github-agile

Pass

Audited by Gen Agent Trust Hub on Sep 17, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONPRIVILEGE_ESCALATION
Full Analysis
  • [COMMAND_EXECUTION]: The skill includes multiple Deno scripts that use the Deno.Command API to interact with the system's git and gh (GitHub CLI) installations.
  • Evidence: scripts/gh-audit.ts, scripts/gh-init-project.ts, scripts/gh-sync-context.ts, and scripts/gh-verify.ts all invoke shell commands to perform repository audits, initialize labels/templates, and sync project status.
  • [INDIRECT_PROMPT_INJECTION]: The skill features an attack surface for indirect prompt injection because it ingests data from external GitHub issues and pull requests and writes them into local Markdown files which the agent subsequently reads.
  • Ingestion points: scripts/gh-sync-context.ts fetches issue titles, labels, and bodies using gh issue list and gh api.
  • Boundary markers: The fetched data is written into context/status.md and context/decision-candidates-{date}.md using Markdown headers and code blocks as delimiters.
  • Capability inventory: The skill has shell execution capabilities via its own provided Deno scripts.
  • Sanitization: gh-sync-context.ts truncates issue body previews to 500 characters when generating decision candidate files.
  • [PRIVILEGE_ESCALATION]: The skill documentation provides standard setup guidance for Linux environments that includes the use of administrative privileges.
  • Evidence: SKILL.md contains the instruction sudo apt install gh in the environment setup section for GitHub CLI installation.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 17, 2026, 01:07 PM
Security Audit — agent-trust-hub — github-agile