Analysis Orchestrator

/lp-do-analysis is the convergence layer between fact-finding and planning.

It owns:

1. Discovery/intake/dedupe
2. Analysis gates
3. Track routing to specialized analysis modules
4. Analysis artifact persistence using external templates
5. Mandatory critique via /lp-do-critique
6. Optional handoff to /lp-do-plan

It does not decompose executable tasks. That is /lp-do-plan's job.

Global Invariants

Operating mode

ANALYSIS ONLY

Allowed actions

• Read/search files and docs.
• Run non-destructive validation commands needed for evidence.
• Edit analysis docs.

Prohibited actions

• Production implementation changes, refactors, migrations, or deploy operations.
• Destructive shell/git commands.
• Task decomposition beyond planning handoff notes.
• Re-opening broad discovery that belongs in /lp-do-fact-find unless the fact-find is insufficient.

Inputs Priority

1. Fact-find handoff packet (docs/plans/<slug>/fact-find.packet.json) when present
2. Fact-find brief (docs/plans/<slug>/fact-find.md)
3. Existing analysis (docs/plans/<slug>/analysis.md)
4. User request/topic/card ID
5. Current repository evidence

Use the packet-first load order from docs/business-os/startup-loop/contracts/do-stage-handoff-packet-contract.md. Read the full fact-find.md only when the packet is insufficient or a path-specific claim needs verification.

Phase 1: Intake and Mode Selection

Determine analysis mode early:

• analysis+auto (default — proceeds to /lp-do-plan after analysis gates pass)
• analysis-only when any of the following apply:
  • --notauto flag passed explicitly by the user
  • user explicitly says "analysis only", "just analyze", "don't plan yet", or equivalent

Phase 2: Discovery and De-duplication

Fast path (argument provided)

• Slug or fact-find path -> open matching fact-find/analysis paths directly.

Discovery path (no argument)

• Scan docs/plans/*/fact-find.md files for entries with Status: Ready-for-analysis.
• Present results as a short table (slug | title | track | effort).
• Ask user to select a slug or topic.

De-duplication rule

Before creating a new analysis file:

• search docs/plans/ for overlapping analysis docs,
• update existing canonical analysis when possible,
• avoid duplicate analysis files for the same feature slug.

Phase 3: Analysis Gates

Populate and evaluate these gates in the analysis doc:

• Evidence Gate
• Option Gate
• Planning Handoff Gate

Evidence Gate

Require:

• fact-find exists and is Status: Ready-for-analysis
• Deliverable-Type, Execution-Track, Primary-Execution-Skill
• ## Outcome Contract
• for Execution-Track: code | mixed, a completed ## Engineering Coverage Matrix in the fact-find
• enough current-state evidence to compare viable approaches

If missing, stop analysis and return to /lp-do-fact-find, or mark Status: Needs-input with explicit blockers.

Option Gate

Require one of:

• at least 2 viable approaches compared explicitly, or
• one viable approach with explicit elimination rationale for rejected/infeasible alternatives

Do not skip option comparison by jumping straight to recommendation.

Planning Handoff Gate

Require:

• chosen approach stated decisively
• rejected options documented with rationale
• planning handoff notes present (validation implications, sequencing constraints, risk transfer)
• for Execution-Track: code | mixed, engineering-coverage implications carried forward explicitly
• ## End-State Operating Model present with area-by-area delivered flow, or explicit None: no material process topology change
• only operator-only questions remain open

If the recommendation is still hedged, analysis is not complete.

Phase 4: Track Classification and Routing

Determine from the fact-find:

• Execution-Track: code | business-artifact | mixed
• Deliverable-Type
• Primary-Execution-Skill and Supporting-Skills

Route to one module only (or mixed pair):

• code -> modules/analyze-code.md
• business-artifact -> modules/analyze-business.md
• mixed -> modules/analyze-mixed.md

Phase 4.5: Decision Boundary

/lp-do-analysis owns approach selection, not task planning.

• Resolve high-level approach choices here.
• Leave executable task design, validation contracts, and dependency sequencing to /lp-do-plan.
• Keep operator questions only for genuine undocumented preference or real-world information the operator holds.

Phase 5: Produce Analysis Artifact with External Template

Use:

• Analysis template: docs/plans/_templates/analysis.md

Rules:

• Compare approaches explicitly.
• Make the recommendation decisive.
• Keep the artifact decision-grade, not task-grade.
• Carry forward the outcome contract from the fact-find without fabricating operator-authored rationale.
• For Execution-Track: code | mixed, fill ## Engineering Coverage Comparison using ../_shared/engineering-coverage-matrix.md.

