thkt/probe
skills/probe/SKILL.md
Extract repository spec while detecting bugs, spec gaps, and consistency drift via dual-purpose documentation. OUTCOME.md-axis question-driven exploration with ephemeral output. Do NOT use for code review (use /audit or /polish), feature implementation (use /code), planning only (use /think), or single-bug fix (use /fix).
$ install --global
skillsauthnpx skillsauth add thkt/claude-config probeInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
Security Scan Results
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
3
Scanners Passed
9
Scanners in report
Clean
TrivyContainer and dependency vulnerability scanner
95%
Clean
SemgrepStatic code analysis for vulnerabilities
95%
Clean
mcp-scan (Snyk)Model Context Protocol security validation
95%
Skipped
Snyk (dep)Open source security scanning
50%
Skipped
Socket.devSupply chain security analysis
50%
Skipped
VirusTotalMulti-engine malware detection
50%
Skipped
CrowdStrikeAdvanced threat intelligence
50%
Skipped
OSV-ScannerOpen Source Vulnerability database check
50%
Skipped
OWASP Dep-Check
50%
Last scanned: Jun 1, 2026, 2:31 AM94.2s2 files scanned
SKILL.md
- name:
- probe
- description:
- Extract repository spec while detecting bugs, spec gaps, and consistency drift via dual-purpose documentation. OUTCOME.md-axis question-driven exploration with ephemeral output. Do NOT use for code review (use /audit or /polish), feature implementation (use /code), planning only (use /think), or single-bug fix (use /fix).
- when_to_use:
- 仕様抽出, 一貫性チェック, ムラ検知, ADR drift, repository spec, バグ発掘, spec gap detection, 暗黙知発掘
- allowed-tools:
- Read Write Edit Bash LS Glob Grep
probe
Trace the repository aspect-by-aspect and surface friction points as issue candidates. The act of "trying to explain" exposes contradictions and gaps. Final output is Issues, no severity rating. Single-file output: findings.md.
Inputs
No arguments. Target is the entire repository at current working directory. Scope auto-derived from .claude/OUTCOME.md.
| Input | Required | Source |
| ---------- | -------- | ------------------------------------------------------------------------------ |
| OUTCOME.md | Yes | .claude/OUTCOME.md (generate via rules/core/OUTCOME.md workflow if absent) |
| ADR | Optional | docs/decisions/ |
| Spec | Optional | .claude/workspace/planning/*/spec.md etc. |
Output
File: /tmp/probe-{repo}-{YYYYMMDD}/findings.md (ephemeral)
Template: ${CLAUDE_SKILL_DIR}/templates/findings.md
Process
Step 1: Outcome Loading
Add each item under Behavior / Non-goals / Constraints in .claude/OUTCOME.md to the aspect list. If ADR exists, enumerate headings with status.
Step 2: Pass-by-Aspect Exploration
Sequential per aspect:
- Locate implementation matching the aspect
- Summarize the spec mentally while reading (do not write out)
- Stop when stuck
- Record as issue candidate in findings.md Issues section (inline current / problem / proposal / impact in the body)
- Ask one question if needed
- Move to next aspect
Step 3: Inconsistency Sweep
After each aspect pass, detect via grep.
| Target | Example | | ------------------- | ------------------------------------------------------------- | | Terminology drift | ADR/README term vs implementation symbol vs test assertion | | Concept overlap | Same name carrying different meaning across fields | | Logging hygiene | Redact wrapper vs raw value in log statements | | ADR Call Site Index | Function names listed in ADR still exist / line numbers drift |
Step 4: Question Phase
Ask one question only when stuck. Mark as "unresolved" and proceed if no answer. Durably record in findings.md Open Questions.
Step 5: Self-Reflection
Record skill behavior at the end of findings.md (question count, friction patterns, Positive issue presence, aspect pass breakage).
Issue Categories
| Category | What | | ------------------ | ------------------------------------------------------------------- | | Bug candidate | Implementation flaw leading to misbehavior | | Spec gap | Expected by OUTCOME / Spec but not reflected in implementation | | Non-goals leakage | Implementation crossing OUTCOME.md Non-goals | | Constraints breach | OUTCOME.md Constraints not honored | | Inconsistency | Pattern divergence among same-responsibility code | | ADR drift | ADR Call Site Index / function names / numbers diverge from current | | Implicit knowledge | Exists in code but undocumented | | Positive | Dynamic verification that OUTCOME / ADR / implementation align |
Related Skills
thkt/use-workflow-plan-preview
tools
VerifiedTrustedCommunity
Internal helper for /think Step 11. Renders SOW.md + Spec.md as an integrated Astro view and returns a dev server URL.
9SKILL.mdUpdated Jun 1, 2026
thkt/audit-adr-gaps
development
VerifiedTrustedCommunity
Discover undocumented design decisions and challenge each candidate via critic-design before promotion. Rank by impact and reversibility, produce ADR promotion candidates. Treat each candidate as a position arguing for ADR status, not a fact to be filed. Pairs with audit-adr-drift, which scans existing ADRs for drift against code.
9SKILL.mdUpdated Jun 1, 2026
thkt/audit-adr-drift
development
VerifiedTrustedCommunity
Scan ADR Decision sections against current code and report drift with modification direction and priority. Do NOT use for repos without ADRs (use audit-adr-gaps instead).
9SKILL.mdUpdated Jun 1, 2026
thkt/reflect
data-ai
VerifiedTrustedCommunity
Surface the current session's tentative reflection (realization / judgment / counterfactual) and optionally promote it to the durable knowledge store. Escape hatch for when Stop hook reflection extraction failed.
9SKILL.mdUpdated May 31, 2026