pack-prerequisite-checker

Purpose

Block "half-ready" pack execution: confirm the user's brief, chosen pack/recipe, runtime understanding, and evidence workflow are sufficient to start planners and validators without rework loops.

Quick start

• identify target pack/recipe
• run checklist rows from references/prerequisite-matrix.md
• mark each row Ready / Needs input / Blocked
• only then authorize entry into execution-heavy skills

When to use

• immediately after recipe-fit-assessor + intent-to-3d-brief-writer
• before first scene planner for a pack workflow
• when user jumps mid-pack without status context

When not to use

• Goal or entry point is ambiguous and no pack is chosen yet → use ../blendops-help/SKILL.md (Triages ambiguous entry requests at precedence Rank 1)
• Evaluating which recipe or pack fits the user's goal → use ../recipe-fit-assessor/SKILL.md (Scores candidate recipes/packs at precedence Rank 2 before preflight)
• Workflow is already past preflight with existing stage or artifact inventory → use ../workflow-stage-router/SKILL.md (Routes in-flight workflows at precedence Rank 4)
• Detailed Blender MCP runtime connection preflight → use ../runtime-readiness/SKILL.md (Inspects Path 1 or Path 2 runtime readiness in detail)
• Active MCP bridge port or add-on session conflicts → use ../bridge-conflicts/SKILL.md (Resolves port 9876 and dual-bridge conflicts)
• Debugging Blender scene or runtime execution errors → use ../blender-troubleshooting/SKILL.md (Diagnoses runtime or scene failures)
• Auditing external 3D asset license compliance → use ../asset-license-checker/SKILL.md (Performs dedicated asset license and provenance verification)

Trigger phrases

• "are we ready to start"
• "pre-flight for the pack"
• "what's missing before we build"
• "check prerequisites"

Prerequisites / readiness

• intended pack/recipe named OR explicitly "exploratory"
• user OK receiving Blocked rows

Input schema

Required inputs

Optional inputs

Output schema

Primary output

• row table: Area | Status (Ready / Needs input / Blocked) | Owner action
• go / no-go summary
• one next skill (downstream domain planner if go; upstream remediation skill if no-go)

Evidence / caveat output

Runtime status: Not Run | Attempted | Produced | Verified | Failed
Artifact status: Not Produced | Produced | Verified | Failed
Evidence used: <brief text, doc paths>
Limitations: <unknowns>

Truth-state rules:

• Artifact status defaults to Not Produced.
• States are downgrade-only (never upgrade).
• Preflight gating never executes packs, planners, or runtime, and never produces artifacts (a gate run alone stays Runtime status: Not Run and Artifact status: Not Produced).

Required laws

• ../../laws/evidence-before-done.md
• ../../laws/non-blender-user-language.md
• ../../laws/no-arbitrary-python-interface.md
• ../../laws/official-runtime-only.md

Official runtime boundary

Preflight gate only — never executes packs, domain planners, or Blender runtime. Uses the 2-path + CLI appendix model per docs/runtime-stack-strategy.md:

• Path 1 always implies the Lab MCP add-on inside Blender 5.1+; Connector is never standalone.
• Path 2 and CLI appendix require explicit scope caveats and readiness confirmation via runtime-readiness.
• Defer runtime detail to runtime-readiness, bridge conflicts to bridge-conflicts, and runtime errors to blender-troubleshooting.

Operating procedure

1. Load matrix references/prerequisite-matrix.md and evaluate brief completeness, runtime posture, skill coverage, and evidence habits.
2. Mark each row Ready, Needs input, or Blocked; apply references/blocked-start-playbook.md for any Blocked row.
3. Emit references/pack-readiness-template.md with explicit go/no-go and downgrade-only truth states (gating never produces artifacts).
4. Recommend exactly one next skill: if go, hand off to the target domain planner; if no-go, hand off to the single blocking upstream skill (never run pack execution, runtime debugging, or license checks inside this skill).

