feat: path command, expanded benchmarks, docs updates#121
Merged
carlos-alm merged 10 commits intomainfrom Feb 26, 2026
Merged
Conversation
Contributor
Greptile SummaryThis PR adds A→B symbol pathfinding ( Core changes:
Testing:
Documentation:
The implementation is clean, well-tested, and properly integrated across CLI, programmatic API, and MCP interfaces. Confidence Score: 5/5
Important Files Changed
Flowchart%%{init: {'theme': 'neutral'}}%%
flowchart TD
A[path command entry] --> B[symbolPath]
B --> C[pathData]
C --> D{Find from/to nodes}
D -->|Not found| E[Return error]
D -->|Found| F{Same node?}
F -->|Yes| G[Return 0-hop path]
F -->|No| H[BFS traversal]
H --> I{Direction?}
I -->|Forward| J[source_id → target_id]
I -->|Reverse| K[target_id → source_id]
J --> L[Build neighbor query]
K --> L
L --> M[Queue traversal up to maxDepth]
M --> N{Target found?}
N -->|Yes| O[Reconstruct path]
N -->|No| P[Return not found]
O --> Q[Count alternates]
Q --> R[Return result]
S[CLI: src/cli.js] --> A
T[MCP: src/mcp.js] --> C
U[API: src/index.js] --> C
V[Tests] --> W[queries.test.js]
V --> X[cli.test.js]
V --> Y[mcp.test.js]
Z[Benchmarks] --> AA[benchmark.js]
AA --> AB[Query latency: path]
Last reviewed commit: d55f8ea |
![greptile-apps[bot]](https://avatars.githubusercontent.com/in/867647?s=60&v=4){kind=small}
docs/recommended-practices.md
Outdated
| Enable `--multi-repo` to let the agent query any registered repository, or use `--repos` to restrict access to a specific set of repos. | ||
|
|
||
| The server exposes 17 tools: `query_function`, `file_deps`, `impact_analysis`, `find_cycles`, `module_map`, `fn_deps`, `fn_impact`, `context`, `explain`, `where`, `diff_impact`, `semantic_search`, `export_graph`, `list_functions`, `structure`, `hotspots`, and `list_repos` (multi-repo only). See the [AI Agent Guide MCP reference](./ai-agent-guide.md#mcp-server-reference) for the full tool-to-CLI mapping table. | ||
| The server exposes 20 tools: `query_function`, `file_deps`, `impact_analysis`, `find_cycles`, `module_map`, `fn_deps`, `fn_impact`, `symbol_path`, `context`, `explain`, `where`, `diff_impact`, `semantic_search`, `export_graph`, `list_functions`, `structure`, `hotspots`, `node_roles`, `co_changes`, and `list_repos` (multi-repo only). See the [AI Agent Guide MCP reference](./ai-agent-guide.md#mcp-server-reference) for the full tool-to-CLI mapping table. |
Contributor
There was a problem hiding this comment.
tool count mismatch with README — README says "19 MCP tools" (single-repo default), this says "20 tools" which includes list_repos (multi-repo only)
Suggested change
| The server exposes 20 tools: `query_function`, `file_deps`, `impact_analysis`, `find_cycles`, `module_map`, `fn_deps`, `fn_impact`, `symbol_path`, `context`, `explain`, `where`, `diff_impact`, `semantic_search`, `export_graph`, `list_functions`, `structure`, `hotspots`, `node_roles`, `co_changes`, and `list_repos` (multi-repo only). See the [AI Agent Guide MCP reference](./ai-agent-guide.md#mcp-server-reference) for the full tool-to-CLI mapping table. | |
| The server exposes 19 tools (20 in multi-repo mode): `query_function`, `file_deps`, `impact_analysis`, `find_cycles`, `module_map`, `fn_deps`, `fn_impact`, `symbol_path`, `context`, `explain`, `where`, `diff_impact`, `semantic_search`, `export_graph`, `list_functions`, `structure`, `hotspots`, `node_roles`, `co_changes`, and `list_repos` (multi-repo only). See the [AI Agent Guide MCP reference](./ai-agent-guide.md#mcp-server-reference) for the full tool-to-CLI mapping table. |
Contributor
Author
There was a problem hiding this comment.
Good catch — fixed. Updated to "19 tools (20 in multi-repo mode)" to match README and clarify that list_repos is multi-repo only.
