synthesis-writer
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes study data from an external CSV file, creating a surface where embedded instructions could influence the resulting synthesis.
- Ingestion points: The script reads data from papers/extraction_table.csv using the read_csv_rows function.
- Boundary markers: No delimiters or explicit 'ignore instructions' markers are used to isolate the data from the synthesis logic.
- Capability inventory: The script performs local file system reads and writes to output/SYNTHESIS.md, with no network or shell execution access.
- Sanitization: The orchestration script does not perform validation or escaping of the CSV content before passing it to the renderer.
- [DYNAMIC_EXECUTION]: The script dynamically modifies the Python search path to resolve its internal dependencies.
- Evidence: scripts/run.py searches for an AGENTS.md file in parent directories to calculate a repo_root and then adds that path to sys.path to enable importing the tooling package.
Audit Metadata