github-agile
Pass
Audited by Gen Agent Trust Hub on Mar 29, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes
gh(GitHub CLI) andgitcommands through Deno'sCommandAPI to perform repository operations, triage issues, and manage pull requests. This logic is implemented inscripts/gh-verify.ts,scripts/gh-audit.ts,scripts/gh-init-project.ts, andscripts/gh-sync-context.ts. - [EXTERNAL_DOWNLOADS]: The documentation provides guidance for installing the official GitHub CLI through standard platform package managers (Homebrew, apt, winget), as documented in the
GH0state intervention section ofSKILL.md. - [PROMPT_INJECTION]: The skill ingests untrusted content from GitHub issues and pull requests to extract decision candidates and update project status files.
- Ingestion points: Issue and PR bodies processed in
scripts/gh-sync-context.tsandscripts/gh-audit.ts. - Boundary markers: Absent; the scripts do not use specific delimiters or warnings to isolate ingested content.
- Capability inventory: Subprocess execution via
ghandgit, and local file system writes within thecontext/and.github/directories. - Sanitization: Absent; external content is parsed via regex or JSON without specific security validation. This represents a surface for indirect prompt injection, which is handled as a low-severity risk due to its necessity for the skill's primary diagnostic purpose.
Audit Metadata