See it in action
One of 10 test cases the skill fixed. Same model, same prompt, with and without the skill.
PromptWhat is the exact local file path specified as the root directory for the analysis-target repository fixture?
What was checked, without and with the skill| Checked | Without skill | With skill |
|---|
| Identifies `./tests/fixtures/analysis-target` as the root directory. | Missed | Passed |
|---|
What's inside
The instructions your model reads when the skill loads.
SKILL.md74 linessingle file
analysis-target Repository Skill
Use When
-
You are working inside analysis-target.
-
You need repository-specific commands, validation checks, or configuration references.
-
You want evidence-backed onboarding context instead of inferred workflow assumptions.
-
Root Directory: ./tests/fixtures/analysis-target
-
Detected Package Manager: pnpm
Steps
- Review relevant config files first:
.env.example, package.json, vite.config.ts.
- Start code navigation from evidenced directories:
src.
- Use only the detected commands below; do not invent package scripts.
- Before finishing, run the relevant validation commands listed below.
Maintenance Workflow
Before Editing
- Treat repository files as evidence; inspect the relevant source and config before changing behavior.
- Do not trust generated instructions blindly; compare them against the referenced repository files.
Implementation Discipline
- Keep changes scoped to the requested behavior and the files directly needed for that behavior.
- Match local patterns before introducing new abstractions.
- Do not invent package scripts, entrypoints, or environment variables that are not evidenced below.
Validation Ladder
- Run the narrowest relevant detected check first.
- Broaden to additional detected checks when touching shared code, config, or public behavior.
- Report any checks that could not be run.
When Tests Are Missing
- Use the smallest reproducible manual check tied to the changed behavior.
Track record
Benchmark
17 / 20
test cases passed, vs 8 / 20 without
Adoption
- Users
- 0
- Times used
- 0
Safety and source details
SkillSafety
✓ 3 of 3 passed✓Code scan15 of 15 checks clean
0 critical · 0 warning · 0 info — scanner v4 — 8/2/2026
View 15 scanner checks
live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing
✓IntentSafe
LLM judge · does the skill do what it claims
✓ContentClean
LLM judge · policy & harmful-content review
Source
ImportedLicense MITImported 8/2/2026Version v1