chief-of-staff
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and summarize data from various external and untrusted sources into a persistent knowledge base (wiki).\n
- Ingestion points: The instructions in SKILL.md direct the agent to read from 'subagent reports', 'research notes', 'logs from critical running processes', and 'test environment databases'.\n
- Boundary markers: There are no instructions or delimiters provided to ensure the agent ignores potentially malicious instructions embedded within these external data sources.\n
- Capability inventory: The agent is empowered to coordinate subagents, access browsers, read from databases, and write to a persistent wiki directory.\n
- Sanitization: The skill lacks any mechanism or instructions for sanitizing external content before it is processed or stored in the wiki.- [DATA_EXFILTRATION]: The skill explicitly instructs the agent to seek access to sensitive information environments to fulfill its orchestration role.\n
- Evidence: The instructions in SKILL.md state the agent needs 'Logs from critical running processes' and 'Access to test environment databases'.\n
- Context: While these capabilities are requested to facilitate the 'Chief of Staff' persona, they represent a significant surface for the exposure of sensitive system and user information.
Audit Metadata