planning-github-issue-tasks

Pass

Audited by Gen Agent Trust Hub on Apr 17, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill implements a robust, multi-stage orchestrator for decomposing GitHub issues into actionable plans.
  • [SAFE]: No remote code execution patterns, external package dependencies, or unauthorized network operations were detected.
  • [SAFE]: File system access is limited to the expected project directory (docs/) for reading snapshots and writing planning artifacts.
  • [SAFE]: Structural validation is performed at every stage boundary by a dedicated validator subagent to ensure output consistency and requirement coverage.
  • [PROMPT_INJECTION]: The skill processes external data in the form of GitHub issue snapshots, which is an inherent surface for indirect prompt injection.
  • Ingestion points: docs/ISSUE_SLUG.md.
  • Boundary markers: The skill uses strict markdown templates with defined headings to structure and isolate external content.
  • Capability inventory: Limited to file I/O within the local docs/ directory; no shell execution or network access identified.
  • Sanitization: Employs multiple validation subagents to verify the structure and presence of required fields, mitigating the risk of malicious input influence.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 17, 2026, 09:13 AM