recipe-fit-assessor

Purpose

Map a user's stated goal to the closest documented BlendOps recipe or pack before expensive planning or runtime work, returning a fit verdict (Strong / Partial / Poor), rationale, explicit gap list, and one recommended next skill. This skill is planning-only and never runs Blender, packs, or runtime.

Quick start

• capture goal + constraints + deliverables
• compare against catalog rows in references/recipe-catalog.md
• assign fit verdict (Strong / Partial / Poor)
• list gap fixes or recommend a different workflow entrypoint

When to use

• user asks "which recipe / pack fits my idea"
• scope drifts between product hero, portrait, interior, web GLB, batch renders
• before locking intent-to-3d-brief-writer output to a pack

When not to use

• Diverging raw creative ideas before a delivery shape exists → use ../blender-brainstorming/SKILL.md (Explores creative directions and clarifies vague ideas before recipe selection)
• General entry routing across BlendOps when the goal itself is unclear → use ../blendops-help/SKILL.md (Handles ambiguous entry triage at precedence Rank 1)
• Checking whether a selected pack's preflight prerequisites are met → use ../pack-prerequisite-checker/SKILL.md (Evaluates preflight readiness once the recipe/pack is already selected)
• Picking the next workflow stage after preflight passed and artifacts already exist → use ../workflow-stage-router/SKILL.md (Routes active in-flight workflows at precedence Rank 4)
• Writing the structured 8-slot 3D brief after recipe fit is settled → use ../intent-to-3d-brief-writer/SKILL.md (Authors the formal brief after recipe fit is determined)
• Detailed scene planning once a domain and brief are locked → use ../product-hero-scene-planner/SKILL.md (Builds the domain-specific scene plan after preflight passes)
• Runtime setup, path selection, or bridge conflict troubleshooting → use ../runtime-path-picker/SKILL.md (Handles MCP runtime selection and setup outside recipe scoring)
• Legal or asset license verification → use ../asset-license-checker/SKILL.md (Audits asset provenance and license compatibility)

Trigger phrases

• "what recipe should I use"
• "does product hero fit"
• "closest workflow for X"
• "which pack"

Prerequisites / readiness

• rough goal statement exists
• intended outputs known (still image, GLB, animation, web, print)
• audience skill level known (non-Blender user assumed default)

Input schema

Required inputs

Optional inputs

Assumptions to confirm

• User accepts switching packs if fit is Poor.
• Recipes under docs/recipes/ and packs under packs/ are the authoritative catalog.

Output schema

Primary output

• Recommended closest documented recipe or pack name
• Fit verdict (Strong | Partial | Poor) + plain-language rationale
• Gap list (what must change or be accepted to reach a Strong fit, or none)
• One recommended next skill from the official catalog (e.g., intent-to-3d-brief-writer or the single best-matching skill)

Evidence / caveat output

Assessing fit is planning-only and never executes runtime or produces artifacts. Truth states are downgrade-only (never upgrade):

• Allowed Runtime states: Not Run | Attempted | Produced | Verified | Failed (always Not Run for fit assessment unless reporting prior state)
• Allowed Artifact states: Not Run | Not Produced | Produced | Verified | Failed (defaults to Not Produced)

Runtime status: Not Run
Artifact status: Not Produced
Evidence used: <documented recipe/pack paths consulted>
Limitations: <catalog coverage gaps or unverified user constraints>

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

Planning-only skill. Does not configure MCP, Blender, or hosts. When mentioning runtime, use the 2-path + CLI appendix model from docs/runtime-stack-strategy.md only as context for whether a workflow expects interactive MCP vs batch CLI — never claim runtime success.

Operating procedure

1. Normalize goal using references/scope-gap-rules.md.
2. Score against each catalog row in references/recipe-catalog.md.
3. Pick Strong fit if any; else best Partial with explicit gaps.
4. Emit recommendation block from references/workflow-recommendation-template.md.
5. Hand off to brief writer or narrow planner per fit.

