Agent Skills: Can I Help: interest-routed contribution matching

Can I Help: interest-routed contribution matching

Turn repository state into contributor action. Collect project context, ask the developer what kind of contribution they want, map that interest to the strongest native signals, then return exact file/line recommendations with data-backed rationale and an executable first step.

The invariant: no vague "look around src/". Every recommendation names a file and preferably a line, states why the target matters, explains the local code in 2 to 3 sentences, and gives the first command or edit.

When to Apply / NOT

Apply:

• A contributor asks where to start, what needs work, or what would be a good first issue.
• A maintainer wants to route a helper toward tests, bugs, docs, or cleanup using repo evidence.
• An OSS project has open issues but unclear contribution paths.
• The user wants quick, low-risk cleanup and accepts the verification gate before deletion claims.

NOT:

• The user already named the exact task or issue; solve that task instead.
• The repo cannot be read locally and no public issue tracker is available; report the missing substrate.
• The user asks for maintainer-only triage, release planning, or architectural roadmap work.
• The request is pure project orientation with no contribution decision; produce orientation, not recommendations.

Workflow

1. Resolve target and collect base context. Default target = current repo unless the user supplied a path. Capture the project shape before ranking anything:

   • Manifests: fd '^(package.json|pyproject.toml|Cargo.toml|go.mod|pom.xml|build.gradle|deno.json|bun.lockb|pnpm-lock.yaml|requirements.txt)$' <repo>.
   • Top-level structure: fd --max-depth 3 --type f <repo>; exclude generated/vendor directories.
   • README / contributing docs: fd '^(README|CONTRIBUTING|DEVELOPMENT|HACKING)(\..*)?$' <repo> then read the relevant file ranges.
   • Test roots: fd '(^test$|^tests$|__tests__|spec$|\.test\.|\.spec\.)' <repo>.
   • Build/test commands: derive from manifest scripts, Makefile targets, CI config, or existing docs; mark certainty MEDIUM unless a command is explicitly declared.

2. Collect contribution signals with native recipes. Prefer indexed codegraph when available; otherwise use ast-grep, rg, fd, and git history recipes. Keep every signal as {kind, file, line?, metric, confidence, evidence}.

   • Good-first areas: low blast radius, clear adjacent patterns, nearby tests, recent maintainer activity, low bug density. Codegraph: codegraph_impact on candidate symbols and codegraph_files for local neighborhoods. Fallback: count importers with rg -n 'from .*/<module>|require\(.*/<module>|use .*<module>|import .*<module>' and prefer files with few dependents plus visible neighboring tests.
   • Test gaps: hot source files with no co-changing test file and no nearby test. Native history recipe: git --no-pager log --since='180 days ago' --name-only --format='commit:%H' -- <src-paths>; rank source files by touches, then subtract files whose commits include a matching test|tests|spec|__tests__ path. Certainty HIGH when source churn ≥5 and zero matching test co-change; MEDIUM when no test root exists but naming conventions are unclear.
   • Doc drift: docs with zero or weak code coupling, stale inline identifiers, or examples importing paths that no longer exist. History recipe: git --no-pager log --since='365 days ago' --name-only --format='commit:%H' -- docs README* CONTRIBUTING*; compute doc commits with no source files. Symbol recipe: extract backticked identifiers/import paths from docs, check via codegraph search, else rg -n '<identifier-or-path>' <repo>. Certainty HIGH for broken import/path; MEDIUM for zero code coupling.
   • Bugspots: files repeatedly touched by fix commits. History recipe: git --no-pager log --since='365 days ago' --regexp-ignore-case --grep='fix|bug|regression|crash|panic|race|leak|broken' --name-only --format='commit:%H' -- <repo>; bug-fix rate = fix_touches / max(total_touches, 1). Certainty HIGH when fix_touches ≥3 and rate ≥0.25.
   • Open issues: gh issue list --state open --limit 15 --json number,title,labels. Treat labels bug, good first issue, help wanted, documentation, test, testing, cleanup, refactor as routing hints. If gh fails, mark issue signal unavailable and continue.
   • Slop-deletion candidates: commented-out code, orphan exports, passthrough wrappers, and always-true/always-false conditions. Use AST where possible; never promise zero-behavior cleanup until the slop verification gate passes.

