adr-relationships

Pass

Audited by Gen Agent Trust Hub on Jul 15, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection by processing external data (ADRs) without sanitization.
  • Ingestion points: The skill reads metadata from markdown files located in docs/adrs/*.md.
  • Boundary markers: No specific boundary markers or instructions to ignore embedded commands are present in the processing logic.
  • Capability inventory: The skill utilizes Bash and Grep to perform file system operations based on the ingested data.
  • Sanitization: Metadata extracted from files (e.g., the supersedes field) is passed directly into shell subshells without escaping or validation.
  • [COMMAND_EXECUTION]: The validation logic uses shell scripts to process ADR references. Specifically, the command ls docs/adrs/*-"$num"-*.md uses the variable $num, which is derived directly from the content of ADR files. If an ADR file contains a maliciously crafted reference (e.g., supersedes: ADR-$(command)), the shell will execute the embedded command when the validation script runs.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 15, 2026, 01:40 PM
Security Audit — agent-trust-hub — adr-relationships