grace-ask

Pass

Audited by Gen Agent Trust Hub on Sep 17, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest data from local project files (XML artifacts and file-local anchors). This creates an attack surface for indirect prompt injection if the project files being analyzed contain malicious instructions intended to influence the agent's behavior.
  • Ingestion points: Local files at .grace/context/*.xml, .grace/graph/index.xml, .grace/verification/index.xml, .grace/changes/active/C-*, and file-local semantic markers (e.g., MODULE_CONTRACT, LINKS:, START_BLOCK_).
  • Boundary markers: No specific delimiters or "ignore instructions" warnings are defined to separate data from instructions.
  • Capability inventory: The skill specifies reading and analyzing local file content; no network access or shell execution capabilities are defined in the provided files.
  • Sanitization: No sanitization or validation of the ingested content is defined.
  • [NO_CODE]: The skill does not contain any executable scripts, binaries, or external code dependencies. It relies purely on the agent's native ability to read files and follow instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 17, 2026, 03:49 PM
Security Audit — agent-trust-hub — grace-ask