bootstrap

Pass

Audited by Gen Agent Trust Hub on Mar 23, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it processes untrusted data from the current directory to generate project-specific instructions.
  • Ingestion points: Reads project configuration files such as pubspec.yaml, package.json, build.sbt, Cargo.toml, and go.mod to understand project details.
  • Boundary markers: No explicit delimiters or warnings are used when interpolating project details into the CLAUDE.md template.
  • Capability inventory: Performs file system write operations to create CLAUDE.md and project memory files.
  • Sanitization: No explicit sanitization of content read from project configuration files is mentioned.
  • [DATA_EXFILTRATION]: The skill accesses local paths including ~/git2/claude-config/templates/ and ~/.claude/projects/. However, there are no network operations (e.g., curl, wget, fetch) detected, and data remains within the local agent environment.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 23, 2026, 10:58 AM
Security Audit — agent-trust-hub — bootstrap