pstack-mode
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes Git and GitHub CLI commands to manage repository state and pull requests. Evidence includes calls to
subprocess.runinscripts/worktree-audit.pyandspawninscripts/watch-pr/github.ts. These operations are limited to standard development tools likegitandghfor repository maintenance and pull request monitoring. - [EXTERNAL_DOWNLOADS]: The skill uses
bun installinscripts/bootstrap.tsto set up its execution environment by fetching standard Node.js/TypeScript dependencies from official registries. It also references its own source repository on GitHub for maintenance and updates. - [INDIRECT_PROMPT_INJECTION]: The skill triages external data such as pull request comments and automated review feedback. The attack surface is mitigated by instructions in
references/bugbot-triage.mdwhich guide the agent to verify automated findings before acting. Ingestion occurs inscripts/watch-pr/github.tsvia the GitHub API, and capabilities include local file writes and command execution. - [DYNAMIC_EXECUTION]: The
scripts/bootstrap.tsscript prepares a local runtime by copying skill scripts to a cache directory and executing them. This is used to ensure dependencies are installed and to run the TypeScript-based tools for PR watching and orchestration.
Audit Metadata