runtime-path-picker
Purpose
Produce a single primary runtime execution choice among Path 1 (Blender Lab MCP), Path 2 (community ahujasid/blender-mcp), or the CLI fallback appendix — with explicit tradeoffs and no false "standalone Connector" claims.
Quick start
- collect Blender version band, host app (Claude Desktop vs coding agent vs CI), interactivity needs
- score against
references/path-constraints-table.md - walk
references/decision-tree.md - document choice using
references/anti-patterns.mdguardrails
When to use
- user unsure Lab vs community vs CLI
- environment pinned to Blender below 5.1 (Path 1 unavailable)
- automation / batch render discussion
When not to use
- port collision diagnostics →
runtime-bridge-conflict-resolver - detailed install steps →
official-runtime-setup-guide - readiness probing →
official-runtime-readiness-checker
Trigger phrases
- "which MCP should I use"
- "Lab MCP or ahujasid"
- "Path 1 or Path 2"
- "CLI vs MCP"
Prerequisites / readiness
- target Blender major/minor known or explicitly unknown
- whether interactive steering vs batch matters
Input schema
Required inputs
| Input | Why it is required |
|---|---|
| Blender version (or uncertainty) | Gates Path 1 |
| Host environment category | Connector vs generic MCP vs CI |
Optional inputs
| Input | Use |
|---|---|
| Org IT constraints | May force CLI appendix |
| Prior tooling investment | Path loyalty |
Output schema
Primary output
- primary path recommendation + rationale
- rejected paths + why
- host option note for Path 1: (a) Anthropic Blender Connector or (b) manual MCP client — both require Lab add-on inside Blender
Evidence / caveat output
Runtime status: Not Run | Attempted | Produced | Verified | Failed | Blocked / Not Run
Artifact status: Not Run | Not Produced | Produced | Verified | Failed
Evidence used: <user-stated constraints only unless logs provided>
Limitations: <unknown Blender build>
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
Decision skill only — does not configure hosts. Must reflect BlendOps model: two MCP execution paths plus CLI fallback appendix documented upstream as a first-class Blender CLI surface; No in-repo evidence file yet for BlendOps-run CLI examples. Path 2 is third-party community (ahujasid/blender-mcp) — mature prior art with manual configuration responsibility; see docs/unofficial-runtime-bridges.md. Never describe Anthropic Connector as replacing the Lab add-on inside Blender.
Operating procedure
- Gate Path 1 on Blender 5.1+ feasibility.
- If Path 1 impossible and interactive MCP wanted → Path 2 if acceptable risk profile.
- If batch / headless / CI → prefer CLI appendix plan (
cli-appendix-job-outline). - Warn on Single-bridge constraint if multiple bridges detected.
Mode handling
Text-only mode
Hypothetical picks labeled clearly.
Runtime-ready mode
Cross-check suggestions with readiness checker output if present.
Blocked runtime mode
Still deliver decision table for future attempt.
Validation checklist
- Primary path singular
- Connector not standalone assertion present when Path 1 + Connector mentioned
- Path 2 labeled community third-party when recommended
- CLI appendix framed per policy strings above when mentioned
Pass / Warn / Fail rubric
| Verdict | Criteria |
|---|---|
| Pass | Coherent pick + constraints mapped |
| Warn | Missing Blender version — conditional branches explicit |
| Fail | Claims Connector alone suffices for Path 1 |
Cross-skill handoff
- Setup detail →
../official-runtime-setup-guide/SKILL.md - Conflicts →
../runtime-bridge-conflict-resolver/SKILL.md - CLI shape →
../cli-appendix-job-outline/SKILL.md - Host specifics Path 1 →
../path-one-host-outline/SKILL.md
References
references/path-constraints-table.mdreferences/decision-tree.mdreferences/anti-patterns.md../../docs/runtime-stack-strategy.md