Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when diagnosing Superloopy doctor, install, wrapper, plugin cache, hook bootstrap, bundled agents, marketplace, Codex, Claude Code, stale-version, evidence-floor, or host-wiring health problems.
.claude/skills/hashgraph-online-superloopy-doctor/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-13 | ✗→✓ | ▲ Improved | 21% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 139% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 76% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 31% | 0% |
| case-17 | ✗→✓ | ▲ Improved | 60% | 0% |
SUPERLOOPY DOCTOR ENABLED
Use this skill to prove whether a local Superloopy surface can support the Superloopy loop: repo-local .superloopy/ state, artifact-backed evidence under .superloopy/evidence/, bundled crew agents, hook steering, and the deterministic completion floor. It is read-only by default. Do not mutate plugin config, wrappers, agents, caches, or source files unless the user explicitly asks for repair after seeing the report.
This is a Superloopy health check, not a generic plugin drift audit.
superloopy doctor --scope source --json as the source-checkout truth surface and superloopy doctor --scope installed --json as the machine-local installation truth surface. The JSON scope field records which verdict contract was used.superloopy loop status --json, superloopy loop check, superloopy loop guide --json, and evidence files under .superloopy/evidence/ can show current plan and artifact readiness.superloopy doctor reports the plugin root it checked: by default it uses source scope in a recognized Superloopy checkout and installed scope elsewhere. Pin a checkout with superloopy doctor --root <checkout> --scope source --json; pin an installed cache with superloopy doctor --root <cache> --scope installed --json. Source verdicts still return machine-local diagnostics, but installedModelPolicy and wrapper do not gate source health.installedPluginTruth runs the read-only codex plugin list --json authority probe for superloopy@beefiker; it never infers the installed version from cache directory names. A confirmed version mismatch is informational in source scope, but fails installed scope. Missing Codex, no registered plugin, and invalid authority output are informational and do not fail doctor. Report these four states as current, version_mismatch, not_registered, or authority_unavailable; never repair from this probe without explicit approval.
pwd, git rev-parse --show-toplevel when it is a checkout, and the installed plugin cache path when the wrapper points into one.command -v superloopy, which -a superloopy, and the wrapper text or symlink target.codex plugin list --json when Codex is available; use it only to locate superloopy@beefiker and its version.scope and root. For checkout health use superloopy doctor --root <checkout> --scope source --json; for a machine-local wrapper, managed-agent fleet, or installed cache use superloopy doctor --root <cache> --scope installed --json.skills, fileAudit, reviewability, dispatchCoherence, hostContract, modelPolicy, and claudeHostWiring.superloopy loop status --json if a plan exists, otherwise note "no active plan" instead of creating one. Use superloopy loop check only to validate trace/artifact readiness for an existing plan; read-only diagnosis cannot prove completion safety because it does not re-run recorded commands. If the user asks whether completion is actually safe, point at the real gates: superloopy loop review, superloopy loop checkpoint, or superloopy loop finish.franky, zoro, usopp, jinbe, robin, and nami; judge by files and doctor checks, not by role labels alone.--force; never infer ownership from names alone.model_unverified unless the host attests both the resolved role and model.PASS: wrapper/cache/checkout, superloopy doctor --json, and relevant loop/evidence surfaces agree within read-only limits.WARN: Superloopy can run, but a wrapper points at an older cache, no active plan exists for loop readiness, read-only mode cannot prove completion safety, host behavior is advisory only, or manual evidence cannot be command-replayed.FAIL: doctor named checks fail, wrapper is absent or points at the wrong root, required agents are missing, .superloopy/ state is corrupt, or the deterministic completion floor cannot be evaluated.Recommend repair commands only after the read-only report explains the failure:
codex plugin marketplace upgrade beefiker.installedPluginTruth mismatch after approval: codex plugin add superloopy@beefiker --json, then start a new Codex session.node src/cli.js install --json.node src/cli.js install --force --json.Never run install --force, plugin remove, marketplace remove, cache deletion, or wrapper deletion without explicit user approval.
