Copy the password to a password manager (1Password, Bitwarden, etc.) right now.
:::Suggestions[S-01] No Troubleshooting Section for Conversational Install MethodFiles: Description: The new Method 1 (conversational install) has no troubleshooting guidance. The CLI methods each link to detailed OfficialServerAccess pages with troubleshooting checklists. Conversational install introduces new failure modes: the AI client may not support shell command execution, Node.js might be missing on the machine hosting the agent, or the AI might misinterpret the pasted prompt. A short Suggestion: Add a collapsible "Troubleshooting conversational install" section covering:
[S-02] i18n Mirror Files Should Be Noted as Auto-GeneratedFiles: Description: The diff includes 10 mirrored i18n files that are byte-for-byte copies of the English source (they appear to be the same content, not yet translated). This is standard Docusaurus i18n scaffold behaviour. However, if future contributors aren't aware that these mirrors exist, they may update the English docs and forget to sync the i18n copies, leading to stale translated content. A brief note in Suggestion: Add a CI step using [S-03] BANK-OF-AI QuickStart Missing Verification Link After "agent install everything"File: Description: The QuickStart tells users to ask the AI Suggestion: After the verification step, add: If the AI can't list any skills, see [SKILLS Quick Start — Troubleshooting](../McpServer-Skills/SKILLS/QuickStart.md#troubleshooting)
or [TRON MCP Troubleshooting](../McpServer-Skills/MCP/TRONMCPServer/OfficialServerAccess.md#troubleshooting).Positive Observations
Checklist Results
DisclaimerThis is an automated code review. It supplements but does not replace human review. The reviewer analysed only the diff between Report generated by Code Review Skill v1.0.0 |
Code Review ReportProject: BANK OF AI Documentation Site ( PR OverviewBranch Information
Commit HistoryReview SummaryVerdict
Findings at a Glance
Summary
Change Summary1. New BANK OF AI Section (docs/BANK-OF-AI/)Two entirely new files:
2. MCP Server Installation Guides RestructuredBoth 3. SKILLS QuickStart Restructured
4. Agent-Wallet QuickStart Restructured
5. Navigation and Configuration
Detailed FindingsMajor FindingsFinding M-1: "Restart your AI client" instruction silently removed from MCP Server installation guides
Description: Additionally, the
This bullet was removed entirely. MCP clients must restart to load a newly added MCP server into their context. Removing the restart instruction is a functional regression that will cause users to believe the installation failed when the server is simply not yet loaded. Diff evidence (SUNMCPServer/OfficialServerAccess.md): -Once installation is complete, restart your AI Agent, and you can start interacting with SunSwap via SUN MCP Server.
+Once installation is complete, you can start interacting with SunSwap via SUN MCP Server.-3. AI client has been fully quit and restarted (not just refreshed)Recommendation: Finding M-2: Sample output in Agent-Wallet QuickStart contains an exposed, realistic-looking password
Description: Even if this is an illustrative example, presenting a specific password string (rather than a clearly masked placeholder like Diff evidence: +🔐 Wallet password
+
+$mU80m9fLYB5Recommendation: This is consistent with how Finding M-3:
|
| Product | Provider | Capabilities | Install |
|---|---|---|---|
| Etherscan | Etherscan Official | Query / Tracking / Analytics, 60+ chains | npx skills add https://docs.etherscan.io |
The install command references https://docs.etherscan.io, which is the Etherscan documentation site, not a GitHub repository or npm package. npx skills add is expected to accept a GitHub user/repo slug or a GitHub URL. An HTTPS URL pointing to a documentation site is almost certainly incorrect and will fail for any user who tries to run it.
Recommendation:
Verify and replace with the correct install reference for the Etherscan skill, whether that is a GitHub repo slug (etherscan/etherscan-skills) or the correct URL. If the skill is not yet publicly available, remove or annotate the install command accordingly.
Finding m-5: SKILLS QuickStart — Method 2 description says "tell your AI Agent to execute" but this is a terminal command, inconsistent with the method's intent
- Severity: Minor
- Category: Docs / Clarity
- File:
docs/McpServer-Skills/SKILLS/QuickStart.md
Description:
Method 2 is titled "Quick Auto-Install (Command Line)", implying the user runs this in a terminal. However, its description says:
If you have Node.js installed and prefer the command line, simply tell your AI Agent to execute the following command:
Instructing the user to "tell your AI Agent to execute" a terminal command conflates the conversational-install flow (Method 1) with the direct CLI flow (Method 2). For Method 2, the user should run the command themselves in a terminal.
Recommendation:
Revise the description to read: "If you have Node.js installed and prefer the command line, run the following directly in your terminal:" — consistent with how the same method is described in SUNMCPServer/OfficialServerAccess.md and TRONMCPServer/OfficialServerAccess.md.
Suggestions
Suggestion S-1: Consider adding a "Windows Users" tab in Agent-Wallet QuickStart Method 2
- Severity: Suggestion
- Category: Docs / UX
- File:
docs/Agent-Wallet/QuickStart.md
Description:
The "2.1 Save and Activate the Password" section provides tabs for "Mac Users (Zsh)" and "Linux Users" but not for Windows users. Given the conversational-install method (Method 1) is now front-and-center for non-technical users, the CLI method targeting more technical users should still cover Windows to be complete.
Recommendation:
Add a Windows tab with $env:AGENT_WALLET_PASSWORD = 'your-password' (PowerShell) or instructions for setting environment variables via System Properties. Alternatively, note explicitly that Method 2 currently only supports macOS/Linux.
Suggestion S-2: The package.json version bump is not accompanied by a CHANGELOG entry
- Severity: Suggestion
- Category: Docs / Maintainability
- File:
package.json
Description:
The version was bumped from 1.2.8 to 1.2.9. No CHANGELOG.md file exists in the repo (and none is expected given this is a docs-only project), but the commit messages are generic ("add bankofai", "Add general introduction and quick installation."). For a documentation site, a brief entry noting what changed in the version is helpful for release tracking.
Recommendation:
Consider adding a CHANGELOG.md or using git tags to mark this release, or at minimum using more descriptive commit messages.
Suggestion S-3: Architecture SVG file uses inline <foreignObject> with HTML — verify cross-browser compatibility
- Severity: Suggestion
- Category: Quality
- File:
docs/BANK-OF-AI/image/bankofai-architecture.svg
Description:
The newly added SVG architecture diagram uses <foreignObject> elements to embed HTML content. <foreignObject> rendering in SVG is inconsistently supported across browsers and SVG renderers (e.g., it can fail in Safari, certain PDF exporters, and social media link previews). Docusaurus renders this inline or as an <img> tag depending on usage, and the <img> tag path will not render <foreignObject> HTML.
Recommendation:
Verify the SVG renders correctly in all major browsers and in the Docusaurus build output. If cross-browser consistency is needed, consider generating a rasterized PNG version of the diagram as a fallback.
Suggestion S-4: No i18n Chinese translations provided for the new BANK-OF-AI content
- Severity: Suggestion
- Category: Docs / Maintainability
- Files:
i18n/zh-Hans/docusaurus-plugin-content-docs/current/BANK-OF-AI/Intro.mdi18n/zh-Hans/docusaurus-plugin-content-docs/current/BANK-OF-AI/QuickStart.md
Description:
The i18n mirror files for BANK-OF-AI under i18n/zh-Hans/docusaurus-plugin-content-docs/current/BANK-OF-AI/ appear to be copies of the English originals (based on the diff showing the same content added to the i18n directory). If the Chinese locale is actively maintained, these files should be translated. If they are intentionally English-only stubs, a comment should indicate this.
Recommendation:
Either translate the content or add a front-matter comment noting the file is a stub pending translation.
Positive Observations
-
Well-structured tiered installation approach. Presenting "Conversational Install → Quick CLI → Interactive CLI" as three distinct methods with clear labels ("Easiest", "Command Line", "Most Control") is an excellent progressive disclosure pattern. It makes the documentation accessible to both non-technical users and developers.
-
Strong security guidance in Agent-Wallet documentation. The caution blocks around password backup, the explicit warning about not using double quotes for environment variables with special characters, and the explanation of why
echoto shell history is dangerous are all accurate, well-placed, and valuable guidance for users dealing with crypto wallet credentials. -
Consistent structure across MCP server documentation. The TRON MCP Server and SUN MCP Server installation guides now share an identical three-method structure with consistent phrasing, making it easy for users to transfer knowledge between sections.
-
Comprehensive BANK OF AI introduction document.
BANK-OF-AI/Intro.mdprovides a thorough layered overview of the product stack with working cross-references to all sub-product documentation. The "Is BANK OF AI for Me?" section with audience-specific messaging is a good documentation practice. -
Version bump is correctly scoped. The
package.jsonversion bump from1.2.8to1.2.9is appropriate for a documentation-only content addition. -
i18n files are kept in sync. All changed English documentation files have corresponding changes in the
i18n/zh-Hans/mirror directory, demonstrating good i18n hygiene.
Checklist Results
| Category | Check | Result | Notes |
|---|---|---|---|
| Correctness | Logic/flow correct | FAIL | Restart instructions removed from MCP install guides (M-1) |
| Correctness | Cross-references valid | PARTIAL | Etherscan install command likely invalid (m-4); heading anchors may break (m-3) |
| Correctness | Claude Code sidebar preserved | FAIL | Claude Code category missing in updated sidebars.js (M-3) |
| Security | No credentials/secrets exposed | PARTIAL | Realistic password in sample output (M-2) |
| Security | Crypto wallet guidance is accurate | PASS | Strong guidance throughout |
| Performance | No unbounded operations | N/A | Documentation-only PR |
| Quality | Consistent terminology | PARTIAL | Instruction prompts differ between pages (m-2); Method 2 framing inconsistent (m-5) |
| Quality | Heading hierarchy correct | PARTIAL | Minor inconsistency in Agent-Wallet QuickStart (m-3) |
| Testing | Build/render tested | UNKNOWN | SVG foreignObject compatibility not verified (S-3) |
| Docs | i18n entries added | PARTIAL | BANK OF AI label untranslated in zh-Hans (m-1); new content files are English stubs (S-4) |
| Docs | All new sidebar entries registered | PARTIAL | New BANK OF AI section registered; Claude Code category accidentally dropped (M-3) |
Disclaimer
This is an automated code review generated by AI Code Reviewer. All findings should be validated by a human engineer before action is taken. Severity ratings are advisory. Security-related findings in particular should be reviewed by a qualified security engineer.