Add `codegraph path <from> <to>` — BFS shortest-path search on the call graph. Given two symbol names, finds the shortest call chain with hop count, intermediate nodes, edge kinds, and alternate path count. Supports --reverse, --max-depth, --kinds, --from-file/--to-file, -T, -j, -k flags. Exposed as symbol_path MCP tool. Impact: 4 functions changed, 3 affected
- Create docs/use-cases/titan-paradigm.md — maps Johannes R.'s multi-agent codebase cleanup architecture (RECON, GAUNTLET, GLOBAL SYNC, STATE MACHINE) to codegraph commands, roadmap items, and post-LLM-integration recommendations - Update roadmap/BACKLOG.md: mark #4 (node classification), #9 (git change coupling), #1 (dead code), #2 (shortest path), #12 (execution flow) as DONE; add 6 new Titan Paradigm-inspired items (#21-#26): composite audit, batch querying, triage priority queue, change validation predicates, graph snapshots, MCP orchestration tools - Update README.md: add roles + co-change to features table, differentiators, commands section, agent template, common flags, comparison table; update MCP tool count 18 → 19 - Update docs/recommended-practices.md: update MCP tool count and tool list, add roles/co-change/path to CLAUDE.md template and developer workflow, add "Understand architectural roles" and "Surface hidden coupling" sections, add co-change step to setup checklist - Add full examples with real output for roles, co-change, and path to docs/examples/CLI.md and docs/examples/MCP.md - Update GitHub repo description with new capabilities
- Restore Phase 3 (Architectural Refactoring) to ROADMAP - Renumber phases 4-8 and all cross-references - Fix MCP tool count per Greptile review
carlos-alm
force-pushed
the
feat/path-command
branch
from
34d4548 to
53f9a83
Compare
Contributor
Author
|
Claude finished @carlos-alm's task —— View job PR Review Complete ✅Todo List:
Phase 1 Root Cause Analysis ✅Problem Being Solved: Adding BFS pathfinding between symbols to trace calling relationships (A calls B calls C). This fills a gap in the query capabilities - users can see direct callers/callees but couldn't trace transitive paths. Why This Approach: The implementation uses proper BFS algorithm with depth limiting, bidirectional search (forward/reverse), and edge-kind filtering. This is the correct approach for shortest-path finding in dependency graphs. Risk Assessment: Low risk - this is additive functionality with no breaking changes to existing APIs. Phase 2 Configuration Change Detection ✅No High-Risk Changes Detected:
This is purely additive functionality with proper testing. Phase 3 Test Integrity ✅Excellent Test Coverage:
Phase 4 Code Quality ✅Excellent Implementation Quality: Security:
Performance:
Code Quality:
Documentation:
Root Cause AnalysisProblem Being Solved: Adding graph pathfinding to enable tracing call chains across codebases - essential for dependency analysis and refactoring workflows. Critical ConcernsNone identified. The implementation follows security best practices, uses efficient algorithms, and has excellent test coverage. Final Recommendation
|
docs/roadmap/BACKLOG.md
Outdated
Comment on lines
+36
to
+46
| | 21 | Composite audit command | Single `codegraph audit <file-or-function>` that combines `explain`, `fn-impact`, and code health metrics into one structured report per function. Core version uses graph data; enhanced version includes Phase 4.4 `risk_score`/`complexity_notes`/`side_effects` when available. Inspired by [Titan Paradigm](../docs/use-cases/titan-paradigm.md) Gauntlet phase. | Orchestration | Each sub-agent in a multi-agent swarm gets everything it needs to assess a function in one call instead of 3-4 — directly reduces token waste and round-trips | ✓ | ✓ | 4 | No | | ||
| | 22 | Batch querying | Accept a list of targets (file or JSON) and return all query results in one JSON payload. Applies to `audit`, `fn-impact`, `context`, and other per-symbol commands. Inspired by [Titan Paradigm](../docs/use-cases/titan-paradigm.md) swarm pattern. | Orchestration | A swarm of 20+ agents auditing different files can be fed from a single orchestrator call instead of N sequential invocations — reduces overhead and enables parallel dispatch | ✓ | ✓ | 4 | No | | ||
| | 23 | Triage priority queue | Single `codegraph triage` command that merges `map` connectivity, `hotspots` fan-in/fan-out, node roles, and optionally git churn + `risk_score` into one ranked audit queue. Inspired by [Titan Paradigm](../docs/use-cases/titan-paradigm.md) RECON phase. | Orchestration | Orchestrating agent gets a single prioritized list of what to audit first — replaces manual synthesis of 3+ commands, saves RECON phase from burning tokens on orientation | ✓ | ✓ | 4 | No | | ||
