- name
- agent-audit-log
- description
- Lightweight operational audit logging for AI assistants, agent workspaces, and personal automation systems. Use when you need a structured way to record high-value actions such as installs, config changes, updates, repository operations, external publishing, secret injection, deletions, export-safety checks, and follow-up risks. Also use when designing or improving an audit trail with JSONL logs, risk levels, target indexes, human summaries, and open-item tracking.
Agent Audit Log
Create and maintain a lightweight audit trail for high-value actions.
Core rule
Log only actions that matter for safety, traceability, or later review. Do not turn the audit log into noise.
Default layers
- Raw fact log (
YYYY-MM-DD.jsonl) - Date summary (
index.json) - Target/project index (
by-target.json) - Risk index (
by-risk.json) - Human-readable summary (
latest.md) - Export-safety events (
export_safety_check) - Open items (
open-items.json) - Status transition history (
open-items-history.json)
Read references as needed
- Read
references/schema.mdfor the log schema and event fields. - Read
references/risk-model.mdfor how to classify low / medium / high risk. - Read
references/export-safety.mdbefore logging publish/export actions. - Read
references/open-items.mdwhen tracking unresolved risks or follow-up work. - Read
references/examples.mdwhen you need concrete event, export-safety, or open-item examples.
Use scripts as needed
- Use
scripts/init_audit.shto create the basic audit directory and starter files.
Operating rules
- Do not store plaintext secrets in audit logs.
- Prefer concise, human-readable summaries.
- Record target, result, and non-sensitive references.
- Use
warnwhen something needs attention but did not fail. - Use open items for follow-up risk, not for routine noise.