Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 3 additions & 3 deletions .claude-plugin/plugin.json
Original file line number Diff line number Diff line change
Expand Up @@ -3,9 +3,9 @@
"description": "Conversational discovery and install of GitHub agent workflows (gh-aw), with auth setup that respects your Claude subscription.",
"version": "0.1.0",
"author": {
"name": "TBD"
"name": "verkyyi"
},
"homepage": "https://github.com/TBD/github-agent-runner",
"repository": "https://github.com/TBD/github-agent-runner",
"homepage": "https://github.com/verkyyi/github-agent-runner",
"repository": "https://github.com/verkyyi/github-agent-runner",
"license": "MIT"
}
78 changes: 66 additions & 12 deletions skills/discover/catalog.md
Original file line number Diff line number Diff line change
@@ -1,20 +1,74 @@
# Curated workflow catalog

STUB — fill during scope-lock hour (memo: 2026-04-19, task 3).
7 hand-picked entries from [`githubnext/agentics`](https://github.com/githubnext/agentics/tree/main/workflows).
All require Claude auth — see `../install/auth.md` for the OAuth vs API-key decision.
Comment on lines +3 to +4
Copy link

Copilot AI Apr 19, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The catalog’s setup-friction blurbs don’t mention the extra OAuth-only post-compile .lock.yml tweak required by the documented install flow (skills/install/SKILL.md + skills/install/auth.md). As written, a reader could assume “one secret” is sufficient regardless of auth path. Consider noting here that the listed friction assumes the API-key path, and that the OAuth path adds the mandatory lockfile patch step (and re-application after any recompile).

Copilot uses AI. Check for mistakes.

Target: 5-8 entries from the `githubnext/agentics` catalog, chosen because the author would personally recommend them. Quality of recommendation > catalog breadth.
---

## Entry template
## issue-triage

Each entry should capture:
- **Upstream source**: `githubnext/agentics/workflows/issue-triage.md`
- **One-line purpose**: Labels new issues, detects spam, and posts an analysis comment with debugging tips and reproduction steps — automatically, the moment an issue lands.
- **Fits repos that**: receive issues regularly; have a label set already defined; maintainers who are time-poor and want triage handled before they even look at their inbox.
- **Setup friction**: low — one secret (Claude auth), no YAML to edit; customize the prompt body after install if you want custom label logic.
Copy link

Copilot AI Apr 19, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

“No YAML to edit” is inaccurate for the OAuth auth path described in ../install/auth.md, which requires patching the generated .lock.yml after gh aw add. Consider rephrasing to something like “no workflow YAML to author by hand” or explicitly calling out the OAuth-only lockfile tweak.

Suggested change
- **Setup friction**: low — one secret (Claude auth), no YAML to edit; customize the prompt body after install if you want custom label logic.
- **Setup friction**: low — one secret (Claude auth), no workflow YAML to author by hand; customize the prompt body after install if you want custom label logic.

Copilot uses AI. Check for mistakes.
- **Auth requirement**: either

- **Name**: as it appears in `gh aw` CLI
- **Upstream source**: path in `githubnext/agentics` or URL
- **One-line purpose**: what it does for the repo owner
- **Fits repos that**: concrete signals (e.g. "have a `tests/` directory", "use pnpm", "have open issues older than 30 days")
- **Setup friction**: low / medium / high — how many secrets, how much YAML, how many decisions
- **Auth requirement**: needs ANTHROPIC_API_KEY / works with subscription OAuth / either (answer depends on auth research — see `../install/auth.md`)
---

## Candidates to evaluate (not yet committed)
## pr-nitpick-reviewer

TODO: Work through the `agentics` repo and pick 5-8. No guesses here — leaving this empty so future-me doesn't anchor on placeholder names.
- **Upstream source**: `githubnext/agentics/workflows/pr-nitpick-reviewer.md`
- **One-line purpose**: On-demand nitpicky code review (style, naming, complexity, test gaps) triggered by a `/nit` slash command on any PR.
- **Fits repos that**: have active PR flow; want a second set of eyes on style points that linters miss; already do code review but want the small stuff caught automatically.
- **Setup friction**: low — one secret (Claude auth), slash-command trigger so it never runs unsolicited.
- **Auth requirement**: either

---

## markdown-linter

- **Upstream source**: `githubnext/agentics/workflows/markdown-linter.md`
- **One-line purpose**: Runs Super Linter against all Markdown files on a weekday schedule and files a structured issue when violations are found.
- **Fits repos that**: have docs, READMEs, or wiki-style Markdown; care about consistent formatting; want documentation quality tracked automatically.
- **Setup friction**: low — one secret (Claude auth), schedule runs automatically; no extra config needed.
- **Auth requirement**: either

---

## pr-fix

- **Upstream source**: `githubnext/agentics/workflows/pr-fix.md`
- **One-line purpose**: Analyzes CI failures in a PR, implements a fix, and pushes the corrected commit — invoked via `/pr-fix` slash command.
- **Fits repos that**: have CI (tests, linters, type-checks); contributors whose PRs sometimes get stuck on red CI; maintainers who want to unblock PRs without context-switching.
- **Setup friction**: medium — one secret (Claude auth) plus `push-to-pull-request-branch` permission; review the pushed fix before merging.
- **Auth requirement**: either

---

## weekly-issue-summary

- **Upstream source**: `githubnext/agentics/workflows/weekly-issue-summary.md`
- **One-line purpose**: Posts a weekly GitHub Discussion with issue-activity trend charts (opened vs closed, resolution time) and actionable recommendations.
- **Fits repos that**: have a `Discussions` feature enabled and an `audits` category; track issues actively; want a team-visible pulse on backlog health every Monday.
- **Setup friction**: medium — one secret (Claude auth) plus Discussions must be enabled; create an `audits` discussion category if it doesn't exist.
- **Auth requirement**: either

---

## daily-malicious-code-scan

- **Upstream source**: `githubnext/agentics/workflows/daily-malicious-code-scan.md`
- **One-line purpose**: Scans all commits from the last 3 days for secret-exfiltration patterns, out-of-context code, obfuscation, and supply-chain red flags; surfaces findings as GitHub code-scanning alerts.
- **Fits repos that**: accept third-party contributions; care about supply-chain security; want threat detection without paying for a dedicated SAST tool.
- **Setup friction**: low — one secret (Claude auth); findings appear in the Security tab automatically.
- **Auth requirement**: either

---

## daily-repo-status

- **Upstream source**: `githubnext/agentics/workflows/daily-repo-status.md`
- **One-line purpose**: Creates a daily GitHub issue summarizing recent activity (issues, PRs, releases, code changes) with productivity insights and recommended next steps.
- **Fits repos that**: have a small team or solo maintainer who wants a morning briefing; track multiple concurrent workstreams; close older status issues automatically.
- **Setup friction**: low — one secret (Claude auth), fully automatic once installed.
- **Auth requirement**: either