Decision tree

Is the user's goal still vague or unformed with no deliverable shape?
  Yes → route to blender-brainstorming (creative divergence) or blendops-help (Rank 1 entry triage)
  No → continue
Is a recipe/pack already selected and ready for preflight or stage routing?
  Recipe selected, preflight not run → route to pack-prerequisite-checker (Rank 3)
  Preflight passed, stage/artifact inventory exists → route to workflow-stage-router (Rank 4)
  No recipe/pack selected yet → score candidate entries in `references/recipe-catalog.md`
Does the prompt request multiple incompatible deliverables at once (e.g., studio hero + full interior arch + character rig)?
  Yes → assign Partial or Poor fit, list scope-conflict gaps (`references/scope-gap-rules.md`), and route to blender-recipe-decomposer
  No → evaluate single-domain alignment:
    - Deliverable + style + budget match `product-hero-v0` or documented workflow → Strong fit → recommend `intent-to-3d-brief-writer` (or `pack-prerequisite-checker` if brief is already locked)
    - Domain matches a specialized planner (portrait, interior, environment, product grid) with manageable adaptations → Partial fit with explicit gap bullets → recommend `intent-to-3d-brief-writer`
    - Request violates Draft v0 scope (e.g., multi-user live editing) → Poor fit → document mitigations and alternative entrypoint

Playbooks

Playbook A: Direct product hero match (Strong fit)

• Match single-product focal stills or web GLB spin deliverables against packs/product-hero-v0/PACK.md and docs/workflows/product-hero-workflow.md (references/recipe-catalog.md).
• Assign Fit verdict: Strong with Gaps: none (or minor notes) using references/workflow-recommendation-template.md.
• Recommend ../intent-to-3d-brief-writer/SKILL.md as the single next skill and keep Runtime status: Not Run and Artifact status: Not Produced.

Playbook B:Adjacent domain or hybrid deliverable (Partial fit)

• When the user wants an architectural interior, character portrait, environment establishing shot, or multi-item product grid, score product-hero-v0 as Partial (or anchor on docs/workflows/full-non-blender-user-workflow.md + the matching domain planner).
• List each scope gap and mitigation explicitly per references/scope-gap-rules.md (for example, switching the downstream scene planner to ../interior-scene/SKILL.md or prioritizing the hero still before optional GLB export).
• Recommend ../intent-to-3d-brief-writer/SKILL.md to lock the adapted 8-slot brief.

Playbook C: Bundled multi-project ask (Poor fit / decomposition needed)

• When a user combines incompatible recipes in one request, assign Fit verdict: Poor (or Partial with blocking scope split) and explain why forcing a single pack causes rework.
• Recommend ../blender-recipe-decomposer/SKILL.md as the single immediate next skill to split the deliverables into ordered sub-projects.

Mode handling

Text-only mode

Fit from stated intent only; flag unknowns as gaps.

Runtime-ready mode

Same as text-only; optionally note if user demands CLI-only automation.

Blocked runtime mode

Still usable; note runtime unavailable does not change recipe fit.

Validation checklist

• Documented recipe/pack catalog consulted
• Fit verdict (Strong / Partial / Poor) and plain-language rationale assigned
• Explicit gap list included (or none)
• Exactly one next skill recommended from the valid 48-skill catalog
• Truth states follow downgrade-only rules with Artifact status defaulting to Not Produced
• No Blender execution, Python code, pack installation, or runtime success claims

Pass / Warn / Fail rubric

Failure handling

• If the user's goal statement is too ambiguous to score against references/recipe-catalog.md, do not guess a pack; keep Runtime status: Not Run and Artifact status: Not Produced and route to ../blender-brainstorming/SKILL.md or ../blendops-help/SKILL.md.
• If the prompt bundles multiple incompatible recipes and the user refuses to prioritize, assign Fit verdict: Poor, record the unmitigated scope collision under Limitations, and route to ../blender-recipe-decomposer/SKILL.md.
• Never claim a recipe or pack is "installed", "executed", or Verified during fit assessment.

Troubleshooting

Best practices

• Consult references/recipe-catalog.md and references/scope-gap-rules.md on every run so fit verdicts remain grounded in documented repository paths (packs/product-hero-v0/PACK.md, docs/recipes/cyberpunk-shoe-hero.md, docs/workflows/product-hero-workflow.md, docs/workflows/full-non-blender-user-workflow.md).
• Always emit a clear 3-tier verdict (Strong, Partial, or Poor) plus an explicit gap list (none only when genuinely Strong).
• Recommend exactly one next catalog skill so the agent does not fan out into parallel planners prematurely.
• Keep Runtime status: Not Run and Artifact status: Not Produced as the default truth states.

