ai-decision-logger
Find model-based decision points and check if they are logged with sufficient context. Trigger: /ai-log
SKILL.mdskills/ai-ml/ai-decision-logger/SKILL.md
GitHub/ai-log
AI decisions must be auditable. This skill finds every model-based decision point and checks logging coverage.
What this is for
- Model output used in conditionals without audit logging
- Classification results used for actions without context
- Approval/recommendation decisions missing human review path
- Read-only skill. No code changes.
PROTECTION RULE - never ~/.claude/
Read-only skill. Guard required if write mode added later.
What You Must Do When Invoked
During analysis, assign a confidence level to each finding: proven (confirmed by evidence), likely (strong signal, needs review), or suspected (weak signal).
If /ai-log -help or /ai-log -h (without further arguments)
is invoked: output the ## Usage section unchanged and stop.
Otherwise follow these steps in order, skipping none.
Step 1 - Help check
If invoked with -help or -h, output the ## Usage section unchanged and stop.
Step 2 - Clarify target
Clarify -ProjectDir. Get confirmation.
Step 3 - Scan
& "<SKILL_DIR>/scripts/ai-decision-scan.ps1" -ProjectDir "<path>"
Step 4 - Classification
For each decision point:
- Auditable: full context logged + human review option
- Partially-logged: some context captured
- Black-box: no log, no audit trail
Step 5 - Write report
File ai-decision-report.md in current working directory:
- Summary - decisions by classification.
- Decision table - black-box first. Per decision: file, line, decision type, has audit log, includes context, has human review, risk.
- Open questions.
Step 6 - Summarize
State report path, highlight black-box decisions.
Usage
/ai-log # interactive
/ai-log <dir> # scan project
/ai-log -help