| | 24 | Change validation predicates | `codegraph check --staged` with configurable predicates: `--no-new-cycles`, `--max-blast-radius N`, `--no-signature-changes`, `--no-boundary-violations`. Returns exit code 0/1 for CI gates and state machines. Inspired by [Titan Paradigm](../docs/use-cases/titan-paradigm.md) STATE MACHINE phase. | CI | Automated rollback triggers without parsing JSON — orchestrators and CI pipelines get first-class pass/fail signals for blast radius, cycles, and contract changes | ✓ | ✓ | 4 | No | | ||
| | 26 | MCP orchestration tools | Expose `audit`, `triage`, and `check` as MCP tools alongside existing tools. Enables multi-agent orchestrators (Claude Code agent teams, custom MCP clients) to run the full Titan Paradigm loop through the MCP protocol without CLI overhead. Inspired by [Titan Paradigm](../docs/use-cases/titan-paradigm.md). | Embeddability | Agents query the graph through MCP with zero CLI overhead — fewer tokens, faster round-trips, native integration with AI agent frameworks | ✓ | ✓ | 4 | No | | ||
| | 5 | TF-IDF lightweight search | SQLite FTS5 + TF-IDF as a middle tier (~50MB) between "no search" and full transformer embeddings (~500MB). Provides decent keyword search with near-zero overhead. Inspired by codexray. | Search | Users get useful search without the 500MB embedding model download; faster startup for small projects | ✓ | ✓ | 3 | No | | ||
| | 13 | Architecture boundary rules | User-defined rules for allowed/forbidden dependencies between modules (e.g., "controllers must not import from other controllers"). Violations flagged in `diff-impact` and CI. Inspired by codegraph-rust, stratify. | Architecture | Prevents architectural decay in CI; agents are warned before introducing forbidden cross-module dependencies | ✓ | ✓ | 3 | No | | ||
| | 15 | Hybrid BM25 + semantic search | Combine BM25 keyword matching with embedding-based semantic search using Reciprocal Rank Fusion. Better recall than either approach alone. Inspired by GitNexus, claude-context-local. | Search | Search results improve dramatically — keyword matches catch exact names, embeddings catch conceptual matches, RRF merges both | ✓ | ✓ | 3 | No | | ||
| | 18 | CODEOWNERS integration | Map graph nodes to CODEOWNERS entries. Show who owns each function, surface ownership boundaries in `diff-impact`. Inspired by CKB. | Developer Experience | `diff-impact` tells agents which teams to notify; ownership-aware impact analysis reduces missed reviews | ✓ | ✓ | 3 | No | | ||
| | 22 | Manifesto-driven pass/fail | User-defined rule engine with custom thresholds (e.g. "cognitive > 15 = fail", "cyclomatic > 10 = fail", "imports > 10 = decompose"). Outputs pass/fail per function/file. Generalizes ID 13 (boundary rules) into a generic rule system. | Analysis | Enables autonomous multi-agent audit workflows (GAUNTLET pattern); CI integration for code health gates with configurable thresholds | ✓ | ✓ | 3 | No | | ||
| | 25 | Graph snapshots | `codegraph snapshot save <name>` / `codegraph snapshot restore <name>` for lightweight SQLite DB backup and restore. Enables orchestrators to checkpoint before refactoring passes and instantly rollback without rebuilding. After Phase 4, also preserves embeddings and semantic metadata. Inspired by [Titan Paradigm](../docs/use-cases/titan-paradigm.md) STATE MACHINE phase. | Orchestration | Multi-agent workflows get instant rollback without re-running expensive builds or LLM calls — orchestrator checkpoints before each pass and restores on failure | ✓ | ✓ | 3 | No | |
Contributor
There was a problem hiding this comment.
Duplicate item IDs — new items 21-26 collide with existing items 21-22
The backlog already had:
- ID 21: "Cognitive + cyclomatic complexity" (line 50)
- ID 22: "Manifesto-driven pass/fail" (line 45)
New items need different IDs. Suggest renumbering to 27-32:
- 27: Composite audit command
- 28: Batch querying
- 29: Triage priority queue
- 30: Change validation predicates
- 31: MCP orchestration tools (currently 26)
- 32: Graph snapshots (currently 25)
Contributor
Author
There was a problem hiding this comment.
Fixed in ab57fb6 — renumbered the new orchestration items to IDs 27-32 to avoid colliding with existing IDs 21-22.
| @@ -158,7 +158,7 @@ Full agent setup: [AI Agent Guide](docs/guides/ai-agent-guide.md) · [CLAU | |||
| | 🔍 | **Symbol search** | Find any function, class, or method by name — exact match priority, relevance scoring, `--file` and `--kind` filters | | |||
Contributor
There was a problem hiding this comment.
Tool count is incorrect — says "19 MCP tools" but BASE_TOOLS in src/mcp.js contains 21 tools after adding symbol_path
Should be "21 MCP tools" (22 in multi-repo mode with list_repos)
Contributor
Author
There was a problem hiding this comment.
