本文へ移動
cccskills
無料GitHub で公開

gaia-fuse-full-suite

Fuse all of a contributor's named skills into a single fusion node (suite capstone) in the registry. Use when the user says "fuse <contributor>'s skills", "create a suite for <contributor>", "build a capstone for <user>", "fuse all of <user>'s named skills", "create a suite fusion", "consolidate <contributor> into one skill", or explicitly types /gaia-fuse-full-suite. Also triggers when a contributor has accumulated 3+ named skills and the user asks what to do next with them, or when someone asks "can we make a suite from these?". This skill: collects component IDs from registry/named/<contributor>/, verifies nodes exist, researches evidence and appraises Trust Magnitude, runs `gaia dev fuse` to write the fusion node and suite manifest, validates, and opens a PR on a review/meta/ branch.

インストール方法を見る

含まれるファイル(1)

  • SKILL.md8.5 KB

SKILL.md(原文)

インストールする前に、エージェントに与えられる指示の中身を確認できます。

gaia-fuse-full-suite

Fuse every named skill attributed to a contributor into a new fusion node, link it as the suite capstone, record a fuse timeline event, and open a PR.

A fusion node is the right abstraction when a contributor has built a coherent body of named skills that together form a recognisable suite. The fusion step makes that relationship explicit in the graph: prerequisites point to the fusion node, and the fusion node's evidence anchors the whole cluster's credibility.

Taxonomy (Yggdrasil II)

Ratified 2026-07-07. Read CONTEXT.md § Taxonomy v6 before writing any copy.

  • The type axis is starless-only and collapsed to exactly {basic, fusion}. extra, ultimate, and unique are retired as types — never write them to a node.
  • Ultimate (5★ Suite) and Apex (6★ Suite) survive only as rank names, never as taxonomy words. A fused capstone is a Fusion, not "an Ultimate".
  • Branch (standard / suite / unique) is derived at read time, never declared. A Named Skill carrying suiteComponents is suite at any rank.
  • The per-star Evidence Floor is retired. Trust Magnitude is the sole promotion gate.

Programmatic-First — use the CLI

Per CLAUDE.md, all registry mutations go through gaia dev verbs. Do not hand-write node JSON — gaia dev fuse writes the fusion node, the suite manifest, the back-links, and the timeline event, and it always emits type: fusion into registry/nodes/fusion/. Hand-edits skip timeline logging and are how schema-invalid nodes reach the registry.

Inputs

Gather these before starting. Infer from context when obvious; ask when ambiguous.

InputDescriptionExample
contributorGitHub username of the named contributorobra
fusion-idkebab-case ID for the new fusion nodesuperpowers
fusion-nameHuman-readable display nameSuperpowers
capstone-idNamed capstone to link (contributor/slug)obra/superpowers
source-urlCanonical repo or landing page (for evidence research)https://github.com/obra/superpowers

If the contributor already has a fusion node with this ID and it has a fuse timeline entry, abort early and report the existing entry — there is nothing to do.

Workflow

1. Collect components

ls registry/named/<contributor>/

For each .md file, parse the YAML frontmatter and extract:

  • genericSkillRef — the component skill ID
  • level — used in the summary table
  • title — the named title

Fail loudly (do not silently skip) if any file is missing genericSkillRef, or has status: awakened. Awakened skills have not been named yet and cannot participate in a fusion — their inclusion would undercount what the suite represents.

2. Verify component nodes exist

For each genericSkillRef, confirm a node JSON file exists under registry/nodes/basic/ or registry/nodes/fusion/. Report any missing IDs and abort — missing nodes must be registered via gaia dev add before the fusion can proceed. Do not create stub nodes inline.

3. Check for an existing fusion node

gaia dev list --generic | grep '<fusion-id>'
  • Already exists + has fuse timeline entry → abort (report URL).
  • Already exists, no fuse entry → update mode; gaia dev fuse merges rather than clobbers.
  • Does not exist → proceed to creation.

