claude-settings-audit
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONREMOTE_CODE_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill analyzes untrusted repository configuration files (such as package manifests, lock files, and build configs) to generate recommended settings and instructions.
- Ingestion points: The skill uses
ls,find, andcatin Phase 1, 2, and 4 to read local project files and dependency manifests. - Boundary markers: There are no explicit markers or instructions to the agent to ignore embedded instructions within the ingested project files.
- Capability inventory: The skill utilizes filesystem inspection commands (
ls,find,cat) and recommends specific bash commands based on the detected stack. - Sanitization: The instructions explicitly require the redaction of settings or MCP secret values before displaying the generated proposal.
- [REMOTE_CODE_EXECUTION]: The skill suggests the addition of an MCP server for Linear using the official package.
- Evidence: Recommends
npx -y @linear/mcp-server, which fetches and executes code from a well-known service (Linear).
Audit Metadata