Skip to content

openJiuwen-ai/sciencediscovery / skills/idea-tree-team

idea-tree-team

Verified

Prepare supplied inputs for autonomous Idea Tree research and explain its saved results.

FileKindSize
SKILL.mdskill md2.6 KB
references/evaluation-principles.mdmarkdown1.7 KB
references/workflow.mdmarkdown1.1 KB

references/evaluation-principles.md

View on GitHub
# Optional Evaluation Principles

Use these principles only when the user prompt selects a multi-perspective evaluation workflow. They are not Runtime requirements.

## General Rules

1. Keep every reported score finite and inside the configured range.
2. State which rubric, evidence, or judgment produced the final leaf score.
3. Preserve disagreements between evaluators instead of silently averaging them away.
4. Do not fabricate sources, measurements, Specialist outputs, or Artifact identities.
5. When durable provenance is useful, declare exact outputs as Artifacts and checkpoint them; otherwise these steps may be skipped.
6. `idea_tree_finalize` accepts the claimed execution identity, final score, semantic insight, and optional exact Artifact identities. It does not require a particular evaluator count or schema.

## Optional Catalyst Triad

For prompts explicitly requesting the traditional catalyst review, three independent perspectives can cover:

- activity: catalytic activity, mechanism, selectivity, efficiency;
- stability: structural stability, durability, recyclability, lifetime;
- sustainability: environmental safety, lifecycle, disposal, resource use.

The workflow or prompt must define how these perspectives are combined. The Runtime does not assume the former `0.35 / 0.35 / 0.30` weighting.

## Artifact Guidance

If Artifacts are supplied to `idea_tree_finalize`:

- use exact `artifact_id`, `version_id`, and a meaningful workflow-defined `role`;
- submit at most one version of each Artifact;
- ensure each version comes from a completed Specialist execution in the same Session and executor contract;
- treat checkpoints as optional retry aids, not mandatory completion gates.
add-product-e2e-evalpaperclipai/paperclipAdd or extend a Paperclip full-stack runner E2E workflow, fixture, matcher, or report evidence path for local or Daytona execution.Stars: 115kUpdated Claude CodeCodexVerifiedInstall
add-runner-evalpaperclipai/paperclipAdd or extend a Paperclip Runner protocol evaluation definition, roster, assertion, or report fixture with provenance and narrow validation.Stars: 115kUpdated Claude CodeCodexVerifiedInstall
diagnose-why-work-stoppedpaperclipai/paperclipCompatibility entry point for failed, stalled, looping, or over-recovered Paperclip work. Use fix-failed-paperclip-run for the full diagnosis, regression, repair, real verification, and green-PR workflow; honor an explicitly requested diagnosis-only scope.Stars: 115kUpdated Claude CodeCodexVerifiedInstall
fix-failed-paperclip-runpaperclipai/paperclipDiagnose and fix failed, stuck, slow, or incorrectly blocked Paperclip tasks and runs. Use with a task/run URL or failure report to trace the root cause, prove a regression red then green, check second-order effects, verify the real workflow, and deliver a PR with passing checks.Stars: 115kUpdated Claude CodeCodexVerifiedInstall