4. Research evidence and appraise Trust Magnitude

Fetch source-url with WebFetch and record the signals that will feed TM scoring:

  • GitHub star count and trajectory
  • Multi-platform adoption (unrelated hosts shipping it)
  • Active release tag / version ≥ 1.0
  • Published academic paper, peer review, or official specification

Log each source through the CLI so it is graded and timestamped:

gaia dev evidence <fusion-id> "<url>" --type <evidence-type> --evaluator <your-handle>

Then appraise:

gaia appraise <fusion-id>

Rank decision — Trust Magnitude is the sole gate. There is no per-star Evidence Floor under Yggdrasil II; do not assert a star level from an evidence grade alone. For the Suite branch the published gates are:

  • 4★ Extra — Origin Contributor recorded + TM ≥ 100
  • 5★ Ultimate — Origin in suiteComponents + 5 A-graded origins in suiteComponents + TM ≥ 250
  • 6★ Apex — the full 6-predicate Apex Gate (see META.md)

If the appraised TM does not clear a gate, record the rank the TM supports and say so in the PR. Do not round up.

5. Run the fusion

gaia dev fuse <fusion-id> \
  --name "<fusion-name>" \
  --description "<one sentence synthesising what the suite achieves>" \
  --prereqs <component-id-1>,<component-id-2>,... \
  --named-capstone <capstone-id> \
  --suite-components <named-id-1>,<named-id-2>,...

This writes registry/nodes/fusion/<fusion-id>.json with type: fusion, upserts the suite manifest at registry/suites/<contributor>/<slug>.json, back-links derivatives on every component node, appends the fuse timeline event, and rebuilds the graph and docs.

Notes:

  • --type is not a flag. gaia dev fuse always produces a fusion node; the type is derived structurally from the presence of prerequisites.
  • suiteComponents is a Named-Skill-only field. Never write it onto the starless node.
  • Do not write a branch field — branch is computed at read time.
  • Pass --no-build only if you are batching several fusions; run one build at the end.

6. Validate

PYTHONIOENCODING=utf-8 gaia dev validate

All checks must pass. Common failures and fixes:

  • Reference integrity — a missing derivative or prerequisite; re-run the fuse with the complete --prereqs list.
  • DAG cycle — a component points back to itself through the fusion node; remove the self-referencing derivative.
  • Fusion constraints — the recorded rank exceeds what Trust Magnitude supports; lower the rank via gaia dev calibrate, or log stronger evidence and re-appraise.
  • Meta sync — registry/ and the bundled snapshot diverged; re-run the build step.

Do not open a PR until validation is clean.

7. Commit and open PR

Branch: review/meta/<fusion-id>-fusion

Per CLAUDE.md Guard E, a change to registry/nodes/ or registry/named/ must ship the regenerated Class S artifacts (docs/graph/*) in the same PR.

git checkout -b review/meta/<fusion-id>-fusion origin/main
git add registry/nodes/fusion/<fusion-id>.json \
        registry/nodes/basic/<component-id>.json \
        registry/suites/<contributor>/ \
        registry/named-skills.json \
        docs/graph/
git commit -m "feat(registry): fuse <contributor>/<suite-name> into /<fusion-id> <rank> fusion

- Add /<fusion-id> fusion node capstoning N named <contributor> skills
- Back-link derivatives on all N component nodes
- Add fuse timeline event (action: fuse, <today>)
- Regenerate Class S artifacts (docs/graph/*)

gaia dev validate: all checks pass (<total> skills, <edge> edges)."
git push -u origin review/meta/<fusion-id>-fusion

PR body template:

## Summary

- New fusion node **/<fusion-id>** (<rank>) capstoning N named <contributor> skills
- Fuse timeline event recorded; components back-linked via `derivatives`
- Evidence: <one-line summary of source and signals>
- Trust Magnitude: <TM> (gate for <rank> is <threshold>)

## Component skills

| ID | Named Title | Rank |
|---|---|---|
| <id> | <title> | <rank> |

## Rank rationale

<Which TM gate the appraisal cleared — cite the appraised TM from step 4>

## Validation

- `gaia dev validate` — all checks pass

Output

Report back concisely:

  • PR URL
  • Fusion node ID and rank
  • Number of components fused
  • Appraised Trust Magnitude and the key signals behind it
  • Validation result

レビュー

まだレビューはありません。使ってみた感想をお寄せください。

同じリポジトリのスキル

概要と使いどころ

ci-churn

無料

Measure CI churn cost for a pull request: how many commits and how many CI-compute seconds were burned on avoidable "fix the CI" push rounds. Use when you want to understand the efficiency cost of a PR's iteration cycle, get actionable pre-push check suggestions, or include a churn summary in a pipeline report. Triggers: "ci churn", "how much CI did we burn", "avoidable commits", "ci efficiency", "wasted CI time", "churn report", "/ci-churn". Called automatically by /fp-drift at pipeline close.