Fixed — updated all 3 occurrences in README.md from "19-tool" to "21-tool" (lines 100, 147, 173). BASE_TOOLS has 21 tools, 22 in multi-repo with list_repos.
docs/guides/recommended-practices.md
Outdated
| Enable `--multi-repo` to let the agent query any registered repository, or use `--repos` to restrict access to a specific set of repos. | ||
|
|
||
| The server exposes 17 tools: `query_function`, `file_deps`, `impact_analysis`, `find_cycles`, `module_map`, `fn_deps`, `fn_impact`, `context`, `explain`, `where`, `diff_impact`, `semantic_search`, `export_graph`, `list_functions`, `structure`, `hotspots`, and `list_repos` (multi-repo only). See the [AI Agent Guide MCP reference](./ai-agent-guide.md#mcp-server-reference) for the full tool-to-CLI mapping table. | ||
| The server exposes 19 tools (20 in multi-repo mode): `query_function`, `file_deps`, `impact_analysis`, `find_cycles`, `module_map`, `fn_deps`, `fn_impact`, `symbol_path`, `context`, `explain`, `where`, `diff_impact`, `semantic_search`, `export_graph`, `list_functions`, `structure`, `hotspots`, `node_roles`, `co_changes`, and `list_repos` (multi-repo only). See the [AI Agent Guide MCP reference](./ai-agent-guide.md#mcp-server-reference) for the full tool-to-CLI mapping table. |
Contributor
There was a problem hiding this comment.
Tool count is incorrect — says "19 tools (20 in multi-repo mode)" but should be "21 tools (22 in multi-repo mode)"
BASE_TOOLS array in src/mcp.js has 21 tools (including the newly added symbol_path), plus list_repos in multi-repo mode makes 22
Contributor
Author
There was a problem hiding this comment.
Fixed in ab57fb6 — now reads "21 tools (22 in multi-repo mode)" which matches the actual BASE_TOOLS count.
.claude/skills/dogfood/SKILL.md
Outdated
| - What testing approaches am I missing? | ||
| - **Cross-command pipelines:** Have I tested `build` → `embed` → `search` → modify → `build` → `search`? Have I tested `watch` detecting changes then `diff-impact`? | ||
| - **MCP server:** Have I tested the `mcp` command? Initialize via JSON-RPC on stdin, send `tools/list`, verify all 17 tools are present. Test single-repo mode (default — `list_repos` should be absent, no `repo` parameter on tools) vs `--multi-repo` mode. | ||
| - **MCP server:** Have I tested the `mcp` command? Initialize via JSON-RPC on stdin, send `tools/list`, verify all 18 tools are present. Test single-repo mode (default — `list_repos` should be absent, no `repo` parameter on tools) vs `--multi-repo` mode. |
Contributor
There was a problem hiding this comment.
Tool count is incorrect — says "18 tools" but should be "21 tools"
BASE_TOOLS in src/mcp.js contains 21 tools after adding symbol_path in this PR
Contributor
Author
There was a problem hiding this comment.
Fixed in ab57fb6 — now reads "21 tools" matching BASE_TOOLS count in src/mcp.js.
Address Greptile review comments on #121: - Update MCP tool counts from 18/19 to 21 (22 in multi-repo mode) across README, recommended-practices, dogfood skill, titan-paradigm - Add missing execution_flow and list_entry_points to tool enumeration - Renumber new backlog items 21-26 → 27-32 to avoid collision with existing items 21-22
|
Claude finished @carlos-alm's task —— View job Claude finished @carlos-alm's task —— View job PR Review Complete ✅Todo List:
Phase 1 Root Cause Analysis ✅Problem Being Solved: Adding BFS pathfinding between symbols to trace calling relationships (A calls B calls C). This fills a gap in the query capabilities - users can see direct callers/callees but couldn't trace transitive paths. Why This Approach: The implementation uses proper BFS algorithm with depth limiting, bidirectional search (forward/reverse), and edge-kind filtering. This is the correct approach for shortest-path finding in dependency graphs. Risk Assessment: Low risk - this is additive functionality with no breaking changes to existing APIs. Phase 2 Configuration Change Detection ✅No High-Risk Changes Detected:
This is purely additive functionality with proper testing. Phase 3 Test Integrity ✅Excellent Test Coverage:
Phase 4 Code Quality ✅Excellent Implementation Quality: Security:
Performance:
Code Quality:
Documentation:
Root Cause AnalysisProblem Being Solved: Adding graph pathfinding to enable tracing call chains across codebases - essential for dependency analysis and refactoring workflows. Critical ConcernsNone identified. The implementation follows security best practices, uses efficient algorithms, and has excellent test coverage. Final Recommendation
|