Code Review ReportProject: BANK OF AI Documentation ( PR OverviewBranch Information
Commit History
Review SummaryVerdict
Findings at a Glance
SummaryThis PR is a documentation-only release for a Docusaurus-based docs site. It introduces the new "BANK OF AI" product section (Intro + QuickStart pages) in both English and Chinese, retrofits all existing MCP Server and Skills install guides with a new "conversational install" method (Method 1), and bumps the package version from 1.2.8 to 1.2.9. The overall quality is high — the content is well-structured, bilingual parity is maintained, and the new conversational install flow is clearly explained with example dialogs. Two major issues require attention before merge: a realistic-looking wallet password and a real-format wallet address are embedded in documentation example output without sufficient "sample only" labeling, and the deletion of the "restart AI client" troubleshooting step may leave users stranded when the conversational install path (which eliminates the restart) doesn't apply to their setup. Several minor issues cover documentation inconsistencies, an incorrect ecosystem install command, and SVG accessibility gaps. Change Summary1. New BANK-OF-AI Section (Docs + i18n)
Purpose: Introduces the BANK OF AI brand umbrella across its 4-layer architecture (Model / Protocol / Tool / Ecosystem), with a two-step "paste and reply A" quick-start that installs all Skills and auto-creates a wallet. 2. "Conversational Install" Added to Existing Guides
Purpose: Promotes AI-driven installation (the user pastes a prompt into their AI client and the AI runs 3. Agent-Wallet QuickStart Restructured
Purpose: The original three-step CLI-only wallet setup is now "Method 2: Command-Line Setup." A new "Method 1: Conversational Setup" is prepended for users who already have BANK OF AI Skills installed. 4. Sidebar & Navigation
Purpose: Registers the new section in the doc navigation. 5. Assets & Version
Detailed FindingsMajor[MJ-01] Realistic Wallet Password Embedded in Documentation Without Clear "Sample Only" Disclaimer
Description The "Method 1: Conversational Setup" section embeds a sample AI output block that contains a realistic-looking wallet password: While this is meant to illustrate example output, there is no inline label (e.g., "example password — yours will differ") on or near the password value itself. The surrounding caution admonition is placed after the code block and refers to backing up "your" password, which could be read as referring to the Code ```text
🔐 Wallet password
$mU80m9fLYB5It also removes "restart" language from the Quick Start "Restart and Test" sections (renamed to "Test It Out"), dropping this instruction entirely: -After completing the configuration, restart your MCP client.
-After saving the configuration, **completely quit and restart** your AI client.The rationale for these removals is that the new conversational install path (Method 1) does not require a restart. However, Methods 2 and 3 (CLI and interactive) write to the MCP config file on disk — most MCP-supporting AI clients (Claude Desktop, OpenClaw, Cursor) do require a restart to pick up config-file changes. By removing the restart instruction from both the quick-start and troubleshooting sections without scoping it to the conversational install path, users following the CLI methods will be confused when their installed server does not appear. Code -3. AI client has been fully quit and restarted (not just refreshed)and -## Restart and Test
-After completing the configuration, restart your MCP client. Then try the following query in a conversation:
+## Test It Out
+Try the following query in a conversation:Recommendation Re-add the restart note scoped to Methods 2 and 3: ## Test It Out
> **Method 1 (conversational install):** The AI handles reload automatically — no restart needed.
> **Methods 2 & 3 (CLI / interactive):** After installation completes, **completely quit and restart** your AI client so it picks up the updated MCP config.
Then try the following query…Restore troubleshooting step 3 (or annotate it as applicable only to CLI/interactive install): 3. (CLI/interactive install only) AI client has been fully quit and restarted — not just refreshed.Minor[MN-01] Incorrect Ecosystem Install Command for Etherscan
Description The Ecosystem Layer table lists Etherscan's install command as:
Recommendation Verify the correct install command with the Etherscan team and update the table. If an official skills package does not yet exist, replace the entry with [MN-02] Full Real-Looking Wallet Addresses Embedded in EN Agent-Wallet QuickStart
Description The English Agent-Wallet QuickStart displays full wallet addresses in the example output: These are valid-format blockchain addresses. If they belong to a real wallet (e.g., a developer's wallet), publishing them in public documentation exposes that wallet to monitoring and potential social engineering. The corresponding Chinese file and the Recommendation Replace the full addresses with truncated, clearly illustrative forms consistent with the other pages: | EVM | `0x1339...8366` |
| TRON | `TBisA...LceY` |[MN-03] Inconsistency: "Conversational Install" for SKILLS Uses Shorthand Repo Reference
Description The conversational install prompt for SKILLS uses the shorthand: While the CLI Method 2 on the same page uses the full URL: npx skills add https://github.com/BofAI/skills -y -gIf the Recommendation Validate that [MN-04]
|
| Property | Value |
|---|---|
| Severity | Minor |
| Category | Code Quality / UX |
| File | sidebars.js : Lines 4–12; i18n/zh-Hans/.../sidebars.js : Lines 4–12 |
Description
The new BANK-OF-AI sidebar category is added with collapsed: false:
{
type: 'category',
label: 'BANK OF AI',
collapsed: false,
items: [
'BANK-OF-AI/Intro',
'BANK-OF-AI/QuickStart',
],
},This forces the category to be permanently expanded for all visitors, regardless of whether they are interested in this section. No other category in the existing sidebars uses collapsed: false. As the docs grow and more sections are added, having one category pinned open can make the sidebar feel unbalanced.
Recommendation
Remove collapsed: false (or change to collapsed: true) to be consistent with the rest of the sidebar configuration. If the intent is to highlight the new section, consider using Docusaurus's customProps or a doc badge instead.
[MN-05] "Telegram Bot" Listed as Shell-Command-Capable AI Client Without Qualification
| Property | Value |
|---|---|
| Severity | Minor |
| Category | Documentation / Accuracy |
| File | docs/McpServer-Skills/SKILLS/QuickStart.md : Line 1045; docs/McpServer-Skills/MCP/SUNMCPServer/OfficialServerAccess.md : Line 780; docs/McpServer-Skills/MCP/TRONMCPServer/OfficialServerAccess.md : Line 911; docs/Agent-Wallet/QuickStart.md : Line 26; and Chinese counterparts |
Description
Throughout the new content, "a Telegram bot" is listed alongside OpenClaw, Claude Code, and Cursor as an AI agent that supports running shell commands:
(OpenClaw, a Telegram bot, web chat, Claude Code, Cursor, etc.)
Standard Telegram bots cannot execute shell commands — they operate within Telegram's bot API and do not have access to the host machine's shell. This is presumably a reference to a specific BofAI-built Telegram bot integration that exposes shell execution, but this is not clarified anywhere in the text. Readers with a generic Telegram bot will attempt the conversational install and fail.
Recommendation
Replace the vague "a Telegram bot" with a specific reference, e.g., "the BANK OF AI Telegram bot," and link to its setup page. Alternatively, add a footnote clarifying that only AI clients with shell execution capability are supported.
Suggestions
[S-01] SVG Architecture Diagram Lacks Accessibility Attributes
File: docs/BANK-OF-AI/image/bankofai-architecture.svg (and Chinese counterpart)
Description: The architecture SVG diagram contains no <title>, role="img", or aria-labelledby attributes. Screen readers will either skip or poorly announce the diagram. The text content embedded in the SVG is not selectable in some renderers.
Suggestion: Add SVG accessibility metadata:
<svg … role="img" aria-labelledby="svgTitle svgDesc">
<title id="svgTitle">BANK OF AI Architecture — How Layers Work Together</title>
<desc id="svgDesc">Flow diagram showing a TRC20 transfer moving through: User → AI Agent → Skills (trc20-token-toolkit) → TRON MCP Server → Agent-Wallet → TRON Blockchain, with a dashed result-feedback path returning to the AI Agent.</desc>
…[S-02] Example AI Output in SKILLS QuickStart Uses an OpenClaw-Specific Path
File: docs/McpServer-Skills/SKILLS/QuickStart.md : Lines 1085, 1103
Description: The example AI output hardcodes an OpenClaw-specific installation path:
Location: ~/.openclaw/workspace/.agents/skills/
While OpenClaw is one supported client, readers using Claude Code, Cursor, or other clients will see a path that does not match their environment, which may cause confusion about whether the install succeeded.
Suggestion: Add a parenthetical clarifying the path is client-specific:
Location: ~/.openclaw/workspace/.agents/skills/ ← path varies by AI client
[S-03] BANK-OF-AI/Intro.md BSC MCP Server Install Command Uses Standalone npx Form
File: docs/BANK-OF-AI/Intro.md : Ecosystem Layer table
Description: The BSC MCP Server is listed without an install command in the Ecosystem table (it only says "BSC MCP Server — atomic on-chain operations for BNB Chain") in the product matrix, but the Ecosystem section shows TRON and SUN MCP Servers with npx -y @bankofai/mcp-server-tron style commands while the BSC entry is absent from the ecosystem table entirely. Users who want BSC support have no install path shown at the main overview level.
Suggestion: Either add the BSC MCP Server to the Ecosystem Layer table with its install command, or add a note like (install guide coming soon) so users know to refer to the dedicated BSC MCP docs.
[S-04] Version Bump in package.json Not Reflected in a CHANGELOG
File: package.json : Line 4
Description: The version is bumped from 1.2.8 to 1.2.9 but no CHANGELOG.md or release notes entry is updated or created for this change. For a published documentation package, it is good practice to log what changed between versions.
Suggestion: Add or update a CHANGELOG.md entry summarising the changes in v1.2.9 (new BANK OF AI section, conversational install across all guides, version bump), or document the release notes in the PR description.
