skills/dafang/codestable/cs-audit/Gen Agent Trust Hub

cs-audit

Pass

Audited by Gen Agent Trust Hub on Jul 22, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to use standard file system search and read tools (such as Grep, Glob, and file read operations) to scan project source code for potential issues.
  • [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection because it reads and processes untrusted data (source code and files within the repository).
  • Ingestion points: Files and directories specified by the user during the Phase 2 scanning process.
  • Boundary markers: None explicitly defined; the agent is directed to read the raw content of files.
  • Capability inventory: The agent can read local files and write audit findings to the .codestable/audits/ directory.
  • Sanitization: No explicit sanitization or filtering of the scanned content is mentioned before processing.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 22, 2026, 12:38 PM
Security Audit — agent-trust-hub — cs-audit