Phase 5.5: End-State Operating Model (Non-omittable)

Before persist, write ## End-State Operating Model.

This section may be a single line None: no material process topology change only when the chosen approach does not alter any multi-step process, workflow, lifecycle state, CI/deploy/release lane, approval path, or operator runbook.

For process-affecting work, describe the target state area by area:

• current state
• trigger/start condition
• step-by-step end-state flow after the chosen approach lands
• what remains unchanged
• unresolved seams/risks that planning must absorb

If the target process cannot be explained without hiding behind task-level detail, the analysis is not decision-grade yet.

Phase 6: Persist Analysis

Write/update:

• docs/plans/<feature-slug>/analysis.md
• canonical sidecar after validators pass: docs/plans/<feature-slug>/analysis.packet.json

Status policy:

• Default Status: Draft.
• Set Status: Ready-for-planning only when:
  • mode is analysis+auto (default) or user explicitly wants planning handoff now, and
  • analysis gates pass, and
  • no unresolved blocker remains in the chosen approach

Phase 7: Critique Loop (1–3 rounds, mandatory)

After the analysis is persisted (Phase 6), run the critique loop in analysis mode.

Load and follow: ../_shared/critique-loop-protocol.md

Phase 7.5: Deterministic Validation

Run:

scripts/validate-analysis.sh docs/plans/<feature-slug>/analysis.md

For Execution-Track: code | mixed, also run:

scripts/validate-engineering-coverage.sh docs/plans/<feature-slug>/analysis.md

Rules:

• validate-analysis.sh is required for all analyses.
• validate-engineering-coverage.sh is required for Execution-Track: code | mixed.
• If any required validator fails, analysis is not ready for planning.

After required validators pass, generate the stage handoff packet:

scripts/generate-stage-handoff-packet.sh docs/plans/<feature-slug>/analysis.md

After required validators pass, append workflow-step telemetry:

pnpm --filter scripts startup-loop:lp-do-ideas-record-workflow-telemetry -- --stage lp-do-analysis --feature-slug <feature-slug> --module <loaded-module-relative-to-stage-skill> [--module <additional-module>] [--input-path docs/plans/<feature-slug>/fact-find.packet.json] [--input-path docs/plans/<feature-slug>/fact-find.md] --deterministic-check scripts/validate-analysis.sh [--deterministic-check scripts/validate-engineering-coverage.sh]

Rules:

• Record once per materially updated analysis artifact.
• Include the analysis module(s) actually loaded plus the upstream packet and any full upstream artifact paths that were materially loaded as context.
• Include scripts/validate-analysis.sh on every analysis telemetry record, plus scripts/validate-engineering-coverage.sh when required by track.
• Codex token usage is auto-captured when CODEX_THREAD_ID is available.
• Claude token usage is auto-captured via project session logs (sessions-index.json → debug/latest fallback). Explicit --claude-session-id still takes priority when supplied.

Phase 8: Optional Handoff to Plan

If eligible and mode is analysis+auto, invoke /lp-do-plan <feature-slug>.

Completion Messages

Analysis-only:

Analysis complete. Saved to docs/plans/<feature-slug>/analysis.md. Status: <Draft | Ready-for-planning | Needs-input | Infeasible>. Gates: Evidence <Pass/Fail>, Options <Pass/Fail>, Planning handoff <Pass/Fail>. Critique: <N> round(s), final verdict <credible | partially credible | not credible> (score: <X.X>).

Analysis+auto:

Analysis complete and eligible. Saved to docs/plans/<feature-slug>/analysis.md. Status: Ready-for-planning. Gates passed. Critique: <N> round(s), final verdict credible (score: <X.X>). Auto-continuing to /lp-do-plan <feature-slug>.

Quick Checklist

• No duplicate analysis doc created
• Analysis template loaded from file (not inlined)
• Evidence Gate populated and evaluated
• Option Gate populated and evaluated
• Planning Handoff Gate populated and evaluated
• ## End-State Operating Model present (or explicit None: no material process topology change)
• For code/mixed work, ## Engineering Coverage Comparison is complete
• Chosen approach is decisive, not hedged
• Rejected options have explicit rationale
• Only operator-only questions remain open
• validate-analysis.sh passed
• For code/mixed work, validate-engineering-coverage.sh passed
• analysis.packet.json generated after validators pass
• Workflow-step telemetry appended after validators pass
• Critique loop run (1–3 rounds, mandatory)