---
name: fidget-spinner
description: Use Fidget Spinner as the local system of record for frontier grounding, hypothesis tracking, experiment adjudication, and artifact references. Read health first, ground through frontier.open, and walk the graph deliberately one selector at a time.
---

# Fidget Spinner

Use this skill when working inside a project initialized with Fidget Spinner or
when the task is to inspect or mutate a frontier through the packaged MCP.

Start every session by reading `system.health`.

If the session is unbound, or bound to the wrong repo, call `project.bind`
with the target project root or any nested path inside that project.

If the target root exists and is empty, `project.bind` will bootstrap the local
store automatically.

Then read:

- `project.status`
- `tag.list`
- `frontier.list`
- `frontier.open` for the active frontier

`frontier.open` is the only sanctioned overview surface. It is allowed to give
you the frontier brief, active tags, live metrics, active hypotheses, and open
experiments in one call.

If you need more context, pull it from:

- `hypothesis.read`
- `experiment.read`
- `artifact.read`

## Posture

- `frontier` is scope and grounding, not a graph vertex
- `hypothesis` and `experiment` are the true graph nodes
- every experiment has one mandatory owning hypothesis
- experiments and hypotheses may also cite other hypotheses or experiments as influence parents
- the frontier brief is the one sanctioned freeform overview
- artifacts are references only; Spinner never reads artifact bodies
- token austerity matters more than convenience dumps

## Choose The Cheapest Tool

- `tag.add` when a new campaign or subsystem token is genuinely needed; every tag must carry a description
- `tag.list` before inventing tags by memory
- `frontier.brief.update` when the situation, roadmap, or unknowns need to change
- `hypothesis.record` before core-path work; every experiment must hang off exactly one hypothesis
- `hypothesis.update` when the title, summary, body, tags, or influence parents need tightening
- `experiment.open` once a hypothesis has a concrete slice and is ready to be tested
- `experiment.list` or `experiment.read` when resuming a session and you need to recover open or recently closed state
- `experiment.update` while the experiment is still live and its summary, tags, or influence parents need refinement
- `experiment.close` only for an already-open experiment and only when you have measured result, verdict, and rationale; attach `analysis` only when the result needs interpretation beyond the rationale
- `artifact.record` when preserving an external file, link, log, table, plot, dump, or bibliography by reference
- `artifact.read` only to inspect metadata and attachments, never to read the body
- `metric.define` when a project-level metric key needs a canonical unit, objective, visibility tier, or description
- `metric.keys --scope live` before guessing which numeric signals matter now
- `metric.best` when you need the best closed experiments by one numeric key; pass exact run-dimension filters when comparing one slice
- `run.dimension.define` when a new experiment slicer such as `instance` or `duration_s` becomes query-worthy
- `run.dimension.list` before guessing which run dimensions actually exist in the store

## Workflow

1. Ground through `frontier.open`.
2. State the intended intervention with `hypothesis.record`.
3. Open a live experiment with `experiment.open`.
4. Do the work.
5. Close the experiment with `experiment.close`, including dimensions, metrics, verdict, rationale, and optional analysis.
6. Attach any large markdown, logs, tables, dumps, or links through `artifact.record` instead of bloating the ledger.

Do not dump a whole markdown tranche into Spinner. If it matters, attach it as
an artifact and summarize the scientific upshot in the frontier brief,
hypothesis, or experiment outcome.

## Discipline

1. `frontier.open` is the only overview dump. After that, walk the graph one selector at a time.
2. Pull context from hypotheses and experiments, not from sprawling prompt prose.
3. Do not expect artifact content to be available through Spinner. Open the file or link out of band when necessary.
4. If the MCP behaves oddly or resumes after interruption, inspect `system.health`
   and `system.telemetry` before pushing further.
5. Keep fetches narrow by default; slow is better than burning tokens.
6. Treat metric keys as project-level registry entries and run dimensions as the
   first-class slice surface for experiment comparison; do not encode scenario
   context into the metric key itself.
7. A hypothesis is not an experiment. Open the experiment explicitly; do not
   smuggle planned work into the frontier brief.
8. Experiments are the scientific record. If a fact matters later, it should
   usually live in a closed experiment outcome rather than in freeform text.
9. The ledger is scientific, not git-forensic. Do not treat commit hashes as experiment identity.
10. Porcelain is the terse triage surface. Use `detail=full` only when concise
   output stops being decision-sufficient.
11. When the task becomes a true indefinite optimization push, pair this skill
    with `frontier-loop`.