adr-skill
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes project-specific data such as existing code, records, and instructions to generate documentation. While no malicious behavior was detected, this behavior creates a surface area for indirect prompt injection where malicious instructions embedded in the repository's files could influence the agent's output.
- Ingestion points: The skill reads project instructions, decision indexes, and affected code as instructed in
SKILL.md. - Boundary markers: The instructions do not define explicit boundaries or safety markers to distinguish between data and instructions within the ingested content.
- Capability inventory: The skill possesses file-writing capabilities through its included scripts and can emit system events (
system/adr.sync.requested). - Sanitization: There is no evidence of sanitization for data read from the repository before it is processed by the agent.
- [COMMAND_EXECUTION]: The skill relies on local automation scripts to manage the ADR lifecycle. These scripts (
scripts/bootstrap_adr.js,scripts/new_adr.js,scripts/set_adr_status.js) are used to initialize directories, generate files from templates, and update document status using standard Node.js file system modules. Additionally, the skill utilizes a CLI tool for searching records (joelclaw vault adr). All scripts and tools appear to be functional components provided by the skill author.
Audit Metadata