3. Ask the developer's interest: mandatory and first before recommendations. Use the ODIN ask tool, single-select, exactly one Recommended option. Do this even if signals already look obvious.

   Prompt: What kind of contribution do you want to make?

   Options:

   • New to the stack: Recommended when good-first areas or cleanup candidates exist.
   • Experienced: hard problems, bugspots, architecture-adjacent issues.
   • Want to write tests: test gaps and bugspot overlap.
   • Want to fix bugs: bug-labelled issues, bugspots, suspicious conditions.
   • Want to improve docs: stale references, doc drift, documentation issues.
   • Want quick cleanup: verified deletion-only or tightly-contained cleanup.

4. Route interest to signals. Use references/interest-routing.md as the contract. Lead with the strongest non-empty primary signal for the chosen interest; skip empty subsections with one sentence, not a filler apology. If the chosen interest has no supporting signal, say which signal was empty and pivot to the nearest adjacent interest with data.

5. Read before explaining. For every candidate that survives ranking, read the target file range plus enough surrounding code to understand the local pattern. For docs, read the stale doc and the current code target. For tests, read one nearby existing test pattern. Structural claims without a read are Graft.

6. Emit 2 to 5 recommendations. Each recommendation MUST use the four-field shape:

   • What: exact file and line/range, function, issue number, or doc section.
   • Why: data-backed metric: bug-fix rate, test-gap touch count, zero doc coupling, broken symbol lookup, issue label, confidence score.
   • How: 2 to 3 sentences explaining the local code and the contribution shape.
   • First step: exact command/action, e.g. bat -P -p -n src/parser/expr.ts, rg -n 'parseExpr' tests src, gh issue view 42, or edit src/foo.ts:88-93 after reading context.

7. Offer the next depth step. Close with: Want me to walk you through one of these? I can read the target code, outline the exact diff, or draft the PR description.

Ranking Rules

• Prioritize file-level evidence over directory-level evidence.
• Prefer overlap: test gap ∩ bugspot beats standalone test gap; doc drift ∩ open documentation issue beats standalone stale doc.
• Prefer clear first PRs for newcomers: one file, local pattern, existing tests/docs nearby, low dependents.
• Prefer maintainer value for experienced contributors: bugspot with open issue, high-impact symbol, repeated regression area.
• Cap to 5 recommendations; more is choice overload.
• Certainty labels: HIGH = direct file/line evidence plus metric; MEDIUM = strong heuristic but missing one axis; LOW = only issue label or filename hint. Do not present LOW as fact.

Slop Cleanup Gate

Apply references/slop-cleanup-gate.md in full whenever a recommendation is cleanup-shaped AND claims "zero behavior change" — regardless of whether it surfaced from step 3's Want quick cleanup interest, a slop-deletion signal, or a cleanup-labeled issue. A recommendation that makes no such claim never routes through this gate.

Anti-patterns

• Treating GitHub labels as enough evidence when repo signals disagree.
• Dumping raw signal tables instead of translating them into contribution moves.
• Suggesting generated, vendored, lockfile, snapshot, or build-output files.
• Ranking by personal preference instead of stated contributor interest.

Validation Gates

| Gate | Pass Criteria | Blocking | |---|---|---| | Interest asked | ODIN ask single-select used before recommendations | Yes | | Signals collected | At least base context plus one of test gaps / doc drift / bugspots / issues / cleanup candidates | Yes | | File-level target | Every recommendation includes exact file and line/range when available | Yes | | Data-backed why | Every recommendation has a metric, label, confidence, or observed coupling fact | Yes | | Code read | Each recommendation's How is based on a read source/doc range | Yes | | Cleanup verified | Cleanup claims passed caller + context checks | Yes for zero-behavior wording | | Four-field shape | What / Why / How / First step present for each item | Yes | | Go-deeper close | Final question offers walkthrough/diff/PR help | Yes |