ethos
Pass
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill's primary function is to generate local markdown documentation based on code analysis and user interviews. It operates entirely within the local file system and requires user confirmation before writing changes.\n- [INDIRECT_PROMPT_INJECTION]: The skill reads local files (READMEs, source code, prompts) which are technically untrusted inputs. However, the skill implements a mandatory 'Confirm intent' phase where it asks the user focused questions based on the findings rather than blindly trusting the file content. Furthermore, the output is a documentation file (ETHOS.md) that is presented to the user for review, significantly mitigating the risk of instructions embedded in the analyzed code influencing the agent's behavior beyond the documentation task.
Audit Metadata