skills/sd0xdev/sd0x-harness/recap-doc/Gen Agent Trust Hub

recap-doc

Pass

Audited by Gen Agent Trust Hub on Aug 24, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes git operations and internal Node.js scripts for scope detection and redaction. Risk is mitigated by strict path validation against the project root and temporary directory using realpath resolution to prevent path traversal attacks.- [PROMPT_INJECTION]: The skill processes untrusted source code and diff hunks to generate summaries, creating an indirect prompt injection surface.
  • Ingestion points: Files identified in the ScopeReport and their git history (Phase 2).
  • Boundary markers: The prompt template in references/prompt-template.md uses structured section headers and explicit delimiters to isolate evidence from instructions.
  • Capability inventory: File read/write, git command execution, and internal tool calls (/codex-explain).
  • Sanitization: A dedicated security-redact.js script is invoked as a mandatory gate before writing the final output to disk, identifying and masking potential credentials or secrets.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 24, 2026, 01:44 PM
Security Audit — agent-trust-hub — recap-doc