Adds a benchmark suite that measures how much codegraph reduces token usage when AI agents navigate the Next.js codebase (~4k TS files). - scripts/token-benchmark-issues.js: 5 real Next.js PRs as test cases - scripts/token-benchmark.js: runner using Claude Agent SDK (baseline vs codegraph MCP), with --perf flag for build/query benchmarks - scripts/update-token-report.js: JSON → markdown report generator - docs/benchmarks/: methodology docs and placeholder report Impact: 21 functions changed, 7 affected
|
Claude finished @carlos-alm's task —— View job Claude finished reviewing this PR ✅ PR Review Complete
Phase 1: Root Cause Analysis ✅Problem Being Solved: Adding BFS pathfinding between symbols to trace calling relationships (A calls B calls C). This fills a gap in the query capabilities - users can see direct callers/callees but couldn't trace transitive paths. Why This Approach: The implementation uses proper BFS algorithm with depth limiting, bidirectional search (forward/reverse), and edge-kind filtering. This is the correct approach for shortest-path finding in dependency graphs. Risk Assessment: Low risk - this is additive functionality with no breaking changes to existing APIs. Phase 2: Configuration Change Detection ✅No High-Risk Changes Detected:
This is purely additive functionality with proper testing. Phase 3: Test Integrity ✅Excellent Test Coverage:
Phase 4: Code Quality ✅Excellent Implementation Quality: Security:
Performance:
Code Quality:
Documentation:
Issues Identified (Same as Greptile)1. Duplicate Backlog IDs: ✅ CONFIRMED
2. MCP Tool Count Inaccuracies: ✅ CONFIRMED
Root Cause AnalysisProblem Being Solved: Adding graph pathfinding to enable tracing call chains across codebases - essential for dependency analysis and refactoring workflows. Critical ConcernsMinor Documentation Issues Only: Two instances in README.md incorrectly state "19-tool MCP server" instead of "21-tool". These are simple documentation accuracy issues with no runtime impact. Final Recommendation
|
Contributor
Author
Added: Token Savings BenchmarkNew commit New files:
Usage: |
…verage benchmark.js now measures no-op rebuilds, 1-file rebuilds, and query latency (fn-deps, fn-impact, path, roles) alongside full builds. update-benchmark-report.js renders new Incremental Rebuilds and Query Latency sections in BUILD-BENCHMARKS.md and adds incremental/query rows to the README performance table. All new fields are additive for backward compatibility. Impact: 5 functions changed, 2 affected
|
Claude finished @carlos-alm's task —— View job Claude finished reviewing this PR ✅ PR Review Complete
Phase 1: Root Cause Analysis ✅Problem Being Solved: Adding BFS pathfinding between symbols to trace calling relationships (A calls B calls C). This fills a gap in the query capabilities - users can see direct callers/callees but couldn't trace transitive paths. Why This Approach: The implementation uses proper BFS algorithm with depth limiting, bidirectional search (forward/reverse), and edge-kind filtering. This is the correct approach for shortest-path finding in dependency graphs. Risk Assessment: Low risk - this is additive functionality with no breaking changes to existing APIs. Phase 2: Configuration Change Detection ✅No High-Risk Changes Detected:
This is purely additive functionality with proper testing. Phase 3: Test Integrity ✅Excellent Test Coverage:
Phase 4: Code Quality ✅Excellent Implementation Quality: Security:
Performance:
Code Quality:
Documentation:
Issues Identified (Same as Greptile)1. Duplicate Backlog IDs: ✅ CONFIRMED
2. MCP Tool Count Inaccuracies: ✅ CONFIRMED
Root Cause AnalysisProblem Being Solved: Adding graph pathfinding to enable tracing call chains across codebases - essential for dependency analysis and refactoring workflows. Critical ConcernsMinor Documentation Issues Only: Two instances in README.md incorrectly state "19-tool MCP server" instead of "21-tool". These are simple documentation accuracy issues with no runtime impact. Final Recommendation
|
carlos-alm
changed the title
This was referenced Feb 26, 2026
Extract version from benchmark result JSON and include it in branch names, commit messages, PR titles, and PR bodies across all 4 benchmark jobs (build, embedding, query, incremental).
Contributor
Author
Addressing Review FeedbackGreptile findings — all resolved:
Claude review — acknowledged:
Remaining: merge conflicts in |
Resolve conflicts: - benchmark.yml: keep version-in-branch-name convention - README.md: take main's dual-engine performance table - BUILD-BENCHMARKS.md: main's dual-engine data + incremental/query sections - update-benchmark-report.js: take main's clearer regex comment Impact: 12 functions changed, 13 affected
Contributor
Author
Take main's latest benchmark numbers (PR #125) while preserving the new incremental rebuild and query latency fields from the branch.