Decision tree

Is the user's goal ambiguous or is no recipe/pack selected yet?
  Goal ambiguous → route to blendops-help (Rank 1)
  Goal clear, no recipe/pack selected → route to recipe-fit-assessor (Rank 2)
Has preflight already passed and do stage/artifact inventories already exist?
  Yes → route to workflow-stage-router (Rank 4)
  No → evaluate the 7-row prerequisite matrix (`references/prerequisite-matrix.md`)
Are critical rows (Goal & scope, Deliverables, Evidence habit) all Ready?
  No → verdict = NO-GO; apply `references/blocked-start-playbook.md` and route to the single blocking upstream skill:
    - Brief incomplete → intent-to-3d-brief-writer
    - Conflicting multi-deliverable scope → scope-guard
    - Unchecked third-party asset licenses → asset-license-checker
    - Active MCP bridge collision on port 9876 → bridge-conflicts
  Yes → check whether immediate runtime mutation is requested:
    - Planning-only start → verdict = GO (or GO with Warn on non-critical Needs input); route to target domain scene planner
    - Immediate runtime mutation requested with unverified runtime posture → route to runtime-readiness

Playbooks

Playbook A: Ready-to-plan product hero preflight (GO)

• Score all 7 rows in references/prerequisite-matrix.md (Goal & scope, Deliverables, Asset plan, Quality bar, Runtime posture, Evidence habit, Single-bridge).
• Confirm critical rows are Ready and non-critical runtime rows are either Ready or Needs input (acceptable for pure planning entry).
• Emit Summary: GO using references/pack-readiness-template.md, keep Runtime status: Not Run and Artifact status: Not Produced, and name the single target domain planner (such as ../product-hero-scene-planner/SKILL.md).

Playbook B: Half-baked brief or unprioritized deliverables (NO-GO)

• Mark Goal & scope or Deliverables as Blocked in the prerequisite matrix.
• Follow references/blocked-start-playbook.md: block entry into scene planners, list the exact missing brief slots or conflicting deliverables, and recommend ../intent-to-3d-brief-writer/SKILL.md as the single remediation skill.
• Preserve Runtime status: Not Run and Artifact status: Not Produced.

Playbook C: Mid-pack jump with runtime or bridge confusion (NO-GO)

• When a user jumps directly to pack execution while both Path 1 and Path 2 bridges are enabled on port 9876 (or claims the Anthropic Connector works without Blender 5.1+ and the Lab add-on), mark Runtime posture / Single-bridge as Blocked.
• Route to ../bridge-conflicts/SKILL.md or ../runtime-readiness/SKILL.md before allowing any mutation-dependent pack step.

Mode handling

Text-only mode

Mark runtime rows as Needs input unless user supplied path choice.

Runtime-ready mode

User-declared path must pair with runtime-readiness handoff suggestion.

Blocked runtime mode

Still check non-runtime rows (brief, assets plan, evidence).

Validation checklist

• Matrix covered across brief completeness, runtime posture, skill coverage, and evidence habits
• Every row marked Ready, Needs input, or Blocked with owner action for blocked/missing items
• Explicit go/no-go decision recorded
• Exactly one next skill named from the catalog
• Truth states follow downgrade-only rule with default Not Produced (no artifacts claimed by gate)

Pass / Warn / Fail rubric

Failure handling

• If any critical prerequisite row (Goal & scope, Deliverables, Evidence habit) is Blocked, issue an explicit NO-GO verdict and route to the single upstream remediation skill from references/blocked-start-playbook.md.
• If the user claims Artifact status: Verified before the pack has even started, reject the upgraded truth state, reset Artifact status: Not Produced (downgrade-only), and mark Evidence habit as Blocked until acknowledged.
• If an active MCP port 9876 collision or dual-bridge setup is reported, halt preflight and hand off to ../bridge-conflicts/SKILL.md.
• Never execute domain scene planning, asset license audits, or runtime debugging inside this gate skill.

Troubleshooting

Best practices

• Evaluate every row of references/prerequisite-matrix.md explicitly (Ready, Needs input, or Blocked) rather than giving an informal thumbs-up.
• Keep the gate output strictly decision-focused: one 7-row table, one GO / NO-GO verdict, and one immediate next skill from the catalog.
• Enforce Artifact status: Not Produced and Runtime status: Not Run as the default truth states for preflight gating.
• Always remind users of the single-bridge constraint (Path 1 and Path 2 cannot attach to the same Blender instance simultaneously).

