sw-spec
Pass
Audited by Gen Agent Trust Hub on Nov 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from Jira descriptions, snapshots, and user-provided source documents.
- Ingestion points: Untrusted data enters the agent context through Jira snapshots and imported source text as described in the Purpose and Reads/Writes sections of
SKILL.md. - Boundary markers: The
SKILL.mdbody contains clear instructions to the agent to treat imported text as data and use explicit markers for uncertainty. - Capability inventory: The skill possesses file-write capabilities in specific subdirectories and can execute several local Python scripts (
new_story.py,card_gen.py,render_diagram.py,export_feature.py). - Sanitization: The skill contains robust instructions to the agent to ignore embedded commands, flag suspicious text as potential injection, and never propagate secrets.
- [DYNAMIC_EXECUTION]: The diagram rendering system in
scripts/render_diagram.pyandscripts/draw_engine.pyusesimportlib.import_moduleto load logic for different diagram types. - This dynamic loading is verified against a closed, hardcoded whitelist defined in
scripts/draw_types/__init__.py, preventing the loading of arbitrary or malicious modules.
Audit Metadata