|
Claude finished @carlos-alm's task —— View job PR Review Complete
Phase 1: Root Cause Analysis ✅Problem Being Solved: Adding BFS pathfinding between symbols to trace calling relationships (A calls B calls C). This fills a gap in the query capabilities - users can see direct callers/callees but couldn't trace transitive paths. Why This Approach: The implementation uses proper BFS algorithm with depth limiting, bidirectional search (forward/reverse), and edge-kind filtering. This is the correct approach for shortest-path finding in unweighted graphs. Risk Assessment: Low risk - this is additive functionality with no breaking changes to existing APIs. Phase 2: Configuration Change Detection ✅No High-Risk Changes Detected:
This is purely additive functionality with proper testing. Phase 3: Test Integrity ✅Excellent Test Coverage:
Phase 4: Code Quality ✅Excellent Implementation Quality: Security:
Performance:
Code Quality:
Documentation:
Issues Identified (Same as Greptile)1. Duplicate Backlog IDs: ✅ CONFIRMED
2. MCP Tool Count Inaccuracies: ✅ CONFIRMED
Root Cause AnalysisProblem Being Solved: Adding graph pathfinding to enable tracing call chains across codebases - essential for dependency analysis and refactoring workflows. Critical ConcernsMinor Documentation Issues Only: Two instances in README.md incorrectly state "19-tool MCP server" instead of "21-tool". These are simple documentation accuracy issues with no runtime impact. Final Recommendation
|
carlos-alm
added a commit
that referenced
this pull request
Feb 26, 2026
* feat: add codegraph path for A→B symbol pathfinding Add `codegraph path <from> <to>` — BFS shortest-path search on the call graph. Given two symbol names, finds the shortest call chain with hop count, intermediate nodes, edge kinds, and alternate path count. Supports --reverse, --max-depth, --kinds, --from-file/--to-file, -T, -j, -k flags. Exposed as symbol_path MCP tool. Impact: 4 functions changed, 3 affected * docs: add Titan Paradigm use case, update docs with roles/co-change/path - Create docs/use-cases/titan-paradigm.md — maps Johannes R.'s multi-agent codebase cleanup architecture (RECON, GAUNTLET, GLOBAL SYNC, STATE MACHINE) to codegraph commands, roadmap items, and post-LLM-integration recommendations - Update roadmap/BACKLOG.md: mark #4 (node classification), #9 (git change coupling), #1 (dead code), #2 (shortest path), #12 (execution flow) as DONE; add 6 new Titan Paradigm-inspired items (#21-#26): composite audit, batch querying, triage priority queue, change validation predicates, graph snapshots, MCP orchestration tools - Update README.md: add roles + co-change to features table, differentiators, commands section, agent template, common flags, comparison table; update MCP tool count 18 → 19 - Update docs/recommended-practices.md: update MCP tool count and tool list, add roles/co-change/path to CLAUDE.md template and developer workflow, add "Understand architectural roles" and "Surface hidden coupling" sections, add co-change step to setup checklist - Add full examples with real output for roles, co-change, and path to docs/examples/CLI.md and docs/examples/MCP.md - Update GitHub repo description with new capabilities * docs: restore Architecture Refactoring phase, fix references - Restore Phase 3 (Architectural Refactoring) to ROADMAP - Renumber phases 4-8 and all cross-references - Fix MCP tool count per Greptile review * fix: correct MCP tool counts and backlog ID collisions Address Greptile review comments on #121: - Update MCP tool counts from 18/19 to 21 (22 in multi-repo mode) across README, recommended-practices, dogfood skill, titan-paradigm - Add missing execution_flow and list_entry_points to tool enumeration - Renumber new backlog items 21-26 → 27-32 to avoid collision with existing items 21-22 * feat: add token savings benchmark (codegraph vs raw navigation) Adds a benchmark suite that measures how much codegraph reduces token usage when AI agents navigate the Next.js codebase (~4k TS files). - scripts/token-benchmark-issues.js: 5 real Next.js PRs as test cases - scripts/token-benchmark.js: runner using Claude Agent SDK (baseline vs codegraph MCP), with --perf flag for build/query benchmarks - scripts/update-token-report.js: JSON → markdown report generator - docs/benchmarks/: methodology docs and placeholder report Impact: 21 functions changed, 7 affected * feat: extend benchmarks with incremental builds and expanded query coverage benchmark.js now measures no-op rebuilds, 1-file rebuilds, and query latency (fn-deps, fn-impact, path, roles) alongside full builds. update-benchmark-report.js renders new Incremental Rebuilds and Query Latency sections in BUILD-BENCHMARKS.md and adds incremental/query rows to the README performance table. All new fields are additive for backward compatibility. Impact: 5 functions changed, 2 affected * ci: include version in automated benchmark commits and PRs Extract version from benchmark result JSON and include it in branch names, commit messages, PR titles, and PR bodies across all 4 benchmark jobs (build, embedding, query, incremental). * fix: update remaining 19-tool references to 21-tool in README * docs: remove "viral" from titan paradigm LinkedIn reference --------- Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
