researching-with-deepwiki
Pass
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: SAFECREDENTIALS_UNSAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [CREDENTIALS_UNSAFE]: The
scripts/verify.pyscript accesses sensitive local configuration files where the agent platform stores authentication tokens and server configurations. - Evidence: The script iterates through
settings_pathstargeting~/.claude/settings.jsonand~/.config/claude/settings.jsonto read themcpServersconfiguration. - [INDIRECT_PROMPT_INJECTION]: The skill processes data from external, untrusted sources (GitHub, GitLab, and Bitbucket repositories), which can contain malicious instructions intended to influence the agent.
- Ingestion points: Data is retrieved from external code repositories through the
deepwikiMCP tool and incorporated into the agent's context (SKILL.md). - Boundary markers: The skill instructions do not specify any delimiters or warnings to treat the retrieved repository content as untrusted data.
- Capability inventory: The agent typically has capabilities including file system access and shell command execution which could be targeted by an injection.
- Sanitization: There is no evidence of sanitization or filtering of the repository content before it is processed by the AI model.
Audit Metadata