Install any skill in seconds. Free to start, no credit card required.
Get Started Free →How to find, read, and audit configuration files — includes concrete investigation steps like grepping for env vars, checking for hardcoded secrets, and mapping external service dependencies.
.claude/skills/esurovtsev-config-file-recognition/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-14 | ✗→✓ | ▲ Improved | -19% | 0% |
| case-22 | ✓→✗ | ▼ Worse | -56% | 0% |
| case-10 | ✓→✗ | ▼ Worse | -25% | 0% |
| case-11 | ✓→✗ | ▼ Worse | -21% | 0% |
| case-12 | ✓→✗ | ▼ Worse | -45% | 0% |
When you need to understand how a project is configured, what external services it depends on, what environment variables it requires, or whether there are configuration issues (hardcoded secrets, missing defaults, inconsistent settings).
Use glob_search to locate config files across the project:
glob_search(pattern="**/.env*") — environment filesglob_search(pattern="**/*.json", path="config/") — JSON configglob_search(pattern="**/Dockerfile*") — container configglob_search(pattern="**/*.yaml") or **/*.yml — YAML configglob_search(pattern="**/requirements*.txt") — Python dependenciesglob_search(pattern="**/package.json") — Node.js dependenciesRead in this priority order:
.env.example — safe to read, shows what variables the app expectsconfig/ directory files — JSON/YAML config for services, servers, etc.requirements.txt / package.json — dependencies reveal what services are usedDockerfile / docker-compose.yml — runtime environment, ports, servicesNever read .env files — they may contain real secrets.
Use grep_search to understand how config is consumed:
grep_search(query="os.getenv|os.environ|dotenv") — find env var usage in Pythongrep_search(query="process.env") — find env var usage in JavaScript/Nodegrep_search(query="localhost|127.0.0.1") — find hardcoded local URLsgrep_search(query="port|PORT") — find port configurationgrep_search(query="SECRET|KEY|TOKEN|PASSWORD") — check for sensitive values in codeFrom the config files and grep results, build a picture of:
.env.example and getenv calls)Look for common configuration problems:
.env.example — if code uses env vars but no example file documents themlocalhost or IP addresses that won't work in productionrequirements.txt without version pinsStructure your findings as:
.env files — they contain real secretsrequirements.txt (dependencies) with config.yaml (runtime settings)| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 14,572 | 3,677 | -75% | 1 | 1 | 0% | 2,717 | 1,089 | -60% | 0 | 0 | — |
case-02 | fail→fail | 5,993 | 3,564 | -41% | 1 | 1 | 0% | 1,147 | 1,094 | -5% | 0 | 0 | — |
case-03 | fail→fail | 15,420 | 19,368 | +26% | 1 | 1 | 0% | 2,074 | 1,657 | -20% | 0 | 0 | — |
case-04 | fail→fail | 6,392 | 7,976 | +25% | 1 | 1 | 0% | 265 | 1,163 | +339% | 0 | 0 | — |
case-05 | fail→fail | 10,318 | 8,260 | -20% | 1 | 1 | 0% | 1,683 | 1,102 | -35% | 0 | 0 | — |
case-06 | fail→fail | 12,692 | 4,103 | -68% | 1 | 1 | 0% | 2,773 | 1,081 | -61% | 0 | 0 | — |
case-22 | pass→fail | 10,566 | 5,082 | -52% | 1 | 1 | 0% | 2,307 | 1,014 | -56% | 0 | 0 | — |
case-07 | fail→fail | 19,421 | 3,707 | -81% | 1 | 1 | 0% | 6,156 | 1,025 | -83% | 0 | 0 | — |
case-08 | fail→fail | 2,786 | 3,296 | +18% | 1 | 1 | 0% | 276 | 1,065 | +286% | 0 | 0 | — |
case-09 | fail→fail | 2,998 | 3,856 | +29% | 1 | 1 | 0% | 489 | 1,111 | +127% | 0 | 0 | — |
case-10 | pass→fail | 6,937 | 2,960 | -57% | 1 | 1 | 0% | 1,332 | 995 | -25% | 0 | 0 | — |
case-11 | pass→fail | 7,441 | 4,031 | -46% | 1 | 1 | 0% | 1,395 | 1,101 | -21% | 0 | 0 | — |
case-12 | pass→fail | 10,162 | 3,700 | -64% | 1 | 1 | 0% | 1,915 | 1,059 | -45% | 0 | 0 | — |
case-13 | fail→fail | 9,214 | 4,048 | -56% | 1 | 1 | 0% | 1,843 | 1,093 | -41% | 0 | 0 | — |
case-14 | fail→pass | 8,732 | 2,964 | -66% | 1 | 1 | 0% | 1,808 | 1,466 | -19% | 0 | 0 | — |
case-15 | pass→fail | 11,217 | 3,728 | -67% | 1 | 1 | 0% | 2,084 | 1,098 | -47% | 0 | 0 | — |
case-16 | pass→pass | 6,445 | 4,600 | -29% | 1 | 1 | 0% | 1,275 | 1,749 | +37% | 0 | 0 | — |
case-17 | fail→fail | 12,145 | 21,150 | +74% | 1 | 1 | 0% | 1,613 | 3,134 | +94% | 0 | 0 | — |
case-18 | fail→fail | 5,573 | 3,412 | -39% | 1 | 1 | 0% | 1,111 | 1,019 | -8% | 0 | 0 | — |
case-19 | fail→fail | 12,913 | 3,511 | -73% | 1 | 1 | 0% | 1,779 | 1,023 | -42% | 0 | 0 | — |
case-20 | pass→fail | 7,087 | 4,408 | -38% | 1 | 1 | 0% | 1,698 | 1,015 | -40% | 0 | 0 | — |
case-21 | pass→pass | 9,434 | 5,850 | -38% | 1 | 1 | 0% | 2,341 | 1,955 | -16% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 4 counted toward the lift figure. The other 18 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of -23 percentage points is the difference between those two pass rates over the 4 comparable cases. 8 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.