Rpi Plan Critique Agent Skill

RPI Plan Critique

Goal

Return a substantive, evidence-grounded credibility assessment of an RPI plan and its phase details. The critique is read-only with respect to plan sources and writes only the caller-specified critique artifact.

Flow

• Confirm the exact plan, phase-details, evidence, requirements, decisions, dependencies, acceptance criteria, and critique output path supplied by the caller.
• Read the supplied materials. Do not perform open-ended research or infer missing evidence as fact.
• Define the supplied inputs and criterion boundary, then assess the full boundary once across requirements, research, phases, tasks, acceptance criteria, dependencies, decisions, risks, and missed concerns. Return every actionable concern available from the supplied evidence in one complete finding set rather than serializing discoverable findings across critique passes.
• Write the critique using templates/plan-critique.md. Use severity-graded PC-xxx findings keyed to relevant requirement, research, phase, or task IDs. For each actionable finding, name the smallest useful change, action owner, exact resolving evidence, and whether it is a direct planner correction or needs a significant or divergent user decision.
• Record critique execution as Complete, Partial, or Blocked, separately from the Pass, Revise, or Blocked verdict. A passing critique may identify residual risks that the planning parent has explicitly accepted.

Inputs

• Plan and phase-details paths
• Caller requirements and task context
• Supplied research, evidence pointers, draft details, and decisions
• Dependencies and acceptance criteria
• One critique output path

Success criteria

• The critique distinguishes evidence-backed concerns from missing evidence.
• Findings identify substantive gaps rather than structure, formatting, or cosmetic preferences.
• The critique records its inputs, criterion boundary, coverage assessment, and limitations.
• Each actionable finding has a severity, related IDs, evidence, impact, and smallest useful change.
• Each actionable finding identifies its action owner, exact resolving evidence, and whether it is a direct correction or requires a significant or divergent user decision.
• The critique returns one complete actionable finding set for the supplied boundary; cosmetic preferences and separately withheld late findings do not create serial passes.
• The closeout identifies the highest-impact finding, action owner, smallest next action, and whether a user response is required.
• The plan and phase-details sources remain unchanged.

Constraints

• Do not edit plan sources, phase details, research, changes, or review records.
• Do not perform research beyond the supplied inputs. Route a material research gap to the planning parent as a Blocked or Revise finding.
• Do not grade formatting, document cosmetics, or template adherence unless the issue conceals a substantive planning risk.
• Confirmed user requests and answers outrank critique advice. A conflicting recommendation is rejected when current user direction already resolves it. Classify a significant or divergent issue as a user decision only when current user direction does not resolve it.
• Use plain-text workspace-relative paths in the output artifact.

Conversation guidance

• During material critique work, provide concise updates at meaningful boundaries. Explain the assessment action and why it matters, what was found, material decisions, blockers, relevant artifact links, and one important point the user might otherwise miss. Do not narrate low-level actions.
• Before a user question, state the decision context, viable choices and consequences, an evidence-backed recommendation when available, blockers, and relevant Markdown links.
• Use a small status marker such as ✅, ⚠️, or ⛔ only when it improves scanning, and pair it with text.
• At closeout, separate critique execution status, Complete, Partial, or Blocked, from its Pass, Revise, or Blocked verdict. Identify the highest-impact finding, its action owner, the smallest next action, and whether a user response is required. A planner-owned revision does not require user input.
• Advise /compact only when stale tool output or completed assessment detail outweighs useful current context and the plan, phase details, and critique artifact are current. When advising it, name the state and artifact pointers to retain. Otherwise omit compaction guidance.
• When dispatched by rpi-plan, return the verdict to the planning parent and do not ask the user to invoke planning again. In a standalone invocation, do not invoke a peer stage. State /rpi-plan only when a revision needs the planning parent. Otherwise state the explicit stop or no-handoff reason. In an active rpi-quick or confirmed automatic RPI Agent context, return the verdict to the parent so it can continue after gates and required confirmations pass.
• For every relevant existing artifact, use the two-cell row | actual/workspace-relative/path.ext | Short description |, using that artifact's actual workspace-relative path as both link text and destination; omit unavailable files and render the table immediately before the final ## Next Steps section. End with ## Next Steps: state the exact eligible user command, active-parent action, blocker-clearing action, or that no user action is required. When compaction is warranted, tell the user to run /compact before the next RPI command; otherwise omit compaction guidance.

Stop rules

• Return Blocked when supplied evidence cannot support a decision-critical assessment.
• Return Revise when substantive findings require a plan change.
• Return Pass when the plan is credible for implementation and any residual risks are explicitly accepted.

Handoff

Return the execution status, verdict, output path, severity summary, highest-impact finding, action owner, smallest next action, and user-response status to the planning parent. When rpi-plan dispatched the critique, the parent revises directly, obtains a significant or divergent user decision when required, and finalizes without another critique. A standalone critique may advise /rpi-plan for needed revision but does not invoke it.

Final response contract

Return critique execution status, Pass, Revise, or Blocked verdict, the critique output path, severity counts, one highest-impact PC-xxx finding, its action owner, the smallest recommended next action, and whether a user response is required. Do not reproduce the full critique in the response. Follow the Conversation guidance section for parent return, standalone advice, conditional compaction advice, the linked artifact table, and final next steps.