implementation-task-planner
Pass
Audited by Gen Agent Trust Hub on Jul 27, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill processes untrusted external artifacts (product specs, PRDs, architecture plans) to generate task lists, creating an indirect prompt injection surface. [Ingestion points]: External documentation files provided as input (e.g., specs/001-feature/spec.md). [Boundary markers]: The skill lacks explicit delimiters to separate external content from system instructions. [Capability inventory]: The agent is authorized to write and update task files ('tasks.md', 'tasks.json') based on these inputs. [Sanitization]: The skill mitigates risks through 'Readiness Gate' and 'Guardrails' instructions that enforce grounding and prevent the fabrication of details not present in the source material.
- [SAFE]: The repository includes a validation script ('scripts/validate_skill.py') that performs static and local file system checks; it contains no network calls, obfuscation, or sensitive data access.
- [SAFE]: The installation instructions and tool routing hints align with standard development practices and vendor-owned resources.
Audit Metadata