Positive Observations
| Area | Observation |
|---|---|
| Bilingual parity | Every English content change has a faithful, well-translated Chinese counterpart. Terminology is consistent throughout both locales. |
| Security messaging | Password backup warnings (:::caution, :::danger) are used consistently and prominently wherever a wallet password is involved. The "never paste into email/chat/screenshots" warning is explicitly stated. |
| Progressive disclosure | The new three-method installation pattern (Conversational → CLI → Interactive) cleanly tiered by user expertise level is well-structured and follows good UX documentation practice. |
| Admonition usage | Docusaurus admonitions (:::tip, :::caution, :::danger, :::info, :::note) are used appropriately throughout — tips for prerequisites, cautions for irreversible actions, notes for edge-case expectations. |
| Architecture diagram | The SVG architecture diagram is self-contained (no external dependencies), uses a consistent design language, and clearly illustrates the data flow for a real-world operation. |
| Example dialogs | The <details>-collapsed example AI outputs give users clear expectations of what success looks like without cluttering the main reading path. |
| Internal cross-linking | New pages cross-link to related sections (CLI Reference, MCP Server deployment guides, x402, 8004, LLM Service) thoroughly, reducing documentation dead ends. |
Checklist Results
| Category | Items Checked | Pass | Fail | N/A | Notes |
|---|---|---|---|---|---|
| Correctness | 4 | 2 | 2 | 2 | Incorrect Etherscan command; missing restart scoping |
| Security | 5 | 3 | 2 | 0 | Example password lacks clear placeholder; real addresses in docs |
| Performance | 3 | 3 | 0 | 0 | No performance concerns (docs only) |
| Code Quality | 5 | 3 | 2 | 0 | Sidebar collapse inconsistency; shorthand vs full URL inconsistency |
| Testing | 2 | 0 | 0 | 2 | No automated tests apply to documentation |
| Documentation | 6 | 4 | 2 | 0 | Telegram bot unqualified; BSC install path missing from overview |
| Compatibility | 3 | 3 | 0 | 0 | i18n parity maintained; sidebar structure valid |
| Observability | 2 | 2 | 0 | 0 | N/A for documentation |
Disclaimer
This is an automated code review. It supplements but does not replace human review. The reviewer analyzed only the diff between the specified branches. Runtime behavior, integration testing, and deployment impact are not covered. For documentation PRs, "runtime" concerns include link validity, image rendering, and Docusaurus build correctness — these should be verified via a preview deployment (e.g., npm run build or a staging site) before merge.
Report generated by Code Review Skill v1.0.0
Date: 2026-04-16
Code Review ReportProject: @x402-tron/docs (BANK OF AI Documentation Site) PR OverviewBranch Information
Commit History
Review SummaryVerdict
Findings at a Glance
Summary
Change SummaryNew BANK OF AI Section
Purpose: Introduces a new top-level product overview and entry point for the BANK OF AI ecosystem docs. Agent Wallet QuickStart Expansion
Purpose: Provides a no-terminal "conversational" setup path for users who have already installed the BANK OF AI skills suite. SKILLS QuickStart Expansion
Purpose: Mirrors the conversational-install approach used across MCP Server pages. MCP Server Quick-Start Updates
Purpose: Standardises all MCP Server installation guidance around the new Navigation and Version
Purpose: Wires new pages into site navigation and records the documentation version increment. Detailed FindingsMajor[M-01] Skill Count Inconsistency: "11 skills" vs "8 skills" in Same Page
Description
Code Recommendation
[M-02] Contradictory Quote-Style Instruction in Private Key Option
Description
Code export TRON_PRIVATE_KEY='your_real_or_testnet_private_key'Recommendation
[M-03] Example Output Contains a Realistic-Looking Password Value
Description
Code Recommendation
Minor[m-01] zh-Hans Sidebar Missing
|
| Property | Value |
|---|---|
| Severity | Minor |
| Category | Documentation Maintainability |
| File | i18n/zh-Hans/docusaurus-plugin-content-docs/current/sidebars.js : SKILLS section |
Description
The English
sidebars.jsincludes'McpServer-Skills/SKILLS/QuickStart'in the SKILLS navigation items. The zh-Hanssidebars.jsdoes not include this entry in the SKILLS section (pre-existing omission), and this PR does not correct it despite making other zh-Hans sidebar additions. Chinese locale users will not see a SKILLS QuickStart link in their sidebar navigation, even though the translated filei18n/zh-Hans/.../McpServer-Skills/SKILLS/QuickStart.mdexists and was updated in this PR.
Code
// English sidebars.js — includes QuickStart
items: [
'McpServer-Skills/SKILLS/Intro',
'McpServer-Skills/SKILLS/QuickStart', // <-- present
'McpServer-Skills/SKILLS/BANKOFAISkill',
'McpServer-Skills/SKILLS/Faq',
],
// zh-Hans sidebars.js — missing QuickStart
items: [
'McpServer-Skills/SKILLS/Intro',
// 'McpServer-Skills/SKILLS/QuickStart' <-- absent
'McpServer-Skills/SKILLS/BANKOFAISkill',
'McpServer-Skills/SKILLS/Faq',
],Recommendation
Add
'McpServer-Skills/SKILLS/QuickStart'to the SKILLS items array ini18n/zh-Hans/docusaurus-plugin-content-docs/current/sidebars.js.
[m-02] Mac-Specific Terminal Command in Cross-Platform Private Key Documentation
| Property | Value |
|---|---|
| Severity | Minor |
| Category | Documentation Correctness |
| File | docs/McpServer-Skills/SKILLS/QuickStart.md : Lines 275–283 |
Description
Option 2 (Paste Your Private Key Directly) instructs the user: "In Terminal (the black window), type
open -e ~/.zshrcand press Enter." Theopen -ecommand is macOS-specific and will fail on Linux. The section offers no platform-specific guidance. This contrasts with the Agent-Wallet QuickStart (Method 2, Step 2.1), which correctly uses<Tabs>to present Mac-specific (open -e ~/.zshrc) and Linux-specific (nano ~/.bashrc) instructions separately. Furthermore,.zshrcis the default shell config on macOS, but Linux users (who default to Bash) would need.bashrc.
Recommendation
Either apply the same tab-based, platform-specific pattern used in
docs/Agent-Wallet/QuickStart.md, or at minimum add a parenthetical note: "(Mac users:open -e ~/.zshrc; Linux users:nano ~/.bashrc)".
[m-03] Unclear Commit Message History
| Property | Value |
|---|---|
| Severity | Minor |
| Category | Code Quality / Maintainability |
| File | Git history |
Description
Three consecutive commits are all titled "add bankofai" (
a72b661,2248ef7,86022be). Commit messages provide no indication of what changed in each, making the history difficult to bisect or review granularly. This is especially important given the PR modifies 24 files and these three commits appear to contain the bulk of the new BANK-OF-AI section content.
Recommendation
Squash or reword the three "add bankofai" commits into descriptive messages before merging, e.g.:
- "Add BANK-OF-AI Intro and QuickStart pages (en + zh-Hans)"
- "Update SKILLS QuickStart with conversational install method"
[m-04] package.json name Field Mismatch with Project Content
| Property | Value |
|---|---|
| Severity | Minor |
| Category | Documentation Maintainability |
| File | package.json : Line 2 |
Description
The
package.jsonnamefield is"@x402-tron/docs"and thedescriptionis"x402-tron documentation". This PR significantly repositions the documentation site as BANK OF AI product documentation, not x402-tron-specific documentation. The package name and description no longer reflect the actual scope of the documentation. While this is not a build-breaking issue, it can cause confusion for developers and in generated metadata.
Recommendation
Update
nameanddescriptioninpackage.jsonto reflect the actual project scope, for example:"name": "@bankofai/docs", "description": "BANK OF AI documentation site"
Suggestions
[S-01] BANK-OF-AI Intro Uses Inline Emojis in Headings
| Property | Value |
|---|---|
| Severity | Suggestion |
| Category | Code Quality |
| File | docs/BANK-OF-AI/Intro.md |
Description
Section headings use emojis (e.g.
## 🧠 Model Layer,## 🛤️ Protocol Layer). While visually appealing in rendered form, emoji in headings can cause issues with: (a) auto-generated anchor links that include the emoji, making fragment URLs fragile across platforms; (b) screen reader accessibility; (c) search indexing. Other documentation pages in this repo do not use emojis in headings.
Recommendation
Consider moving emojis to the body text or a lead paragraph rather than the heading itself, or ensure Docusaurus anchor generation handles emoji consistently.
[S-02] Repetitive Node.js Prerequisites Across Multiple Pages
| Property | Value |
|---|---|
| Severity | Suggestion |
| Category | Code Quality / DRY |
| File | Multiple changed files |
Description
The Node.js prerequisite block is repeated nearly verbatim across multiple new/updated pages:
SKILLS/QuickStart.md,TRONMCPServer/QuickStart.md,TRONMCPServer/OfficialServerAccess.md,SUNMCPServer/QuickStart.md, andSUNMCPServer/OfficialServerAccess.md. Each uses the same wording: "Node.js is required on the machine where the AI Agent runs (the Agent uses npx under the hood)..."
Recommendation
Extract this into a shared Docusaurus snippet (partial) or a dedicated prerequisites page linked from each page. This reduces maintenance burden when the minimum Node.js version changes.
[S-03] Example Wallet Addresses Are Real-Looking Ethereum/TRON Addresses
| Property | Value |
|---|---|
| Severity | Suggestion |
| Category | Documentation |
| File | docs/Agent-Wallet/QuickStart.md : Lines 57–58 |
Description
The example wallet output in Method 1 shows
0x1339Df9ac21E494b39Be47d012A3f33fb9188366(EVM) andTBisApdFay75x3FXNWiG8UE75KzhnFLceY(TRON) as sample addresses. These appear to be real generated addresses from a test wallet. While using real addresses in documentation is common practice, using clearly shortened or annotated addresses (e.g.0x1339...8366) as done in BANK-OF-AI/QuickStart.md (0x7858...4bCA) is a safer convention because it unambiguously signals "this is an example" and does not associate real on-chain addresses with documentation tooling.
Recommendation
Replace the full sample addresses with truncated, clearly example-style values such as
0x1339...8366andTBis...ceY, consistent with the BANK-OF-AI/QuickStart.md pattern.
