session-retro
Pass
Audited by Gen Agent Trust Hub on Nov 11, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill requires the execution of two local Python scripts,
scripts/session_log_analyzer.pyandscripts/action_item_linter.py. These scripts are included in the skill package and use only standard libraries to perform data processing and validation without network access. - [INDIRECT_PROMPT_INJECTION]: The skill processes session logs (.jsonl files) which contain untrusted external data. Instructions embedded in these logs could potentially influence the agent's retrospective analysis.
- Ingestion points: Session logs are read by
scripts/session_log_analyzer.py(line 51) and drafted action items are read byscripts/action_item_linter.py(line 64). - Boundary markers: No programmatic boundary markers or delimiters are implemented to separate log content from control instructions during ingestion.
- Capability inventory: The skill allows the agent to modify project-critical files such as
CLAUDE.md,AGENTS.md,Makefile, andpackage.jsonbased on the processed data. - Sanitization: There is no evidence of sanitization or escaping of the input log content before it is processed by the analyzer or presented to the agent.
Audit Metadata