carlos-alm
added a commit
that referenced
this pull request
Feb 26, 2026
* feat: add codegraph path for A→B symbol pathfinding Add `codegraph path <from> <to>` — BFS shortest-path search on the call graph. Given two symbol names, finds the shortest call chain with hop count, intermediate nodes, edge kinds, and alternate path count. Supports --reverse, --max-depth, --kinds, --from-file/--to-file, -T, -j, -k flags. Exposed as symbol_path MCP tool. Impact: 4 functions changed, 3 affected * docs: add Titan Paradigm use case, update docs with roles/co-change/path - Create docs/use-cases/titan-paradigm.md — maps Johannes R.'s multi-agent codebase cleanup architecture (RECON, GAUNTLET, GLOBAL SYNC, STATE MACHINE) to codegraph commands, roadmap items, and post-LLM-integration recommendations - Update roadmap/BACKLOG.md: mark #4 (node classification), #9 (git change coupling), #1 (dead code), #2 (shortest path), #12 (execution flow) as DONE; add 6 new Titan Paradigm-inspired items (#21-#26): composite audit, batch querying, triage priority queue, change validation predicates, graph snapshots, MCP orchestration tools - Update README.md: add roles + co-change to features table, differentiators, commands section, agent template, common flags, comparison table; update MCP tool count 18 → 19 - Update docs/recommended-practices.md: update MCP tool count and tool list, add roles/co-change/path to CLAUDE.md template and developer workflow, add "Understand architectural roles" and "Surface hidden coupling" sections, add co-change step to setup checklist - Add full examples with real output for roles, co-change, and path to docs/examples/CLI.md and docs/examples/MCP.md - Update GitHub repo description with new capabilities * docs: restore Architecture Refactoring phase, fix references - Restore Phase 3 (Architectural Refactoring) to ROADMAP - Renumber phases 4-8 and all cross-references - Fix MCP tool count per Greptile review * fix: correct MCP tool counts and backlog ID collisions Address Greptile review comments on #121: - Update MCP tool counts from 18/19 to 21 (22 in multi-repo mode) across README, recommended-practices, dogfood skill, titan-paradigm - Add missing execution_flow and list_entry_points to tool enumeration - Renumber new backlog items 21-26 → 27-32 to avoid collision with existing items 21-22 * feat: add token savings benchmark (codegraph vs raw navigation) Adds a benchmark suite that measures how much codegraph reduces token usage when AI agents navigate the Next.js codebase (~4k TS files). - scripts/token-benchmark-issues.js: 5 real Next.js PRs as test cases - scripts/token-benchmark.js: runner using Claude Agent SDK (baseline vs codegraph MCP), with --perf flag for build/query benchmarks - scripts/update-token-report.js: JSON → markdown report generator - docs/benchmarks/: methodology docs and placeholder report Impact: 21 functions changed, 7 affected * feat: extend benchmarks with incremental builds and expanded query coverage benchmark.js now measures no-op rebuilds, 1-file rebuilds, and query latency (fn-deps, fn-impact, path, roles) alongside full builds. update-benchmark-report.js renders new Incremental Rebuilds and Query Latency sections in BUILD-BENCHMARKS.md and adds incremental/query rows to the README performance table. All new fields are additive for backward compatibility. Impact: 5 functions changed, 2 affected * ci: include version in automated benchmark commits and PRs Extract version from benchmark result JSON and include it in branch names, commit messages, PR titles, and PR bodies across all 4 benchmark jobs (build, embedding, query, incremental). * fix: update remaining 19-tool references to 21-tool in README * docs: remove "viral" from titan paradigm LinkedIn reference * fix: use endLine for scope-aware caller selection in nested functions Nested/closure functions (e.g. nodeId inside exportMermaid) were incorrectly classified as [dead] because the caller selection loop picked the last definition where line <= call.line, creating self-call edges that got filtered out. Now uses endLine to find the innermost enclosing scope, so calls within an outer function correctly attribute the outer function as caller rather than the nested function itself. Fixes false-positive [dead] for nodeId in branch-compare.js, export.js, and queries.js. Impact: 1 functions changed, 17 affected --------- Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
carlos-alm
added a commit
that referenced
this pull request
Feb 26, 2026