Positive Observations
| Area | Observation |
|---|---|
| Content structure | Consistent use of Docusaurus callout types (:::tip, :::caution, :::danger) with appropriate severity mapping throughout all new content |
| Cross-references | All internal links from new BANK-OF-AI pages to existing docs (Agent-Wallet, MCP Server, Skills, x402, 8004, llmservice) are correct and bidirectional |
| Security guidance | The Agent-Wallet QuickStart includes strong, explicit warnings about password backup, master password risks, and the danger of using double quotes vs single quotes for passwords — materially important for a cryptocurrency wallet tool |
| Translation parity | New English content is fully mirrored in zh-Hans translations including images, maintaining complete bilingual documentation coverage |
| Installation UX | The new three-method install pattern (Conversational / Quick Auto-Install / Interactive) is consistent across all updated pages and represents a clear improvement in accessibility for non-technical users |
| Version management | package.json version bump is correctly included in the PR |
| Sidebar navigation | New BANK-OF-AI section is properly registered in both sidebars.js and i18n/zh-Hans/.../sidebars.js, with correct i18n label added to current.json |
Checklist Results
| Category | Items Checked | Pass | Fail | N/A | Notes |
|---|---|---|---|---|---|
| Correctness | 6 | 4 | 2 | 0 | Skill count inconsistency (M-01); quote instruction contradiction (M-02) |
| Security | 4 | 3 | 1 | 0 | Realistic example password value (M-03) |
| Performance | 3 | 3 | 0 | 0 | Static documentation site — no runtime performance concerns |
| Code Quality | 5 | 3 | 2 | 0 | Commit messages (m-03); package.json name (m-04); repeated Node.js prereqs (S-02) |
| Testing | 2 | 2 | 0 | 0 | Documentation-only PR; no code logic to test |
| Documentation | 5 | 4 | 1 | 0 | zh-Hans sidebar missing SKILLS/QuickStart (m-01) |
| Compatibility | 3 | 2 | 1 | 0 | Mac-only terminal command in cross-platform section (m-02) |
| Observability | 2 | 2 | 0 | 0 | N/A for documentation project |
Disclaimer
This is an automated code review. It supplements but does not replace human review.
Report generated by Code Review Skill v1.0.0
Date: 2026-04-16
Audit Report: PR
|
| Property | Value |
|---|---|
| From Branch | update-mcp-server (remote: origin/update-mcp-server) |
| To Branch | main (remote: origin/main) |
| Commits | 21 commits |
| Files Changed | 24 files |
| Lines Added | ~1,633 |
| Lines Removed | ~101 |
| Binary Assets Added | 2 PNG images, 2 SVG images |
Commit Range Summary
The commits span from foundational MCP server quick-start improvements (620f06b) through iterative introduction updates, culminating in the addition of a new BANK-OF-AI documentation section (c7c6683).
2. Change Summary
Group A: New BANK-OF-AI Documentation Section (New Files)
docs/BANK-OF-AI/Intro.md— 272-line introduction to the BANK OF AI platform (4-layer architecture overview, product matrix, ecosystem table)docs/BANK-OF-AI/QuickStart.md— 231-line end-to-end quick-start guide covering 2-step onboardingdocs/BANK-OF-AI/image/bankofai-architecture.svg— Architecture diagram SVG
Group B: Navigation & Sidebar Updates (Modified)
sidebars.js— AddsBANK OF AIas the first sidebar category withIntroandQuickStartitemsi18n/zh-Hans/docusaurus-plugin-content-docs/current/sidebars.js— Mirrors the sidebar update for Chinese localei18n/zh-Hans/docusaurus-plugin-content-docs/current.json— Adds sidebar category label entry forBANK OF AIpackage.json— Version bump from1.2.8→1.2.9
Group C: MCP Server Documentation Revamp (Modified)
docs/McpServer-Skills/MCP/SUNMCPServer/OfficialServerAccess.md— Adds "Method 1: Conversational Install" and renames existing methods; removes "restart AI client" stepdocs/McpServer-Skills/MCP/SUNMCPServer/QuickStart.md— Rewrites install section to conversational-first approach; removes restart stepdocs/McpServer-Skills/MCP/TRONMCPServer/OfficialServerAccess.md— Same pattern as SUN MCPdocs/McpServer-Skills/MCP/TRONMCPServer/QuickStart.md— Same pattern as SUN MCP; removes restart step
Group D: SKILLS Documentation Revamp (Modified)
docs/McpServer-Skills/SKILLS/QuickStart.md— Adds Node.js prerequisite tip; converts to 3-method install structure (Conversational / Auto-CLI / Interactive)docs/McpServer-Skills/SKILLS/image/conversational-install.png— New screenshot for conversational install
Group E: Agent-Wallet Quick Start Update (Modified)
docs/Agent-Wallet/QuickStart.md— Adds "Method 1: Conversational Setup" as primary path; renames existing CLI steps as "Method 2"; generalizes "OpenClaw" references to "your AI agent"
Group F: i18n Mirrors (New + Modified)
All docs changes in Groups A–E are mirrored under i18n/zh-Hans/docusaurus-plugin-content-docs/current/ with Chinese translations.
3. Detailed Findings
Critical
No Critical findings identified.
Major
[M-01] Incorrect Install Command for Etherscan in Ecosystem Table (Major)
| Property | Value |
|---|---|
| Severity | Major |
| Category | Correctness / Documentation |
| File | docs/BANK-OF-AI/Intro.md : Line 197 |
Description:
The ecosystem table lists npx skills add https://docs.etherscan.io as the install command for the Etherscan MCP Server/Skills. https://docs.etherscan.io is a documentation website URL, not a valid GitHub repository path or npm package name. Running this command would fail or install an unintended package. The same error is reproduced in the Chinese translation.
Code:
| **Etherscan** | Etherscan Official | Query / Tracking / Analytics, 60+ chains | `npx skills add https://docs.etherscan.io` |Recommendation:
Replace with the correct GitHub repository path or npm package name for the Etherscan Skills package. For example:
| **Etherscan** | Etherscan Official | Query / Tracking / Analytics, 60+ chains | `npx skills add etherscan/etherscan-skills` |Verify the correct install command from Etherscan's official documentation before merging.
[M-02] Removal of Restart Instructions May Leave Users Stuck (Major)
| Property | Value |
|---|---|
| Severity | Major |
| Category | Correctness / User Experience |
| File | docs/McpServer-Skills/MCP/TRONMCPServer/QuickStart.md, docs/McpServer-Skills/MCP/SUNMCPServer/QuickStart.md, docs/McpServer-Skills/MCP/TRONMCPServer/OfficialServerAccess.md, docs/McpServer-Skills/MCP/SUNMCPServer/OfficialServerAccess.md |
Description:
The PR removes all instructions to restart the AI client after MCP Server installation. For CLI-based installs (Methods 2 and 3 in the updated docs), MCP configuration is written to a file that the AI client reads only on startup. Without restarting, the newly installed MCP Server will not be visible or usable. The conversational install (Method 1) may handle this differently by triggering a reload, but no explanation is provided. The removal also deletes a troubleshooting checklist item ("AI client has been fully quit and restarted") that helped users diagnose connection failures.
Code (removed lines):
-After the command completes, restart your MCP client.
-Once installation is complete, restart your AI Agent, and you can start interacting...
-3. AI client has been fully quit and restarted (not just refreshed)Recommendation:
Either:
- Re-add the restart instructions for Methods 2 and 3 (CLI-based), limiting the "no restart needed" claim to Method 1 (conversational), OR
- Explain clearly why a restart is no longer needed (e.g., the tool now hot-reloads the config). Without justification, this is a factual regression for CLI users.
[M-03] Inconsistent Install Commands Between BANK-OF-AI and SKILLS QuickStarts (Major)
| Property | Value |
|---|---|
| Severity | Major |
| Category | Correctness / Consistency |
| File | docs/BANK-OF-AI/QuickStart.md : Lines 15–16 vs docs/McpServer-Skills/SKILLS/QuickStart.md : Lines 23–27 |
Description:
The BANK-OF-AI QuickStart instructs users (via the AI chat prompt) to install Skills into the Agent's own skills directory (no -g flag), while the SKILLS QuickStart Method 2 uses -y -g (global install). A user who follows the BANK-OF-AI QuickStart would have Skills only for that Agent, while a user following SKILLS QuickStart Method 2 would have a global install. This inconsistency is compounded by subtly different wordings in the two conversational prompts:
- BANK-OF-AI:
"Note: please install into this Agent's own skills directory." - SKILLS:
"Note: Please install to the skill directory corresponding to the current Agent."
These instructions say similar things but may produce different AI behaviors depending on how the AI client interprets them.
Recommendation:
Align the two documents. Decide which install scope (global vs. agent-specific) is the recommended default and use a single canonical prompt text that both docs reference or quote identically.
Minor
[m-01] BANK OF AI Sidebar Category Label Not Translated in Chinese Locale (Minor)
| Property | Value |
|---|---|
| Severity | Minor |
| Category | Documentation / i18n |
| File | i18n/zh-Hans/docusaurus-plugin-content-docs/current.json : Lines 6–9 |
Description:
The new sidebar category entry for BANK OF AI in the Chinese locale file has its message set to "BANK OF AI" (the English brand name), which is acceptable since it is a proper noun. However, all other categories in the same file have translated labels (e.g., "x402支付协议"). The brand name being untranslated is likely intentional but warrants confirmation.
Code:
"sidebar.docsSidebar.category.BANK OF AI": {
"message": "BANK OF AI",
"description": "The label for category BANK OF AI in sidebar docsSidebar"
},Recommendation:
Confirm with stakeholders whether BANK OF AI should remain in English in the Chinese sidebar (acceptable for brand names), or if a Chinese equivalent (e.g., "BANK OF AI ——— AI银行") is preferred.
[m-02] Inconsistent Terminology: "BofAI" vs "bankofai" vs "BankOfAI" (Minor)
| Property | Value |
|---|---|
| Severity | Minor |
| Category | Code Quality / Consistency |
| File | docs/BANK-OF-AI/Intro.md, docs/McpServer-Skills/SKILLS/QuickStart.md, docs/BANK-OF-AI/QuickStart.md |
Description:
Three different capitalizations of the organization/brand name are used in install commands and references throughout the new documentation:
BofAI/skills(GitHub shorthand in conversational prompts)https://github.com/BofAI/skills(full GitHub URL in CLI commands)@bankofai/agent-wallet(npm package namespace)bankofai-guide(skill name)
While BofAI as a GitHub org and bankofai as an npm namespace are technically different namespaces, inconsistent casing in user-facing docs can confuse readers.
Recommendation:
Add a brief note or ensure all references consistently use the correct name for each context (GitHub org vs npm package). This also applies to the displayed text where BANK OF AI, BankOfAI, and Bank of AI appear interchangeably in prose.
[m-03] Example Password in Shell Export Section May Mislead Users (Minor)
| Property | Value |
|---|---|
| Severity | Minor |
| Category | Documentation Clarity |
| File | docs/Agent-Wallet/QuickStart.md : Lines 203–208 |
Description:
The documentation correctly warns about using single quotes vs double quotes for passwords containing special characters ($). The example uses P@ss$w0rd! to illustrate shell expansion dangers, but this looks like a placeholder "strong password" example that some users might actually copy and use verbatim as their real password.
