implementation-planner
Pass
Audited by Gen Agent Trust Hub on Sep 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted input from external sources like issue reports or specifications to create plans.\n
- Ingestion points: The skill reads from user requests, issue specifications, and repository documentation as described in the "Gather Enough Evidence" section of
SKILL.md.\n - Boundary markers: The instructions tell the agent to distinguish confirmed repository facts from inferences, but they do not provide technical delimiters to isolate untrusted user input from system instructions.\n
- Capability inventory: The skill is strictly limited to planning functions ("Do not edit files, apply patches, commit, or implement"), which provides a significant safety barrier against any malicious instructions contained in ingested data.\n
- Sanitization: There are no explicit instructions for the validation or sanitization of input data before processing.
Audit Metadata