日本語の概要は準備中です。原文の説明を表示しています。

gaia-research/gaia-skill-tree232026年10月13日 更新

Produce the rendered evidence a human-gated frontend PR owes its reviewer: capture each changed surface (plus the before state for anything modified, plus mobile width), then generate one self-contained local HTML page the founder can open and review without checking anything out. Use when a PR is gated by the "Frontend changes are HUMAN-GATED" section of CLAUDE.md, or when the user says "/design-gate", "screenshot evidence for this PR", "prep this for founder review", "build the design evidence page", "what does this look like before and after". The page comes from a template script, not hand-written markup. Output is session-local and never committed. Do NOT use for non-gated changes: typo/grammar/factual fixes in an unchanged block, dead link targets, alt/ARIA fixes with no visual change, cache-bust bumps, or refactors with no rendered diff.

日本語の概要は準備中です。原文の説明を表示しています。

gaia-research/gaia-skill-tree232026年10月13日 更新

One-fix-at-a-time, approval-gated design fix recovery. Given a checklist of fixes (handover doc, CHECKLIST.md, or inline list), applies each individually, presents a diff, and waits for an explicit APPROVE / REJECT / NO CHANGES decision before committing. Trigger phrases: /design-iteration, "go through the checklist one by one", "fix these one at a time", "bucket recovery", "design fix pass".

日本語の概要は準備中です。原文の説明を表示しています。

gaia-research/gaia-skill-tree232026年10月13日 更新

Playbook for correcting a named skill's star level to match its computed Trust Magnitude grade, with an independent post-calibration verification pass and a CLI-logged timeline event (never hand-written). Use when: recalibrating a skill's level, fixing a stale or drifted star rating, running a batch of `gaia dev calibrate` corrections, or confirming a past calibration actually landed. Triggers: "calibrate <skill>", "recalibrate the trust magnitude backlog", "promote/demote <skill> to N stars", "why is <skill> still N stars", "verify a calibration actually landed", "Hall of Heroes ranking is wrong".

日本語の概要は準備中です。原文の説明を表示しています。

gaia-research/gaia-skill-tree232026年10月13日 更新

Close out a completed EPIC / sprint end-to-end: author a fact-verified meta-post announcing what shipped, archive the EPIC's handovers, snapshot orchestrator memory, gate the integration→main merge on CI-green + explicit human go, drive the version bump (merge-commit conventions that auto-sync classifies), and verify the release reached the registry/PyPI. Use when the user says "close out EPIC <n>", "wrap up the sprint", "ship the meta shift", "/epic-close", or asks to finalize an integration branch that is ready to merge to main. Generalizes the EPIC 1002 (Yggdrasil II) close-out. This skill orchestrates — it authors founder docs directly (superadmin) and delegates code/regen to workers; it NEVER merges to main without the human gate.

日本語の概要は準備中です。原文の説明を表示しています。

gaia-research/gaia-skill-tree232026年10月13日 更新

Run this skill for Phase 3 of the evidence verification pipeline — the adversarial audit step. Use when you need to check the evidence data lake for bad data before ingestion: dead links, wrong URL formats (tree/ vs blob/), subjective wording ("elite", "high-quality"), stale migration notes, benchmark catalog misuse/vendor-claim leakage, or skills whose star evidence conflicts with classified evidence level. Triggers on phrases like: "audit the data lake", "adversarial check", "ev-adversarial-audit", "check for noise in evidence", "flag bad evidence", "run the audit phase", "quality check the by-type files", or any reference to Phase 3 of the pipeline.

日本語の概要は準備中です。原文の説明を表示しています。

gaia-research/gaia-skill-tree232026年10月13日 更新

gaia-research のスキルをすべて見る

このスキルの問題を報告する