Code:
# ✅ Correct — single quotes, password saved as-is
export AGENT_WALLET_PASSWORD='P@ss$w0rd!'
# ❌ Wrong — double quotes, $w0rd gets shell-expanded to empty string, password silently breaks
export AGENT_WALLET_PASSWORD="P@ss$w0rd!" # actual value becomes "P@ss!"Recommendation:
Use a placeholder format that clearly indicates this is illustrative, such as 'Your-Actual-P@ss$w0rd!' or replace the realistic-looking password with a clearer placeholder like '<your-password-here>' in both the correct and incorrect examples.
[m-04] "Quick Start" Link in Intro.md Skills Section Ambiguous (Minor)
| Property | Value |
|---|---|
| Severity | Minor |
| Category | Documentation / Link Accuracy |
| File | docs/BANK-OF-AI/Intro.md : Line 149 |
Description:
In the Skills section of Intro.md, the "Learn more" line links to [Quick Start](./QuickStart.md), which resolves to docs/BANK-OF-AI/QuickStart.md (the BANK-OF-AI overall quick start). However, the natural expectation from a Skills section is a link to the Skills-specific quick start at docs/McpServer-Skills/SKILLS/QuickStart.md.
Code:
👉 Learn more: [Skills Introduction](../McpServer-Skills/SKILLS/Intro.md) · [Quick Start](./QuickStart.md)Recommendation:
Clarify whether the Quick Start link is intentional (linking to the BANK-OF-AI general quick start as the recommended entry point) or should be changed to the Skills-specific quick start. If the former, rename the link text to [BANK OF AI Quick Start](./QuickStart.md) to avoid ambiguity.
Suggestions
[S-01] Conversational Install Prompt Wording Not Standardized (Suggestion)
| Property | Value |
|---|---|
| Severity | Suggestion |
| Category | Documentation Quality |
| File | Multiple files |
Description:
The "conversational install" prompts that users are asked to paste into their AI chat exist in multiple forms across the new docs. They are similar but not identical:
"Run npx skills add BofAI/skills to install every BANK OF AI Skill. After the install completes, run bankofai-guide to walk me through the next steps. Note: please install into this Agent's own skills directory."(BANK-OF-AI QuickStart)"Run npx skills add BofAI/skills to install all BANK OF AI skills. After installation, use bankofai-guide to guide subsequent operations. Note: Please install to the skill directory corresponding to the current Agent."(SKILLS QuickStart)
Slight wording differences ("every" vs "all", "walk me through the next steps" vs "guide subsequent operations") could produce different AI behavior.
Recommendation:
Define a single canonical version of each prompt and reference it consistently, or use a shared include/snippet if the documentation framework supports it.
[S-02] Architecture SVG References Placeholder Transaction Hash (Suggestion)
| Property | Value |
|---|---|
| Severity | Suggestion |
| Category | Documentation Quality |
| File | docs/BANK-OF-AI/Intro.md : Line 243 |
Description:
The architecture walkthrough example shows a placeholder transaction hash 0xabc123...def456 with a TronScan link that uses a generic https://tronscan.org URL without an actual transaction path. This is acceptable for illustration but may confuse readers who try to click the link expecting a real example.
Code:
> Tx hash: `0xabc123...def456` ([view on TronScan](https://tronscan.org))Recommendation:
Either link to a real example transaction on TronScan, or make the placeholder nature explicit (e.g., [view on TronScan](https://tronscan.org/#/transaction/your-tx-hash)).
[S-03] Version Bump in package.json Without Changelog Entry (Suggestion)
| Property | Value |
|---|---|
| Severity | Suggestion |
| Category | Maintainability |
| File | package.json : Line 3 |
Description:
The package version is bumped from 1.2.8 to 1.2.9 in this PR. If this project maintains a changelog or release notes, no corresponding entry is visible in the PR changes.
Recommendation:
Ensure a CHANGELOG.md entry (or equivalent) is updated to document the version bump and the changes it encompasses. If the project does not maintain a changelog, this is informational only.
4. Positive Observations
-
Thorough security warnings on wallet password handling. The
Agent-Wallet/QuickStart.mdchanges include clear:caution:blocks about master password backup, the consequences of loss ("no recovery mechanism, no support team, no backdoor"), and explicit guidance against sharing via chat/email/repos. -
Progressive disclosure structure. The three-method install pattern (Conversational → Auto-CLI → Interactive) is well-designed: beginners get the easiest path first, power users get full control without scrolling past irrelevant steps.
-
Complete i18n mirroring. All English documentation changes are fully mirrored in Chinese (
zh-Hans), including the new BANK-OF-AI section, images, and sidebar configuration — demonstrating good internationalization discipline. -
Comprehensive cross-referencing. The new
BANK-OF-AI/Intro.mdprovides rich internal linking to every referenced sub-component (llmservice,x402,8004,Agent-Wallet,Skills,MCP Server), and all cross-referenced files were confirmed to exist in the branch. -
Good prerequisite documentation. Node.js requirements are consistently surfaced as
:::tip Prerequisiteblocks at the top of each installation page, preventing the most common setup failure. -
Shell quoting security education. The
Agent-Wallet/QuickStart.mdactively educates users on the risk of usingecho "..." >> .zshrc(shell history exposure) and recommends editing config files directly in an editor instead — practical and security-conscious advice. -
SVG architecture diagram. The addition of
bankofai-architecture.svg(an inline SVG) avoids external image hosting dependencies and ensures the diagram renders reliably in all documentation environments.
5. Checklist Results
| Category | Check | Status | Notes |
|---|---|---|---|
| Correctness | Logic errors in changed code | Pass | Documentation only; no application logic |
| Correctness | All cross-referenced file links resolve | Pass | All ../path/file.md links verified to exist |
| Correctness | Install commands accurate | Fail | Etherscan install URL is invalid (M-01) |
| Correctness | Restart instructions accurate | Fail | Removed without justification for CLI installs (M-02) |
| Security | No hardcoded secrets or credentials | Pass | No real keys/passwords in changed files |
| Security | Sensitive data handling instructions | Pass | Explicit warnings about password storage |
| Security | User input / injection risks | N/A | Documentation only |
| Performance | N/A for documentation | N/A | — |
| Code Quality | Naming consistency | Minor Issue | BofAI/bankofai/BankOfAI inconsistency (m-02) |
| Code Quality | Dead code / unused content | Pass | No dead content identified |
| Code Quality | Internal consistency | Fail | Install command scope differs across docs (M-03) |
| Testing | N/A for documentation | N/A | — |
| Documentation | New sections have appropriate cross-links | Pass | BANK-OF-AI section is well cross-linked |
| Documentation | i18n coverage | Pass | Full Chinese translation provided |
| Documentation | Sidebar navigation correct | Pass | Both English and zh-Hans sidebars updated |
| Documentation | Version bump recorded | Suggestion | No changelog entry identified (S-03) |
6. Review Verdict
Decision: Request Changes
Rationale
This PR delivers a well-structured, security-conscious documentation revamp with strong internationalization support. The content quality is high and the user experience improvements (conversational install as the primary path) are meaningful.
However, three Major findings must be addressed before merge:
-
[M-01] — The Etherscan install command (
npx skills add https://docs.etherscan.io) is factually incorrect and would fail if a user followed it. This is a documentation error that actively harms users. -
[M-02] — The removal of AI client restart instructions for CLI-based MCP Server installations removes necessary guidance without explanation. CLI users who install via Method 2 or Method 3 will find their MCP Server invisible until they restart, with no documentation telling them to do so.
-
[M-03] — The inconsistency between the BANK-OF-AI QuickStart (agent-specific install) and the SKILLS QuickStart (global install) creates a confusing and potentially conflicting user experience.
Addressing these three issues would make this PR ready to merge. The Minor and Suggestion items are non-blocking but should be tracked for follow-up.
Code Review ReportProject: @x402-tron/docs PR OverviewBranch Information
Commit History
Review SummaryVerdict
Two major issues require resolution before merging: the removal of restart instructions breaks setup for command-line users (and removes an item from the troubleshooting checklist that users rely on), and the conversational install method makes unverified auto-detection claims without any fallback guidance. Two minor issues — a skill-count discrepancy and a likely orphaned image — should also be corrected. Findings at a Glance
SummaryThis PR is a pure documentation update — no application source code is changed. It introduces a new top-level "BANK OF AI" section (intro + quick-start + architecture diagram), adds documentation for two new skills ( Overall quality is high: the new content is well-structured, security guidance around wallet passwords is thorough and consistently applied, and the new architecture SVG diagram is a useful visual addition. The refactored installation guides correctly present three clearly-labelled methods with graduated complexity. However, two significant accuracy problems were introduced during the refactor: (1) restart instructions were silently removed from command-line install methods where they are still required, and (2) the conversational install method claims the AI will "auto-detect which AI client is currently running" with no documentation of what happens if that detection fails. A minor skill-count inconsistency (the text says "Ten" while 11 are listed) and a potentially orphaned image file also need attention. Change Summary1 · New BANK OF AI Documentation Section
Purpose: Provide a top-level entry point explaining the full BANK OF AI product matrix before users dive into component-specific docs. 2 · Conversational Install Method Added to MCP Server Docs
Purpose: Lower the barrier to entry by making chat-driven installation the primary recommended path. 3 · Conversational Install Method Added to Skills Docs + New Skills Documented
Purpose: Document two new skills and align the Skills installation guide with the new three-method pattern. 4 · Agent-Wallet QuickStart Restructure
Purpose: Add conversational wallet creation as the primary path, consistent with the rest of the documentation. 5 · Package Version Bump
Purpose: Version the documentation release. Detailed FindingsMajor[MJ-01] Restart Instructions Removed from Command-Line Install Methods
Description The PR removes "restart your AI Agent" instructions from every command-line installation path (Methods 2 and 3), and also deletes the troubleshooting checklist item "AI client has been fully quit and restarted (not just refreshed)" from the MCP Servers installed via The original docs even warned explicitly at the Code (removed lines): -Once installation is complete, restart your AI Agent, and you can start interacting
-with SunSwap via SUN MCP Server.
+Once installation is complete, you can start interacting with SunSwap via SUN MCP Server.
-3. AI client has been fully quit and restarted (not just refreshed)And in -After saving the configuration, **completely quit and restart** your AI client.
-After restarting, the client will automatically connect to TRON MCP Server.
