omp-plan

Skill: OMP-Plan

Metadata

Field	Value
ID	omp-plan
Keywords	plan:, /plan, /omp:plan, /omp-plan, --consensus
Tier	Execution Mode
Source	src/skills/omp-plan.mts

Description

Strategic planning mode with optional interview workflow. Supports four modes: Interview (broad requests), Direct (detailed requests), Consensus (iterative Planner/Architect/Critic loop with RALPLAN-DR structured deliberation), and Review (Critic evaluation of an existing plan). Delays implementation until the plan is approved.

Interface

interface SkillInput {
  trigger: string;
  args: string[];
}

interface SkillOutput {
  status: "ok" | "error";
  message: string;
}

export async function activate(input: SkillInput): Promise<SkillOutput>
export function deactivate(): void

Implementation

Activates the plan strategic planning mode by spawning bin/omp.mjs plan [args]. No persistent resources are maintained.

---

Flags

omp-plan [--direct|--consensus|--review] [--interactive] [--deliberate] <task description>

Flag	Description
--direct	Skip interview, generate plan immediately
--consensus	Run Planner → Architect → Critic loop (RALPLAN-DR)
--review	Critic evaluation of an existing plan
--interactive	Enable user prompts at draft review and final approval steps (consensus mode)
--deliberate	Force deliberate RALPLAN-DR mode: adds pre-mortem + expanded test plan

---

Mode Selection

Mode	Trigger	Behavior
Interview	Default for broad requests	Interactive requirements gathering
Direct	--direct, or detailed request	Skip interview, generate plan directly
Consensus	--consensus, "ralplan"	Planner → Architect → Critic loop until agreement with RALPLAN-DR structured deliberation (short by default, --deliberate for high-risk); add --interactive for user prompts at draft and approval steps
Review	--review, "review this plan"	Critic evaluation of existing plan

---

Interview Mode (broad/vague requests)

1. Classify the request: Broad (vague verbs, no specific files, touches 3+ areas) triggers interview mode
2. Ask one focused question using AskUserQuestion for preferences, scope, and constraints
3. Gather codebase facts first: Before asking "what patterns does your code use?", explore the codebase to find out, then ask informed follow-up questions
4. Build on answers: Each question builds on the previous answer
5. Consult Analyst for hidden requirements, edge cases, and risks
6. Create plan when the user signals readiness: "create the plan", "I'm ready", "make it a work plan"

Question Classification:

Type	Examples	Action
Codebase Fact	"What patterns exist?", "Where is X?"	Explore first, do not ask user
User Preference	"Priority?", "Timeline?"	Ask user via AskUserQuestion
Scope Decision	"Include feature Y?"	Ask user
Requirement	"Performance constraints?"	Ask user

---

Direct Mode (detailed requests)

1. Quick Analysis: Optional brief Analyst consultation
2. Create plan: Generate comprehensive work plan immediately
3. Review (optional): Critic review if requested

---

Consensus Mode (--consensus / "ralplan")

RALPLAN-DR modes: Short (default, bounded structure) and Deliberate (for --deliberate or explicit high-risk requests: auth/security, data migration, destructive/irreversible changes, production incident, compliance/PII, public API breakage). Both modes keep the same Planner → Architect → Critic sequence and the same AskUserQuestion gates.

Consensus mode runs fully automated by default; add --interactive to enable user prompts at draft review (step 2) and final approval (step 6).

Steps

1. Planner creates initial plan and a compact RALPLAN-DR summary before any Architect review. The summary MUST include:

   • Principles (3-5)
   • Decision Drivers (top 3)
   • Viable Options (>=2) with bounded pros/cons for each option
   • If only one viable option remains, an explicit invalidation rationale for alternatives that were rejected
   • In deliberate mode: a pre-mortem (3 failure scenarios) and an expanded test plan covering unit / integration / e2e / observability

2. User feedback (--interactive only): If running with --interactive, MUST use AskUserQuestion to present the draft plan plus the RALPLAN-DR Principles / Decision Drivers / Options summary with these options:

   • Proceed to review — send to Architect and Critic for evaluation
   • Request changes — return to step 1 with user feedback incorporated
   • Skip review — go directly to final approval (step 6) If NOT running with --interactive, automatically proceed to review (step 3).

3. Architect reviews for architectural soundness. Adopt the Architect role and review the plan for:

   • Strongest steelman counterargument (antithesis) against the favored option
   • At least one meaningful tradeoff tension
   • A synthesis path when possible
   • In deliberate mode: explicitly flag principle violations

   Wait for this step to complete before proceeding to step 4. Do NOT run steps 3 and 4 in parallel.

   Invoke the architect agent to review the plan. If using a sub-agent: "Adopt the Architect role and review this plan for architectural soundness. Provide the strongest steelman antithesis against the favored option, at least one real tradeoff tension, and a synthesis path where possible."

4. Critic evaluates against quality criteria. Run only after step 3 is complete.

   • Verify principle-option consistency
   • Fair alternative exploration
   • Risk mitigation clarity
   • Testable acceptance criteria
   • Concrete verification steps
   • In deliberate mode: MUST reject missing/weak pre-mortem or missing/weak expanded test plan

   Invoke the critic agent to evaluate the plan. If using a sub-agent: "Adopt the Critic role and evaluate this plan. Verify principle-option consistency, fair alternative exploration, risk mitigation clarity, testable acceptance criteria, and concrete verification steps. Reject shallow alternatives, driver contradictions, vague risks, or weak verification."

