rpi-plan-critique
Pass
Audited by Gen Agent Trust Hub on Jul 30, 2026
Risk Level: SAFE
Full Analysis
- Structured File Operations: The skill reads input documents such as plans, phase details, and evidence from the workspace and writes a critique report to a specified output path. These operations are essential to its function as an assessment tool and are performed using workspace-relative paths.
- Data Ingestion and Processing: The skill ingests untrusted data from external planning documents (documented in SKILL.md under Flow step 1). While this creates a potential surface for indirect prompt injection, the skill's instructions include boundary markers that prohibit editing source files or performing open-ended research. It relies on a deterministic template (templates/plan-critique.md) to structure its findings, which reduces the risk of malicious instructions in the input data influencing the agent's core behavior.
- Read-Only Enforcement: The instructions contain explicit constraints that prevent the skill from modifying the original plan sources or evidence. This focus on non-destructive analysis ensures that the planning data remains consistent and is not subject to unauthorized changes during the critique process.
Audit Metadata