-
-Then enter your first query in the chat:
+## Test It Out
+
+Enter your first query in the chat:Recommendation Restore restart instructions specifically for Method 2 (Quick Auto-Install) and Method 3 (Interactive Install), since the conversational Method 1 may handle this differently. For example: ### Method 2: Quick Auto-Install (Command Line)
...
Once installation is complete, **fully restart your AI client** to load the new MCP
Server configuration, then start asking questions.Also restore the troubleshooting checklist item for the restart step. If the conversational install genuinely avoids the need to restart (e.g. by hot-reloading), document that distinction explicitly rather than silently removing guidance for all paths. [MJ-02] Conversational Install Claims Unverified AI Auto-Detection with No Fallback
Description The new "Conversational Install (Easiest)" method makes the following specific claims that depend on the AI client's behavior, but provides no fallback guidance when they fail:
These behaviors depend on If detection fails, the user has no guidance: there is no "if this doesn't work, do X" path in Method 1. The original docs at least ensured the user knew exactly which file was being edited. Code: 3. The AI handles the entire flow automatically:
- Detects the remote MCP service URL
- Auto-detects which AI client is currently running
- Writes the `tron-mcp-server` entry into the right MCP config file (no manual JSON editing)
- Reports back with a ✅ confirmation when done
Once the AI confirms the install, the TRON MCP Server is ready — you can start asking
questions right away.Recommendation Add a brief fallback/troubleshooting note after the success path to set expectations and give users an escape hatch: :::tip If the conversational install doesn't work as expected
If the AI reports an error, or if MCP tools are unavailable after install,
fall back to **Method 2** (command-line) which gives you direct control over which
config file is updated. See the [Troubleshooting](#troubleshooting) section below.
:::Also consider noting that a restart may still be required after conversational install in some clients (see [MJ-01] above). Minor[MN-01] Skill Count Inconsistency — "Ten" vs 11 Skills Listed Everywhere
Description The PR updates
The count "Ten" was presumably intended to reflect the newly added Code: -Eight core skills covering the most common scenarios in the TRON ecosystem.
+Ten core skills covering the most common scenarios in the TRON ecosystem.Recommendation Update the count to 11 in Eleven core skills covering the most common scenarios in the TRON ecosystem.Alternatively, if [MN-02] Likely Orphaned Image:
|
| Property | Value |
|---|---|
| Severity | Minor |
| Category | Documentation |
| File | docs/McpServer-Skills/SKILLS/image/conversational-install.png (added); i18n/zh-Hans/.../SKILLS/image/conversational-install.png (added) |
Description
Two binary PNG files are added to the image directories for the Skills section, but no corresponding ![]() image reference appears anywhere in the changed portions of docs/McpServer-Skills/SKILLS/QuickStart.md or any other modified file in this PR. The images are therefore either:
- Orphaned (added but not yet referenced — a forgotten step), or
- Referenced in an unchanged line of a file (in which case this is a non-issue, but should be verified)
An unreferenced image adds binary weight to the repository without serving any user-visible purpose.
Recommendation
Verify whether these images are actually referenced anywhere in the current docs:
grep -r "conversational-install" docs/ i18n/If no reference exists, either add the  reference in the appropriate location in the Skills QuickStart guide, or remove the image files from the PR.
Suggestions
[S-01] Clarify That "Conversational Install" Still Requires Node.js Installed Before Use
File: docs/McpServer-Skills/MCP/TRONMCPServer/OfficialServerAccess.md · SUNMCPServer/OfficialServerAccess.md
Description: The Conversational Install method correctly notes in a :::tip Prerequisite callout that Node.js must be installed. However, the prerequisite is nested inside the Method 1 section, making it easy to overlook. If a user skips the prerequisite and the npx call silently fails, they will receive no meaningful error — the AI may simply report that the command failed.
Suggestion: Move the Node.js prerequisite callout to the top of the "How to Connect" section, above all three method descriptions, so it applies globally rather than appearing to be Method 1-specific. This is already done correctly in SKILLS/QuickStart.md where the Node.js tip appears before the method list.
[S-02] Consider Adding the New BANK OF AI Section to the Existing Navigation Structure Note
File: sidebars.js and i18n/zh-Hans/.../sidebars.js
Description: The new BANK OF AI category is inserted as the first item in docsSidebar with collapsed: false. This means it is always expanded when the sidebar loads, pushing all existing navigation entries (x402, 8004, Agent-Wallet, MCP Server, Skills, LLM Service…) further down. For returning users, this is a significant navigation disruption.
Suggestion: Consider whether collapsed: true is more appropriate as a default for a section that experienced users may not need to expand, or add a brief migration/changelog note indicating the new section was added and where existing content moved. Since this is a docs site, a :::note New in v1.2.9 admonition at the top of the main docs landing page would also help orient returning users.
Positive Observations
| Area | Observation |
|---|---|
| Security guidance | Password security warnings are consistent, prominent, and appropriately severe. Every location where a master password is involved uses :::caution or :::danger admonitions with clear instructions to back up externally and never share via chat, email, or repos. |
| Architecture diagram | The new bankofai-architecture.svg is an inline SVG with well-structured layers, meaningful colour coding, bidirectional arrows, and a legend — a substantial improvement over text-only architecture descriptions. Separate localised copies with translated labels are a thoughtful touch. |
| Progressive disclosure | The new three-method installation pattern (Conversational → Command-Line → Interactive) is well-designed: each method is clearly labelled with a difficulty hint, and links between them are consistent. |
| i18n parity | Every English change is correctly mirrored in the i18n/zh-Hans/ tree, including SVG labels and UI copy in the architecture diagram. |
| Heading hierarchy correction | The Agent-Wallet QuickStart correctly demotes sub-steps from ### to #### when they are nested under the new top-level method sections — a subtle but correct structural fix. |
| New skill documentation | usdd-skill and trx-staking-skill sections follow the established pattern (read-only examples, confirmation-required examples, real-world scenarios, cautionary callouts) and include domain-accurate information (e.g. 18-decimal USDD vs. 6-decimal USDT normalisation, 14-day unbonding period, vote replacement semantics). |
| Testnet-first guidance | The :::warning Three Rules for Beginners callout (testnet first, review every transaction, start small) is prominent and well-positioned in the new BANK OF AI QuickStart. |
Checklist Results
| Category | Items Checked | Pass | Fail | N/A | Notes |
|---|---|---|---|---|---|
| Correctness | 5 | 3 | 2 | 0 | Restart steps removed (MJ-01); auto-detection claim unverified (MJ-02) |
| Security | 5 | 5 | 0 | 0 | Password guidance is thorough; no secrets or credentials in changed files |
| Performance | 3 | 0 | 0 | 3 | Not applicable — documentation only |
| Code Quality | 4 | 3 | 1 | 0 | Skill count inconsistency (MN-01); structure and naming otherwise consistent |
| Testing | 2 | 0 | 0 | 2 | Not applicable — documentation only |
| Documentation | 6 | 4 | 2 | 0 | Orphaned image (MN-02); missing fallback guidance (MJ-02) |
| Compatibility | 3 | 2 | 1 | 0 | Sidebar change may disrupt existing navigation for returning users (S-02) |
| Observability | 1 | 0 | 0 | 1 | Not applicable — documentation only |
Disclaimer
This is an automated code review. It supplements but does not replace human review. The reviewer analyzed only the diff between the specified branches. Runtime behavior, integration testing, and deployment impact are not covered. All findings relate to documentation correctness and user-experience quality rather than application code.
Report generated by Code Review Skill v1.0.0
Date: 2026-04-16
Code Review ReportProject: x402-tron / BANK OF AI Documentation Site PR OverviewBranch Information
Commit History
Review SummaryVerdict
Findings at a Glance
SummaryThis PR is a documentation-only update to a Docusaurus site for the BANK OF AI platform. It introduces a new top-level "BANK OF AI" documentation section (Introduction + Quick Start), expands the Skills catalogue with two new entries ( Overall the documentation is well-structured, clearly written, and includes good security warnings (wallet password guidance is thorough). However, three major issues require attention before merging: restart instructions were silently removed from command-line install paths leaving users who follow Methods 2/3 without the key troubleshooting step they need; the MCP Server troubleshooting checklist was shortened in a way that removes the most common fix for connection problems; and a numeric inconsistency in skill counts (8 → 10 in Intro vs 11 shown in install output) will confuse new users. Four minor and four suggestions round out the findings. Change Summary1. New BANK OF AI Section (English + Chinese)
Purpose: Introduce BANK OF AI as the product umbrella with a unified onboarding path. 2. New Skills: usdd-skill & trx-staking-skill
Purpose: Document two newly available skills covering USDD/JUST Protocol operations and TRX staking + SR voting. 3. Conversational Install Refactor (MCP Servers + Skills)
Purpose: Promote conversational AI-driven install as the lowest-friction onboarding path. 4. Agent-Wallet QuickStart Restructure
Purpose: Align wallet creation docs with the new conversational-first philosophy. 5. Navigation & Configuration
Purpose: Wire up new content to site navigation and bump the package version. Detailed FindingsMajor[MJ-01] Restart Requirement Silently Removed from Command-Line Methods
Description
Code -Once installation is complete, restart your AI Agent, and you can start interacting with TRON blockchain via TRON MCP Server.
+Once installation is complete, you can start interacting with the TRON blockchain via TRON MCP Server.Recommendation Once installation is complete, **restart your AI client** so it picks up the updated
MCP config, then you can start interacting with the TRON blockchain via TRON MCP Server.
> **Note for conversational install (Method 1):** No restart is needed — the AI detects
> the new config automatically.Alternatively, if Method 1 genuinely eliminates the need for restart for all methods, add an explicit explanation of why that is the case, so users and future maintainers understand the behaviour change. [MJ-02] Troubleshooting Section Lost the Most Common Fix
Description
Code If you encounter issues, first confirm:
1. Node.js version >= 20.0.0 (run `node --version` to check)
2. Network can reach `sun-mcp-server.bankofai.io`
-3. AI client has been fully quit and restarted (not just refreshed)Recommendation If you encounter issues, first confirm:
1. Node.js version >= 20.0.0 (run `node --version` to check)
2. Network can reach `sun-mcp-server.bankofai.io`
3. AI client has been fully quit and restarted (not just refreshed)
*(not required if you used the Conversational Install — Method 1)*[MJ-03] Skill Count Inconsistency: Intro says 10, Install Output Shows 11
Description
Code -Eight core skills covering the most common scenarios in the TRON ecosystem.