5. Re-review loop (max 5 iterations): If Critic rejects, execute this closed loop: a. Collect all rejection feedback from Architect + Critic b. Pass feedback to Planner to produce a revised plan c. Return to Step 3 — Architect reviews the revised plan d. Return to Step 4 — Critic evaluates the revised plan e. Repeat until Critic approves OR max 5 iterations reached f. If max iterations reached without approval, present the best version to user via AskUserQuestion with note that expert consensus was not reached

6. Apply improvements: When reviewers approve with improvement suggestions, merge all accepted improvements into the plan file before proceeding. Final consensus output MUST include an ADR section with: Decision, Drivers, Alternatives considered, Why chosen, Consequences, Follow-ups. a. Collect all improvement suggestions from Architect and Critic responses b. Deduplicate and categorize the suggestions c. Update the plan file in .omp/plans/ with the accepted improvements d. Note which improvements were applied in a brief changelog section at the end of the plan

7. On Critic approval (--interactive only): If running with --interactive, use AskUserQuestion to present the plan with these options:

   • Approve and implement via team (Recommended) — proceed to implementation via omp:team
   • Approve and execute via ralph — proceed to implementation via omp:ralph
   • Clear context and implement — compact the context window first, then start fresh implementation via omp:ralph with the saved plan file
   • Request changes — return to step 1 with user feedback
   • Reject — discard the plan entirely If NOT running with --interactive, output the final approved plan and stop. Do NOT auto-execute.

8. (--interactive only) User chooses via the structured AskUserQuestion UI (never ask for approval in plain text).

9. (--interactive only) On user approval:

   • Approve and implement via team: Invoke omp:team with the approved plan path from .omp/plans/ as context. Do NOT implement directly.
   • Approve and execute via ralph: Invoke omp:ralph with the approved plan path from .omp/plans/ as context. Do NOT implement directly.
   • Clear context and implement: Compact the context window first, then invoke omp:ralph with the approved plan path.

---

RALPLAN-DR Short Mode Workflow

Used by default in consensus mode. Bounded structure with four mandatory sections:

1. Principles (3-5 guiding constraints for this decision)
2. Decision Drivers (top 3 criteria that will determine the best option)
3. Viable Options (>=2) — for each option:
   • One-sentence description
   • Pros (2-3 bullets)
   • Cons (2-3 bullets)
4. Invalidation rationale (if only one viable option remains after analysis)

---

RALPLAN-DR Deliberate Mode Workflow

Enabled with --deliberate or when the request explicitly signals high risk. Adds to the short mode:

• Pre-mortem: 3 specific failure scenarios (what could go wrong, how to detect it)
• Expanded Test Plan: unit / integration / e2e / observability layers

---

Review Mode (--review)

1. Read plan file from .omp/plans/
2. Adopt the Critic role to evaluate the plan (or invoke critic agent)
3. Return verdict: APPROVED, REVISE (with specific feedback), or REJECT (replanning required)

---

Plan Output Format

Every plan includes:

• Requirements Summary
• Acceptance Criteria (testable, 90%+ concrete)
• Implementation Steps (with file references, 80%+ claims cite file/line)
• Risks and Mitigations
• Verification Steps
• For consensus/ralplan: RALPLAN-DR summary (Principles, Decision Drivers, Options)
• For consensus/ralplan final output: ADR (Decision, Drivers, Alternatives considered, Why chosen, Consequences, Follow-ups)
• For deliberate consensus mode: Pre-mortem (3 scenarios) and Expanded Test Plan (unit/integration/e2e/observability)

Plans are saved to .omp/plans/. Drafts go to .omp/drafts/.

---

Execution Policy

• Auto-detect interview vs direct mode based on request specificity
• Ask one question at a time during interviews — never batch multiple questions
• Gather codebase facts via exploration before asking the user about them
• Plans must meet quality standards: 80%+ claims cite file/line, 90%+ criteria are testable
• Consensus mode runs fully automated by default; add --interactive to enable user prompts
• Consensus mode uses RALPLAN-DR short mode by default; switch to deliberate mode with --deliberate or when the request signals high risk

---

Final Checklist

• [ ] Plan has testable acceptance criteria (90%+ concrete)
• [ ] Plan references specific files/lines where applicable (80%+ claims)
• [ ] All risks have mitigations identified
• [ ] No vague terms without metrics ("fast" → "p99 < 200ms")
• [ ] Plan saved to .omp/plans/
• [ ] In consensus mode: RALPLAN-DR summary includes 3-5 principles, top 3 drivers, and >=2 viable options (or explicit invalidation rationale)
• [ ] In consensus mode final output: ADR section included (Decision / Drivers / Alternatives considered / Why chosen / Consequences / Follow-ups)
• [ ] In deliberate consensus mode: pre-mortem (3 scenarios) + expanded test plan (unit/integration/e2e/observability) included
• [ ] In consensus mode with --interactive: user explicitly approved before any execution; without --interactive: plan output only, no auto-execution
• [ ] CRITICAL — Consensus mode agent calls MUST be sequential, never parallel. Always await Architect review before issuing the Critic evaluation.

---

Escalation and Stop Conditions

• Stop interviewing when requirements are clear enough to plan — do not over-interview
• In consensus mode, stop after 5 Planner/Architect/Critic iterations and present the best version
• Consensus mode without --interactive outputs the final plan and stops; with --interactive, requires explicit user approval before any implementation begins
• If the user says "just do it" or "skip planning", transition directly to omp:ralph for execution — do NOT implement directly in the planning agent
• Escalate to the user when there are irreconcilable trade-offs that require a business decision