| name | blender-agent-benchmark |
| description | Benchmark Blender modeling agents, skills, prompts, scripts, or MCP tools with paired isolated runs. Use for baseline-versus-plugin comparisons, regression suites, skill forward-testing, MCP usefulness evaluation, score calibration, or claims that a Blender workflow improves mesh, visual, structural, animation, cost, or completion quality. |
Blender Agent Benchmark
Measure changes with the same tasks, model, effort, limits, Blender build, and evaluator. Preserve natural agent behavior.
Protect benchmark integrity
- Create isolated directories for every condition and repetition.
- Do not leave the other condition's code, renders, metrics, or expected fixes where the agent can discover them.
- Keep the user-facing task prompt identical except for explicit skill invocation in the plugin condition.
- Use
codex exec --ignore-user-config for the no-plugin baseline.
- Use the installed plugin in a fresh invocation for the plugin condition.
- Record CLI version, model, effort, Blender build, duration, tool calls, failures, and output hashes.
- Evaluate outputs after generation. Do not leak hidden rubric details to the agent.
Read references/methodology.md before changing fixtures, scoring, or comparison claims.
Read references/validated-results.md only
when reviewing the plugin's recorded validation result, not while generating a
benchmark submission.
Run the suites
Use scripts/run_benchmark.ts:
bun "<skill-root>\scripts\run_benchmark.ts" `
--suite quick `
--mode baseline `
--output "<run-root>\baseline"
bun "<skill-root>\scripts\run_benchmark.ts" `
--suite quick `
--mode plugin `
--output "<run-root>\plugin"
Start with a smoke task to validate the harness. Use at least three representative tasks and repeated runs before claiming a general capability gain.
Use scripts/compare_runs.ts for randomized blinded multiview judging. Use
scripts/rescore_run.ts to recompute deterministic scores after a scorer
change without rerunning agents.
Noninteractive Codex cancels MCP tool calls that require approval. If testing
an MCP condition, pass --bypass-approvals to every compared condition and
use isolated benchmark directories. Do not give only the MCP condition broader
permissions.
Use scripts/benchmark_mcp.ts --asset <path> --output <new-dir> to verify that
the MCP transport returns the same deterministic metrics as direct CLI
evaluation. Equivalent results prove transport correctness, not a modeling
quality gain.
Compare conditions
Keep these dimensions separate:
- execution/export validity;
- prompt and structural compliance;
- geometry/game-readiness;
- multiview visual quality;
- physical plausibility;
- animation quality when applicable;
- context/export correctness;
- time, turns, tool failures, and cost.
Use hard gates before the weighted score. Prefer blinded pairwise visual review over uncalibrated absolute aesthetic scores.
Iterate
- Inspect failed metrics, renders, videos, and traces.
- Identify one transferable workflow defect.
- Change the smallest relevant skill, deterministic tool, or MCP surface.
- Rerun the same slice.
- Run a holdout task before retaining the change.
Do not retain benchmark-specific instructions that reveal fixture answers or damage ordinary modeling behavior.