+Ten core skills covering the most common scenarios in the TRON ecosystem.vs. QuickStart install output: Recommendation Audit the actual skill count and make it consistent across all files. If the correct number is 11, update Minor[MN-01]
|
| Property | Value |
|---|---|
| Severity | Minor |
| Category | Security / Documentation |
| File | docs/Agent-Wallet/QuickStart.md : Lines ~47–54 · docs/BANK-OF-AI/QuickStart.md : Lines ~582–586 |
Description
Multiple sections advise users that the master password is auto-saved to
~/.agent-wallet/runtime_secrets.jsonfor convenience. This is excellent UX, but the documentation does not mention that this file should be access-restricted (e.g.,chmod 600). If a multi-user machine or a compromised process reads this file, the wallet can be unlocked without the user's knowledge. The password security warnings are otherwise excellent (backup, never share, etc.), so the omission of file permissions stands out.
Recommendation
Add a note alongside the
runtime_secrets.jsonmention:⚠️ `~/.agent-wallet/runtime_secrets.json` is created with restricted permissions automatically on most systems. If you're on a shared machine, verify with `ls -la ~/.agent-wallet/runtime_secrets.json` that only your user can read it (mode `600`).
[MN-02] BANK OF AI i18n Sidebar Label Not Translated
| Property | Value |
|---|---|
| Severity | Minor |
| Category | Documentation / Consistency |
| File | i18n/zh-Hans/docusaurus-plugin-content-docs/current.json : Lines 5–8 |
Description
The new sidebar category label was added to the Chinese i18n file with
"message": "BANK OF AI"— identical to the key and to the English label. Every other category in this file is properly translated (e.g.,"x402支付协议","当前"). While "BANK OF AI" is a brand name and may deliberately remain in English, the inconsistency is jarring and looks like an oversight.
Code
"sidebar.docsSidebar.category.BANK OF AI": {
"message": "BANK OF AI",
"description": "The label for category BANK OF AI in sidebar docsSidebar"
},Recommendation
If the brand name is intentionally kept in English, add a comment explaining this. If a Chinese rendering is desired (e.g., "BANK OF AI（AI 银行）"), update the message. At minimum, confirm this is intentional to avoid future translators treating it as a missed translation.
[MN-03] Install Prompt Uses Shorthand BofAI/skills but Method 2 Uses Full GitHub URL
| Property | Value |
|---|---|
| Severity | Minor |
| Category | Documentation / Consistency |
| File | docs/McpServer-Skills/SKILLS/QuickStart.md : Lines 25–27 and 47 |
Description
The conversational install prompt (Method 1) tells users to run:
npx skills add BofAI/skillswhile Method 2 (Quick Auto-Install) uses:
npx skills add https://github.com/BofAI/skills -y -gAdditionally, the BANK OF AI QuickStart uses the shorthand form without
-gand with a note to install to the current Agent's directory only. These three variant invocations are presented without explanation of when each form is appropriate, which could lead users to wonder whether they install to different locations.
Recommendation
Add a brief note explaining the equivalence of the shorthand and full URL, and why the -g flag is omitted in the conversational install context. This prevents users from second-guessing the install path.
[MN-04] Hardcoded OpenClaw Path in Example Output of Broadened Documentation
| Property | Value |
|---|---|
| Severity | Minor |
| Category | Documentation / Consistency |
| File | docs/BANK-OF-AI/QuickStart.md : Lines ~419 · docs/McpServer-Skills/SKILLS/QuickStart.md |
Description
The example AI output in the new BANK OF AI Quick Start prominently shows:
Location: ~/.openclaw/workspace/.agents/skills/ (this Agent's directory)The PR's stated goal is to broaden support beyond OpenClaw to Claude Code, Cursor, Telegram bots, etc. The example output uses a path that is specific to OpenClaw, making users of other AI clients wonder if the install worked correctly or went to the wrong location.
Recommendation
Replace the hardcoded path with a placeholder or show multiple examples:
Location: <your-agent-skills-dir> (auto-detected for your Agent) # e.g., ~/.openclaw/workspace/.agents/skills/ (OpenClaw) # ~/.claude/skills/ (Claude Code)
Suggestions
[S-01] PSM Reserve Warning Could Quantify "Low"
File: docs/McpServer-Skills/SKILLS/BANKOFAISkill.md — usdd-skill section
Description: The PSM caution warns users to check reserves before redeeming, but "if USDT reserves are low" is vague. If there is a known threshold or a typical depletion pattern, sharing it would help users calibrate their expectations.
Suggestion: Optionally add: "Reserves below ~X USDT may cause large redemptions to fail; consider splitting into smaller batches." (Fill in the appropriate threshold based on protocol behaviour.)
[S-02] 14-Day Unstaking Clarification — Distinguish from TRX "Unstake" vs. Skill Scope
File: docs/McpServer-Skills/SKILLS/BANKOFAISkill.md — trx-staking-skill danger callout
Description: The :::danger callout says "When TRX is unstaked (via the companion staking flow)". If unstaking is not directly supported by this skill (the description says "read-only" for that operation), the wording might imply users can trigger it through this skill and encounter the 14-day lockup. A clarification that unstaking is handled by a separate flow (and pointing to where) would prevent confusion.
Suggestion: Add: "Note: unstaking itself is handled by the companion trx-staking-management flow, not this skill directly." (Or update to reflect the actual scope.)
[S-03] Architecture SVG Is Duplicated for English/Chinese — Consider Shared Asset
File: docs/BANK-OF-AI/image/bankofai-architecture.svg and i18n/zh-Hans/.../bankofai-architecture.svg
Description: Two nearly identical SVG files are maintained — one with English labels, one with Chinese. SVG text is inline, so there is no shared base asset. Any future change to the diagram layout must be applied twice. Over time this could lead to diagrams diverging.
Suggestion: Consider externalising labels (e.g., using CSS classes, a JS-rendered diagram library, or a shared SVG with localised <text> blocks via Docusaurus's i18n mechanism) to reduce maintenance burden. This is a long-term suggestion and not blocking for this PR.
[S-04] Version Bump Commit Message Does Not Describe Scope
File: package.json
Description: The version was bumped from 1.2.8 to 1.2.9 across multiple commits with generic messages like "add bankofai". The absence of a CHANGELOG entry or a descriptive commit means 1.2.9 carries no human-readable record of what changed.
Suggestion: Add a CHANGELOG.md entry (or equivalent) documenting what was added in 1.2.9 — new BANK OF AI section, new skills, conversational install path, etc.
Positive Observations
| Area | Observation |
|---|---|
| Security warnings | Password security guidance is thorough and prominent — multiple :::caution / :::danger callouts warn about the master password being shown once, the importance of external backups, and never sharing via chat/email/screenshots. |
| User experience | The "conversational install" concept lowers the barrier significantly for non-technical users. The step-by-step with exact prompts to paste is clear and actionable. |
| Bilingual parity | English and Chinese documentation appear to be in sync — every new English file has a corresponding Chinese file, and structural changes are mirrored accurately. |
| SVG architecture diagram | The inline SVG is clean, well-commented, and uses semantic color coding to distinguish flow layers. Both English and Chinese versions correctly localise all visible text. |
| Risk callouts on new skills | The trx-staking-skill danger callout about the 14-day unbonding period and the PSM caution for usdd-skill are well-placed and appropriately prominent for financial operations. |
| Sidebar placement | The new BANK OF AI category is placed at the top of the sidebar with collapsed: false, giving it high visibility — appropriate for an umbrella introduction. |
Checklist Results
| Category | Items Checked | Pass | Fail | N/A | Notes |
|---|---|---|---|---|---|
| Correctness | 5 | 3 | 2 | 0 | MJ-01 (removed restart), MJ-03 (skill count mismatch) |
| Security | 4 | 3 | 1 | 0 | MN-01 (file permissions not mentioned) |
| Performance | 3 | 3 | 0 | 0 | No performance concerns in doc changes |
| Code Quality | 5 | 3 | 2 | 0 | MN-03 (install command inconsistency), MN-04 (hardcoded path) |
| Testing | 2 | 0 | 0 | 2 | N/A — documentation only |
| Documentation | 6 | 4 | 2 | 0 | MJ-02 (truncated troubleshooting), MN-02 (missing i18n translation) |
| Compatibility | 2 | 2 | 0 | 0 | Sidebar additions are backward-compatible |
| Observability | 1 | 0 | 0 | 1 | N/A — documentation only |
Disclaimer
This is an automated code review. It supplements but does not replace human review. The reviewer analysed only the diff between main and update-mcp-server. Runtime behaviour of the Docusaurus site build, link validity, image rendering, and end-to-end installation flows were not tested.
Report generated by Code Review Skill v1.0.0
Date: 2026-04-17
Code Review ReportProject: @x402-tron/docs (BANK OF AI Documentation Site) PR OverviewBranch Information
Commit History
Review SummaryVerdict
Findings at a Glance
SummaryThis PR is a documentation-only update to a Docusaurus-based documentation site for the BANK OF AI Web3/AI platform. There are no changes to application source code, server logic, or security-sensitive runtime components. All 28 changed files are Markdown documentation pages, SVG/PNG assets, sidebar configuration, i18n translation files, and a minor version bump in The PR delivers substantial new content: a full new "BANK OF AI" section with an introduction and quick-start guide (in both English and Chinese), two new skill reference sections ( The one major concern is that the PR removes "restart your AI Agent/MCP client" instructions from the CLI-based installation methods without explanation. These restarts are still required for config changes to take effect for command-line installs, so users following Method 2 or Method 3 may find their MCP Server is not available after installation until they discover they need to restart. Several minor inconsistencies in install commands and skill counts were also identified. Change Summary1. New BANK OF AI Documentation Section
Purpose: Introduces the overarching BANK OF AI product umbrella (4-layer model: Model/Protocol/Tool/Ecosystem) as a new top-level section in the documentation, giving users a unified entry point before diving into individual products. 2. Conversational Install Flow Added to MCP Server Docs
Purpose: Adds a new "conversational install" path (Method 1) for users who already have an AI agent that can run shell commands — the user pastes a prompt into the AI, and the AI installs the MCP Server automatically without the user touching a terminal. Restart instructions were removed from all CLI-based methods. 3. Conversational Install Added to Skills Quick Start
Purpose: Same pattern as MCP docs — introduces the conversational install as the recommended beginner path. 4. Agent-Wallet Quick Start Updated
Purpose: Expands the wallet creation guide with a simpler conversational path for users of the BANK OF AI skill suite. 5. Two New Skill Reference Sections
Purpose: Documents the USDD/JUST Protocol Peg Stability Module skill and the TRX Staking & SR Voting skill that are now included in the BANK OF AI skill suite. 6. Navigation and Version
Purpose: Wires the new BANK OF AI section into the site navigation and bumps the package version. Detailed FindingsMajor[MJ-01] Removed Restart Instructions from CLI Install Methods May Break User Experience
Description The PR removes "restart your AI Agent / MCP client" instructions from all installation methods, including the CLI-based methods (Method 2: Quick Auto-Install and Method 3: Interactive Install). The original text in at least four places read: and in the TRONMCPServer/QuickStart.md: Both were removed in this PR. The conversational install (Method 1) may indeed avoid a manual restart — the AI runs the install and the session may pick up the changes dynamically. However, for Method 2 (direct terminal Similarly, the troubleshooting checklist item "3. AI client has been fully quit and restarted (not just refreshed)" was removed from both the SUN and TRON OfficialServerAccess docs, eliminating a common fix from the debugging path. Code (removed lines, diff context): -Once installation is complete, restart your AI Agent, and you can start interacting
-with the TRON blockchain via TRON MCP Server.