Good examples

• "Fit verdict: Strong — Single sneaker hero still + web GLB spin aligns directly with packs/product-hero-v0/PACK.md and docs/recipes/cyberpunk-shoe-hero.md. Gaps: none. Recommended next skill: ../intent-to-3d-brief-writer/SKILL.md. Runtime status: Not Run. Artifact status: Not Produced."
• "Fit verdict: Partial — Cinematic dining room furniture showcase overlaps docs/workflows/full-non-blender-user-workflow.md and ../interior-scene/SKILL.md, whereas product-hero-v0 assumes studio isolation. Gaps: (1) use interior architectural camera/lighting planner instead of studio sweep; (2) mark optional GLB spin as secondary deliverable. Recommended next skill: ../intent-to-3d-brief-writer/SKILL.md."

Bad examples

• "Fit verdict: Strong — Use the character-rigging-pro-v2 pack and start rendering immediately." — invents a non-existent pack and skips brief/preflight gates.
• "Use product-hero-v0 for your 50-character crowd battle animation; no gaps." — ignores references/recipe-catalog.md weak-fit rules.
• "Recommended next skills: intent-to-3d-brief-writer, pack-prerequisite-checker, product-hero-scene-planner, and lighting-material-plan. Artifact status: Produced." — violates the single-next-skill contract and fabricates artifact status.

User-facing response template

Goal summary: <one-line user goal>
Deliverables & constraints: <deliverable types + time/platform/realism constraints>
Fit verdict: <Strong | Partial | Poor>
Primary catalog anchor: <documented pack or recipe path>
Alternate anchors: <specialized planner or workflow path, or none>
Gaps & mitigations:
  - <explicit gap + mitigation, or "none">
Recommended next skill: <single catalog skill path>
Runtime status: Not Run
Artifact status: Not Produced
Evidence used: <catalog paths consulted>
Limitations: <unconfirmed user constraints or catalog coverage boundaries>

Anti-patterns

• Inventing undocumented packs or recipes that do not exist under packs/, docs/recipes/, or docs/workflows/.
• Assigning Strong fit when the user requests multiple conflicting deliverables without prioritization.
• Recommending more than one immediate next skill in the output block.
• Writing Artifact status: Not Run or upgrading Runtime status / Artifact status during recipe scoring.
• Writing the full 8-slot 3D brief or executing preflight checks inside recipe-fit-assessor instead of handing off to intent-to-3d-brief-writer or pack-prerequisite-checker.

Cross-skill handoff

• [upstream_reference] Upstream creative divergence → ../blender-brainstorming/SKILL.md
• [boundary_delegation] General entry routing → ../blendops-help/SKILL.md
• [conditional] Multi-deliverable scope split when prompt spans multiple incompatible recipes → ../blender-recipe-decomposer/SKILL.md
• [progression] Downstream structured brief → ../intent-to-3d-brief-writer/SKILL.md
• [progression] Pack prerequisite verification → ../pack-prerequisite-checker/SKILL.md
• [conditional] Post-artifact stage routing → ../workflow-stage-router/SKILL.md
• [alternative] Domain scene planner (product hero) → ../product-hero-scene-planner/SKILL.md
• [alternative] Domain scene planner (product grid) → ../product-grid-scene-planner/SKILL.md
• [alternative] Domain scene planner (character portrait) → ../character-portrait/SKILL.md
• [alternative] Domain scene planner (environment establishing shot) → ../environment-shot/SKILL.md
• [alternative] Domain scene planner (interior architectural) → ../interior-scene/SKILL.md

Non-goals

• Brainstorming open-ended creative concepts before a goal exists (use ../blender-brainstorming/SKILL.md).
• Authoring the locked 8-slot 3D brief (use ../intent-to-3d-brief-writer/SKILL.md).
• Running the 7-row pack preflight gate (use ../pack-prerequisite-checker/SKILL.md).
• Selecting MCP runtime paths, configuring Blender, or verifying rendered/exported artifacts.

References

• references/recipe-catalog.md
• references/scope-gap-rules.md
• references/workflow-recommendation-template.md
• ../../docs/recipes/README.md