Skills
skills/oaustegard/claude-skills/semantic-grep/Gen Agent Trust Hub

semantic-grep

Pass

Audited by Gen Agent Trust Hub on May 7, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill uses established patterns to access local files and perform API calls to well-known, trusted service providers (Google and Cloudflare). No unauthorized credential access or data exfiltration to unknown domains was found.
  • [INDIRECT_PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by design as it reads external file content. 1. Ingestion points: The load_corpus function in scripts/semantic_grep.py reads data from paths provided at runtime. 2. Boundary markers: None; the skill processes file content as raw text for embedding. 3. Capability inventory: The skill can read local files and make network requests to trusted domains (googleapis.com and cloudflare.com). No command execution or destructive file operations are present. 4. Sanitization: None beyond length truncation for API compatibility.
Audit Metadata
Risk Level
SAFE
Analyzed
May 7, 2026, 04:36 AM
Security Audit — agent-trust-hub — semantic-grep