pi-cli-workspace
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to base its responses on content from the
pi-monorepository, creating a surface where instructions embedded in those external files could influence agent behavior. - Ingestion points: Documentation and source files within the
pi-mono/directory, including READMEs, docs, and implementation scripts (SKILL.md). - Boundary markers: The skill includes instructional constraints to "Answer only from pi-mono/ sources" and to report if information is missing from the corpus, which provides a logical constraint rather than a technical security boundary (SKILL.md).
- Capability inventory: The agent is documented to have access to a suite of system tools including
bash,edit,write,read,grep,find, andls(SKILL.md). - Sanitization: No specific sanitization or verification logic is defined for the content ingested from the repository files before it is processed by the agent.
- [EXTERNAL_DOWNLOADS]: The skill references an external repository at
https://github.com/badlogic/pi-monoas its primary source of truth for grounding information (CORPUS.md).
Audit Metadata