decision-records

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes a bundled bash script (check-decisions.sh) to perform structural validation of records. The script utilizes standard shell utilities such as grep, awk, and sed to parse markdown files locally without external network calls.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to read and interpret markdown content from an ADR directory to answer user questions and maintain consistency. 1. Ingestion points: Markdown records and indices in the target directory provided by the user. 2. Boundary markers: The skill does not employ specific delimiters to separate untrusted document content from the agent's internal instructions. 3. Capability inventory: The skill has permissions for file read/write/edit operations and shell execution via allowed tools. 4. Sanitization: The validator script performs structural parsing to extract metadata but does not sanitize the record body against embedded instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 05:48 PM
Security Audit — agent-trust-hub — decision-records