wjs-auditing-project
Pass
Audited by Gen Agent Trust Hub on May 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes multiple bash commands to perform a read-only audit of the local development environment, including git status, branch tracking, stash lists, and system log files (DiagnosticReports). Evidence found in Phase 1 of SKILL.md.
- [EXTERNAL_DOWNLOADS]: The skill uses the
ghcommand-line tool to fetch data from GitHub APIs, including pull request details, CI run conclusions, and failed check logs. Evidence found in Phase 1 sections B and C of SKILL.md. - [PROMPT_INJECTION]: The skill is subject to indirect prompt injection risks.
- Ingestion points: The agent reads pull request titles, bodies, and failed CI logs from GitHub via
gh pr listandgh run viewin SKILL.md Phase 1. - Boundary markers: Absent. No explicit instructions are provided to the model to ignore instructions found within processed external content.
- Capability inventory: The agent has capabilities to edit local files, create/merge pull requests, and tag git releases as described in SKILL.md Phase 2.
- Sanitization: Absent. External content from GitHub and local documentation files is processed without validation or sanitization before interpolation into the agent's context.
Audit Metadata