concierge
Pass
Audited by Gen Agent Trust Hub on Sep 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted project metadata from manifest files and documentation to detect capabilities, creating a potential surface for indirect prompt injection.
- Ingestion points: Reads manifest files (package.json, pyproject.toml, go.mod, Cargo.toml, pom.xml, build.gradle, Gemfile, composer.json, pubspec.yaml), directory listings, CI configurations, and repository README content (SKILL.md, Step 2).
- Boundary markers: None identified; the skill directly processes content to form a project snapshot.
- Capability inventory: The skill has access to shell commands (git status), web research tools, browser automation, and subagent spawning (SKILL.md, Step 3).
- Sanitization: No explicit sanitization logic for data extracted from project files is described before it is included in the output snapshot.
Audit Metadata