+Once installation is complete, you can start interacting with the TRON blockchain
+via TRON MCP Server.-After saving the configuration, **completely quit and restart** your AI client.
-After restarting, the client will automatically connect to TRON MCP Server.
-
-Then enter your first query in the chat:
+## Test It Out
+
+Enter your first query in the chat:Recommendation Differentiate restart instructions by method. For Method 1 (conversational), it is acceptable to say no restart is needed. For Methods 2 and 3 (CLI-based), retain the restart instruction: ### Method 2: Quick Auto-Install (Command Line)
...
Once installation is complete, **restart your AI client** to load the new MCP
configuration, then you can start interacting via TRON MCP Server.
### Method 3: Interactive Install (Most Control)
...
Once installation is complete, **restart your AI client** to load the new MCP
configuration, then you can start interacting via TRON MCP Server.Also restore the troubleshooting checklist item: If you encounter issues, first confirm:
1. Node.js version >= 20.0.0 (run `node --version` to check)
2. Network can reach `tron-mcp-server.bankofai.io`
3. AI client has been fully quit and restarted (not just refreshed)Minor[MN-01] Install Command Format Inconsistency Between BANK-OF-AI/QuickStart and Skills/QuickStart
Description The BANK-OF-AI/QuickStart.md (the new top-level entry point) uses the short form install prompt with no flags: The Skills/QuickStart.md Method 2 uses the full URL with npx skills add https://github.com/BofAI/skills -y -gTwo inconsistencies arise:
Recommendation Align the conversational install prompts to explicitly specify all-accept behavior, or add a note clarifying that the AI agent should respond with [MN-02] Skill Count Inconsistency: "Ten core skills" vs. 11 Installed
Description
The "ten" likely refers to user-facing operational skills (excluding Recommendation Either update the description to "eleven skills" (or "11 skills") to match the exact install output, or add a clarifying note such as: "Ten operational skills, plus [MN-03]
|
| Property | Value |
|---|---|
| Severity | Minor |
| Category | Documentation / UX |
| File | sidebars.js : Lines 4–11 |
Description
The new BANK OF AI sidebar category is added at the top of docsSidebar with collapsed: false:
{
type: 'category',
label: 'BANK OF AI',
collapsed: false,
items: [
'BANK-OF-AI/Intro',
'BANK-OF-AI/QuickStart',
],
},Setting collapsed: false means the category will be permanently expanded on first load (and cannot be collapsed by default in Docusaurus unless collapsible: true is also present). While this is intentional for discoverability of the new section, all other sidebar categories use the default (collapsible) behavior, making this one visually inconsistent and potentially pushing other sections further down the page for new visitors.
Recommendation
Confirm this is the intended behavior. If the goal is just to have it expanded by default but still collapsible, the current Docusaurus config is correct — verify that other categories do not specify collapsed: true explicitly. If the intent is to match the style of other categories, remove collapsed: false to use the default.
[MN-04] Duplicate Security Warning Text Across Multiple New Documents
| Property | Value |
|---|---|
| Severity | Minor |
| Category | Documentation Maintainability |
| File | docs/Agent-Wallet/QuickStart.md : Lines 47–51; docs/BANK-OF-AI/QuickStart.md : Lines 582–586 |
Description
The master password backup warning appears verbatim (or near-verbatim) in at least three places:
docs/Agent-Wallet/QuickStart.md(Method 1 section)docs/BANK-OF-AI/QuickStart.md(Step 2 and "A Few Reminders" section)- Both Chinese translations
Example repeated text:
:::caution Your master password is critical — back it up immediately
The master password is only shown once. Although it is auto-saved to
`~/.agent-wallet/runtime_secrets.json`, you should also save it manually to a
password manager (1Password, Bitwarden, etc.). ...
⚠️ Never share this password via chat, email, screenshots, or public repositories.
:::While safety warnings benefit from repetition, having near-identical admonition blocks in multiple documents creates maintenance debt — a future update to the security guidance would need to be applied in multiple places.
Recommendation
This is acceptable for a documentation site where cross-referencing is standard. However, consider whether a single canonical location (e.g., a dedicated "Security" page or a Docusaurus :::caution include via MDX) would reduce duplication. At minimum, add a comment in the source noting the other locations where this text exists.
Suggestions
[S-01] BANK-OF-AI/QuickStart.md Verify Step References Balance Query Which Requires an Activated Account
File: docs/BANK-OF-AI/QuickStart.md : Line 533
Description: The "Verify the Install" step asks users to query TRX and USDT balances of the just-created wallet. The docs note that a brand-new TRON address may be "not activated" and show balance 0. However, the USDT balance query on an unactivated address may return an error rather than 0, which could make users think the install failed even when everything is working correctly.
Suggestion: Update the verify step to use a read-only query that works regardless of wallet activation state — for example, "Check the current price of TRX on SunSwap" — and move the balance query to a secondary "try after funding" section.
[S-02] SVG Architecture Diagrams Use Only Fixed-Width Layout
File: docs/BANK-OF-AI/image/bankofai-architecture.svg
Description: Both SVG files use a fixed viewBox="0 0 720 900" (portrait, 720px wide). The SVG will scale with the container via viewBox, but the text is sized in absolute font units (font-size="18", font-size="12", etc.), which may render very small on narrow mobile viewports. The diagram is embedded in the doc with no explicit width constraint.
Suggestion: Add a wrapping <div> with a max-width in the MDX/Markdown to constrain the SVG display size, or verify that the Docusaurus theme handles SVG scaling adequately on mobile breakpoints.
[S-03] Consider Using Docusaurus admonition Includes for Repeated Node.js Prerequisite
File: docs/McpServer-Skills/SKILLS/QuickStart.md, docs/McpServer-Skills/MCP/SUNMCPServer/OfficialServerAccess.md, docs/McpServer-Skills/MCP/TRONMCPServer/OfficialServerAccess.md
Description: An identical Node.js prerequisite tip box appears in multiple new files:
:::tip Prerequisite
**Node.js is required** on the machine where the AI Agent runs (the Agent uses `npx`
under the hood). If Node.js isn't installed yet, grab the LTS installer from
[nodejs.org](https://nodejs.org) — install once, double-click and follow the prompts.
:::Suggestion: Docusaurus supports MDX partial imports. Consider extracting this tip into a shared partial (e.g., docs/_partials/nodejs-prerequisite.md) and importing it across the files that reference it, reducing duplication and ensuring consistent wording.
Positive Observations
| Area | Observation |
|---|---|
| i18n consistency | Every English change has a corresponding Chinese (zh-Hans) translation. The translations mirror the English structure faithfully, including all code blocks, admonitions, and table layouts — demonstrating strong localization discipline. |
| Security guidance | The password backup warnings are appropriately prominent (placed in :::caution and :::danger admonition blocks) and appear immediately after the step where the password is generated, minimizing the chance of a user missing them. |
| Progressive disclosure | The three-method install pattern (Conversational → Quick CLI → Interactive) provides a clear skill-based ladder — beginners start with Method 1, power users use Method 3. Each method has a clear label summarizing its tradeoff. |
<details> example output |
Using collapsible <details> blocks for example AI outputs is an excellent pattern — it illustrates what to expect without cluttering the main reading flow. |
| Admonition usage | The PR correctly uses Docusaurus admonition types semantically: :::tip for helpful extras, :::caution for reversible risks, :::danger for irreversible consequences (e.g., 14-day unstaking lockup), and :::warning for general best practices. |
| Cross-linking | New docs are well cross-linked to related sections (Quick Start → Intro → individual product docs), and the new sidebar entry is placed at the top as the logical entry point for new users. |
| SVG quality | The architecture SVG is hand-crafted with proper semantic layer organization, clean color coding, directional arrows with markers, a legend, and both a forward-flow path and a return-path — providing genuine visual value over a placeholder image. |
| Version bump | The package.json version increment (1.2.9 → 1.2.10) correctly accompanies the documentation content release. |
Checklist Results
| Category | Items Checked | Pass | Fail | N/A | Notes |
|---|---|---|---|---|---|
| Correctness | 8 | 6 | 1 | 1 | Removed restart instructions for CLI methods (MJ-01) |
| Security | 10 | 9 | 0 | 1 | No hardcoded secrets; password guidance is present and appropriate |
| Performance | 7 | 0 | 0 | 7 | No runtime code changed |
| Code Quality | 10 | 7 | 0 | 3 | Minor inconsistencies in install commands and skill counts |
| Testing | 7 | 0 | 0 | 7 | Documentation only; no test coverage applicable |
| Documentation | 6 | 5 | 0 | 1 | New public features are documented; some duplication noted |
| Compatibility | 5 | 4 | 0 | 1 | Backward-compatible additions only; new sidebar entry uses collapsed: false |
| Observability | 4 | 0 | 0 | 4 | No runtime code changed |
Disclaimer
This is an automated code review. It supplements but does not replace human review. The reviewer analyzed only the diff between the specified branches. Runtime behavior, integration testing, and deployment impact are not covered. As this PR contains exclusively documentation changes, security, performance, and testing categories are not applicable to the runtime application.
Report generated by Code Review Skill v1.0.0
Date: 2026-04-17
2 participants
No description provided.