project-milestone-tracker
Pass
Audited by Gen Agent Trust Hub on Aug 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests milestone titles, notes, and event descriptions from stakeholders or external documents, storing them in a local JSON file. This content is later reflected in status reports and status checks, creating a surface where maliciously crafted strings could attempt to influence the agent's behavior.
- Ingestion points: Data enters the system via command-line arguments in scripts/pm_track.py and is subsequently read from the specified JSON tracker file.
- Boundary markers: The tracker script utilizes Markdown headers and list formatting to separate tracked items, though no specific instructions are used to tell the agent to ignore embedded commands.
- Capability inventory: The skill maintains capabilities for local file read and write operations for state management, but does not possess network access or dynamic code execution functions.
- Sanitization: No character filtering, escaping, or validation is performed on the input text before it is saved or printed in generated reports.
Audit Metadata