textSuperloopy doctor report Summary: <healthy/degraded/broken in one sentence> | Check | Verdict | Evidence | | --- | --- | --- | | Command wrapper | PASS/WARN/FAIL | <path, target, cache or checkout root> | | Doctor checks | PASS/WARN/FAIL | <ok=true or failing check names> | | Evidence floor | PASS/WARN/FAIL | <.superloopy/evidence/, loop check/status readiness, or "completion safety requires the real gate"> | | Crew install | PASS/WARN/FAIL | <agent files and dispatchCoherence evidence> | | Host limits | PASS/WARN/FAIL | <hostContract/claudeHostWiring evidence> | | Reviewability/native boundary | PASS/WARN/FAIL | <reviewability/fileAudit/designAudit evidence> | Recommended repair: - <exact command or "none">
superloopy doctor --json prove it.superloopy loop check alone; those are read-only readiness signals. Completion safety requires the real gate path (superloopy loop review, superloopy loop checkpoint, or superloopy loop finish) because those gates re-derive command-backed criteria.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-13 | fail→pass | 20,511 | 5,451 | -73% | 1 | 1 | 0% | 2,222 | 2,678 | +21% | 0 | 0 | — |
case-02 | fail→fail | 16,506 | 11,609 | -30% | 1 | 1 | 0% | 409 | 2,210 | +440% | 0 | 0 | — |
case-12 | fail→fail | 10,977 | 10,882 | -1% | 1 | 1 | 0% | 1,648 | 2,591 | +57% | 0 | 0 | — |
case-01 | fail→fail | 5,463 | 18,779 | +244% | 1 | 1 | 0% | 287 | 2,481 | +764% | 0 | 0 | — |
case-03 | fail→fail | 22,190 | 13,568 | -39% | 1 | 1 | 0% | 1,783 | 2,238 | +26% | 0 | 0 | — |
case-04 | pass→fail | 11,816 | 14,163 | +20% | 1 | 1 | 0% | 2,406 | 2,334 | -3% | 0 | 0 | — |
case-05 | fail→fail | 8,590 | 19,335 | +125% | 1 | 1 | 0% | 1,369 | 2,169 | +58% | 0 | 0 | — |
case-06 | pass→fail | 22,304 | 7,875 | -65% | 1 | 1 | 0% | 2,593 | 2,101 | -19% | 0 | 0 | — |
case-07 | fail→fail | 5,522 | 9,547 | +73% | 1 | 1 | 0% | 949 | 2,298 | +142% | 0 | 0 | — |
case-08 | fail→fail | 14,315 | 10,412 | -27% | 1 | 1 | 0% | 1,567 | 2,657 | +70% | 0 | 0 | — |
case-09 | fail→pass | 7,231 | 5,302 | -27% | 1 | 1 | 0% | 1,083 | 2,593 | +139% | 0 | 0 | — |
case-10 | fail→pass | 13,847 | 4,925 | -64% | 1 | 1 | 0% | 1,356 | 2,382 | +76% | 0 | 0 | — |
case-11 | fail→pass | 16,477 | 10,087 | -39% | 1 | 1 | 0% | 1,888 | 2,466 | +31% | 0 | 0 | — |
case-14 | fail→fail | 30,033 | 3,665 | -88% | 1 | 1 | 0% | 5,056 | 2,150 | -57% | 0 | 0 | — |
case-15 | fail→fail | 10,835 | 8,863 | -18% | 1 | 1 | 0% | 1,610 | 2,304 | +43% | 0 | 0 | — |
case-16 | pass→pass | 12,593 | 12,010 | -5% | 1 | 1 | 0% | 1,900 | 2,939 | +55% | 0 | 0 | — |
case-17 | fail→pass | 8,280 | 3,433 | -59% | 1 | 1 | 0% | 1,291 | 2,071 | +60% | 0 | 0 | — |
case-18 | pass→pass | 15,023 | 8,010 | -47% | 1 | 1 | 0% | 1,317 | 2,667 | +103% | 0 | 0 | — |
case-19 | fail→pass | 16,612 | 8,847 | -47% | 1 | 1 | 0% | 1,725 | 2,191 | +27% | 0 | 0 | — |
case-20 | fail→pass | 20,290 | 2,752 | -86% | 1 | 1 | 0% | 2,549 | 2,044 | -20% | 0 | 0 | — |
case-21 | fail→pass | 13,965 | 9,033 | -35% | 1 | 1 | 0% | 2,190 | 2,347 | +7% | 0 | 0 | — |
case-22 | fail→pass | 20,321 | 11,016 | -46% | 1 | 1 | 0% | 2,338 | 2,519 | +8% | 0 | 0 | — |
case-23 | fail→fail | 19,419 | 2,594 | -87% | 1 | 1 | 0% | 1,997 | 2,164 | +8% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 23 cases were attempted, and 17 counted toward the lift figure. The other 6 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +30 percentage points is the difference between those two pass rates over the 17 comparable cases. 4 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.