maestro-post-mortem
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a Node.js script located at
${CLAUDE_SKILL_DIR}/../../scripts/maestro-post-mortem.jsto generate a session digest. \n- [DYNAMIC_EXECUTION]: The workflow relies on the runtime execution of a local helper script to process ephemeral session logs. \n- [INDIRECT_PROMPT_INJECTION]: The skill analyzesmaestro_session.log.jsonl, which contains logs of subagent inputs and outputs. This introduces a surface where malicious content within the logs could attempt to influence the agent's post-mortem analysis. \n - Ingestion points:
maestro_session.log.jsonl(processed viamaestro-post-mortem.js). \n - Boundary markers: None specified to differentiate log data from instructions. \n
- Capability inventory: Command execution (
node), file system modifications (.claude/maestro.json, handoff templates), and execution of tool-specific update commands (/maestro-update). \n - Sanitization: No sanitization or validation of log content is mentioned.
Audit Metadata