Good examples

• "Pack / recipe: product-hero-v0 — All 7 matrix rows evaluated (Goal & scope: Ready, Deliverables: Ready, Asset plan: Ready, Quality bar: Ready, Runtime posture: Needs input (planning-only), Evidence habit: Ready, Single-bridge: Ready). Summary: GO. Next skill: ../product-hero-scene-planner/SKILL.md. Runtime status: Not Run. Artifact status: Not Produced."
• "Pack / recipe: product-hero-v0 — Goal & scope: Blocked (missing camera framing and hero subject dimensions), Deliverables: Needs input. Summary: NO-GO. Next skill: ../intent-to-3d-brief-writer/SKILL.md. Runtime status: Not Run. Artifact status: Not Produced."

Bad examples

• "Summary: GO — Brief is still vague, we will figure out deliverables and licenses during rendering. Next skills: product-hero-scene-planner, lighting-material-plan, and render-export-evidence." — bypasses blocked critical rows and recommends multiple skills at once.
• "Summary: GO — Preflight passed so Runtime status: Verified and Artifact status: Produced." — fabricates runtime and artifact truth states from a planning gate.
• "I inspected your third-party FBX EULA and cleared the license." — overreaches into asset-license-checker.

User-facing response template

Pack / recipe: <target pack or recipe name>

Prerequisite matrix:
| Area | Status | Notes / owner action |
|---|---|---|
| Goal & scope | <Ready | Needs input | Blocked> | <notes / action> |
| Deliverables | <Ready | Needs input | Blocked> | <notes / action> |
| Asset plan | <Ready | Needs input | Blocked> | <notes / action> |
| Quality bar | <Ready | Needs input | Blocked> | <notes / action> |
| Runtime posture | <Ready | Needs input | Blocked> | <notes / action> |
| Evidence habit | <Ready | Needs input | Blocked> | <notes / action> |
| Single-bridge | <Ready | Needs input | Blocked> | <notes / action> |

Summary: <GO | NO-GO>
Minimum fixes (if NO-GO): <ordered list or none>
Recommended next skill: <single catalog skill path>
Runtime status: Not Run
Artifact status: Not Produced
Evidence used: <brief text and pack/recipe doc paths>
Limitations: <unverified runtime or asset assumptions>

Anti-patterns

• Authorizing Summary: GO when Goal & scope, Deliverables, or Evidence habit is Blocked.
• Recommending three or four downstream skills simultaneously instead of naming one immediate next skill.
• Writing Artifact status: Not Run or upgrading truth states (Produced / Verified) as a side effect of passing preflight.
• Treating the Anthropic Connector as a standalone Path 1 setup that skips the Blender 5.1+ Lab MCP add-on.
• Performing deep MCP diagnostics, scene planning, or legal license review inside pack-prerequisite-checker.

Cross-skill handoff

• [upstream_reference] Recipe selection → ../recipe-fit-assessor/SKILL.md
• [fallback] Brief completeness → ../intent-to-3d-brief-writer/SKILL.md
• [conditional] Runtime detail → ../runtime-readiness/SKILL.md
• [fallback] Bridge conflicts → ../bridge-conflicts/SKILL.md
• [fallback] Runtime debugging → ../blender-troubleshooting/SKILL.md
• [conditional] License checks → ../asset-license-checker/SKILL.md
• [alternative] Domain planner (product hero when Go) → ../product-hero-scene-planner/SKILL.md
• [alternative] Domain planner (product grid when Go) → ../product-grid-scene-planner/SKILL.md
• [alternative] Domain planner (character portrait when Go) → ../character-portrait/SKILL.md
• [alternative] Domain planner (environment establishing shot when Go) → ../environment-shot/SKILL.md
• [alternative] Domain planner (interior architectural when Go) → ../interior-scene/SKILL.md

Non-goals

• Scoring which recipe or pack fits the user's initial idea (use ../recipe-fit-assessor/SKILL.md).
• Routing active mid-flight workflows that already passed preflight (use ../workflow-stage-router/SKILL.md).
• Writing the 8-slot 3D brief, planning scene geometry/lighting, or auditing asset licenses directly.
• Running Blender, configuring MCP hosts, or verifying rendered/exported files.

References

• references/prerequisite-matrix.md
• references/blocked-start-playbook.md
• references/pack-readiness-template.md