* feat: add codegraph path for A→B symbol pathfinding Add `codegraph path <from> <to>` — BFS shortest-path search on the call graph. Given two symbol names, finds the shortest call chain with hop count, intermediate nodes, edge kinds, and alternate path count. Supports --reverse, --max-depth, --kinds, --from-file/--to-file, -T, -j, -k flags. Exposed as symbol_path MCP tool. Impact: 4 functions changed, 3 affected * docs: add Titan Paradigm use case, update docs with roles/co-change/path - Create docs/use-cases/titan-paradigm.md — maps Johannes R.'s multi-agent codebase cleanup architecture (RECON, GAUNTLET, GLOBAL SYNC, STATE MACHINE) to codegraph commands, roadmap items, and post-LLM-integration recommendations - Update roadmap/BACKLOG.md: mark #4 (node classification), #9 (git change coupling), #1 (dead code), #2 (shortest path), #12 (execution flow) as DONE; add 6 new Titan Paradigm-inspired items (#21-#26): composite audit, batch querying, triage priority queue, change validation predicates, graph snapshots, MCP orchestration tools - Update README.md: add roles + co-change to features table, differentiators, commands section, agent template, common flags, comparison table; update MCP tool count 18 → 19 - Update docs/recommended-practices.md: update MCP tool count and tool list, add roles/co-change/path to CLAUDE.md template and developer workflow, add "Understand architectural roles" and "Surface hidden coupling" sections, add co-change step to setup checklist - Add full examples with real output for roles, co-change, and path to docs/examples/CLI.md and docs/examples/MCP.md - Update GitHub repo description with new capabilities * docs: restore Architecture Refactoring phase, fix references - Restore Phase 3 (Architectural Refactoring) to ROADMAP - Renumber phases 4-8 and all cross-references - Fix MCP tool count per Greptile review * fix: correct MCP tool counts and backlog ID collisions Address Greptile review comments on #121: - Update MCP tool counts from 18/19 to 21 (22 in multi-repo mode) across README, recommended-practices, dogfood skill, titan-paradigm - Add missing execution_flow and list_entry_points to tool enumeration - Renumber new backlog items 21-26 → 27-32 to avoid collision with existing items 21-22 * feat: add token savings benchmark (codegraph vs raw navigation) Adds a benchmark suite that measures how much codegraph reduces token usage when AI agents navigate the Next.js codebase (~4k TS files). - scripts/token-benchmark-issues.js: 5 real Next.js PRs as test cases - scripts/token-benchmark.js: runner using Claude Agent SDK (baseline vs codegraph MCP), with --perf flag for build/query benchmarks - scripts/update-token-report.js: JSON → markdown report generator - docs/benchmarks/: methodology docs and placeholder report Impact: 21 functions changed, 7 affected * feat: extend benchmarks with incremental builds and expanded query coverage benchmark.js now measures no-op rebuilds, 1-file rebuilds, and query latency (fn-deps, fn-impact, path, roles) alongside full builds. update-benchmark-report.js renders new Incremental Rebuilds and Query Latency sections in BUILD-BENCHMARKS.md and adds incremental/query rows to the README performance table. All new fields are additive for backward compatibility. Impact: 5 functions changed, 2 affected * ci: include version in automated benchmark commits and PRs Extract version from benchmark result JSON and include it in branch names, commit messages, PR titles, and PR bodies across all 4 benchmark jobs (build, embedding, query, incremental). * fix: update remaining 19-tool references to 21-tool in README * docs: remove "viral" from titan paradigm LinkedIn reference * fix: use endLine for scope-aware caller selection in nested functions Nested/closure functions (e.g. nodeId inside exportMermaid) were incorrectly classified as [dead] because the caller selection loop picked the last definition where line <= call.line, creating self-call edges that got filtered out. Now uses endLine to find the innermost enclosing scope, so calls within an outer function correctly attribute the outer function as caller rather than the nested function itself. Fixes false-positive [dead] for nodeId in branch-compare.js, export.js, and queries.js. Impact: 1 functions changed, 17 affected * feat: add cognitive & cyclomatic complexity metrics Compute per-function complexity during build via single-traversal DFS of tree-sitter ASTs: cognitive (SonarSource), cyclomatic (McCabe), and max nesting depth. Stores results in new function_complexity table (migration v8) and surfaces them in stats, context, explain, and a dedicated `complexity` CLI command + MCP tool. Adds manifesto config section with warn thresholds (cognitive: 15, cyclomatic: 10, maxNesting: 4) seeding the future rule engine. Phase 1 supports JS/TS/TSX; unsupported languages are skipped gracefully. Impact: 18 functions changed, 32 affected --------- Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
codegraph path: New A→B symbol pathfinding command with BFS traversal, exposed in CLI, programmatic API, and MCPbenchmark.jsnow measures no-op rebuilds, 1-file rebuilds, and query latency (fn-deps, fn-impact, path, roles); report pipeline renders new sections in BUILD-BENCHMARKS.md and READMETest plan
npx vitest run)node scripts/benchmark.jsoutputs JSON withnoopRebuildMs,oneFileRebuildMs, andqueries### Lightweight Footprintsection preserved after update