JPeetz/agent-skills / skills/development/finding-unknowns/skills/reference-hunt
reference-hunt
VerifiedUse when user points at code and says "like this."
| File | Kind | Size |
|---|---|---|
| CHANGELOG.md | markdown | 382 B |
| SKILL.md | skill md | 2.0 KB |
| evals/evals.json | config | 2.6 KB |
evals/evals.json
View on GitHub{
"skill": "reference-hunt",
"version": "1.0.0",
"cases": [
{
"id": "e1",
"description": "Should trigger: user points at code as the spec",
"input": "Make our billing work exactly like the payments module in stripe-python.",
"shouldTrigger": true,
"expect": "Reads reference as a spec, writes semantics summary, confirms, reimplements semantics not syntax",
"assertions": [
"Produces semantics summary before writing code",
"Gets user confirmation of the summary first",
"Reimplements with native idioms, not transliteration",
"Closes loop: each behavior mapped to where it's honored",
"Respects license (no verbatim copying of incompatible code)"
]
},
{
"id": "e2",
"description": "Should trigger: reimplementing behavior from a vendored folder",
"input": "Port the queue logic from vendor/queue-lib into our Go service.",
"shouldTrigger": true,
"expect": "Extracts semantics first, then reimplements in Go",
"assertions": [
"Asks what specifically to extract (behavior/structure/API shape)",
"Flags anything that won't translate with a proposed equivalent"
]
},
{
"id": "n1",
"description": "Not trigger: no reference given, ordinary change",
"input": "Add pagination to our listings endpoint.",
"shouldTrigger": false,
"expect": "Implements directly",
"assertions": [
"No reference-hunt workflow"
]
},
{
"id": "n2",
"description": "Not trigger: reference is merely illustrative, not a spec",
"input": "Our competitor's app has a nice settings page, can we do something in that spirit?",
"shouldTrigger": false,
"expect": "Treats as loose inspiration, not reference-targeted reimplementation",
"assertions": [
"No semantics summary produced"
]
},
{
"id": "e3",
"description": "Should trigger: license-respecting reimplementation from a vendored library",
"input": "The vendored queue library is MIT-licensed. Port its ordering semantics into our Go service, native idioms.",
"shouldTrigger": true,
"expect": "Produces semantics summary first, then reimplements semantics with native Go idioms respecting the license",
"assertions": [
"Reads the reference and writes a semantics summary before any code",
"Gets user confirmation of the summary",
"Does not transliterate line-by-line",
"Notes the license and avoids incompatible copying",
"Closes the loop listing where each behavior is honored"
]
}
]
}
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| receiving-code-reviewobra/superpowersUse when receiving code review feedback, before implementing suggestions, especially if feedback seems unclear or technically questionable - requires technical rigor and verification, not performative agreement or blind implementation | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| requesting-code-reviewobra/superpowersUse when completing tasks, implementing major features, or before merging to verify work meets requirements | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| using-git-worktreesobra/superpowersUse when starting feature work that needs isolation from current workspace or before executing implementation plans - ensures an isolated workspace exists via native tools or git worktree fallback | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| verification-before-completionobra/superpowersUse when about to claim work is complete, fixed, or passing, before committing or creating PRs - requires running verification commands and confirming output before making any success claims; evidence before assertions always | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |