Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when starting hyperflow in a new project, re-initializing analysis, refreshing `.hyperflow/` cache, or installing multi-tool auto-detection shims (AGENTS.md, Cursor rules, GEMINI.md, CLAUDE.md). Trigger phrases — "init hyperflow", "set up hyperflow", "refresh hyperflow", "install hyperflow shims". Standalone setup; does **not** auto-chain into the feature flow.
.claude/skills/ccplugins-scaffold/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | 6% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -3% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -5% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -6% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -24% | 0% |
One-shot project setup. Analyzes the codebase, builds the .hyperflow/ cache, seeds the memory skeleton, and optionally installs detection shims for other AI tools. Does not start the spec → scope → dispatch chain — invoke /hyperflow:spec (or /hyperflow:scope) when you're ready for that.
Check for .hyperflow/ at project root.
If absent — dispatch parallel searchers (single message, six Agent calls):
| Label | File generated | Discovers | |---|---|---| | Searcher — analyzing tech stack | profile.md | Name, language, framework, build commands | | Searcher — mapping folder structure | architecture.md | Dirs, patterns, routing, data flow | | Searcher — extracting conventions | conventions.md | Naming, style, linting rules | | Searcher — scanning dependencies | dependencies.md | UI lib, state, data fetching, DB, auth | | Searcher — auditing test setup | testing.md | Runner, E2E, patterns, commands | | Searcher — reading git workflow | git-workflow.md | Branches, commits, CI/CD, PR conventions |
See project-analysis.md for what each file captures.
If present — staleness check: Compute SHA256 of tracked config files, compare against .hyperflow/.checksums. Refresh only stale files. Print Refreshing — <comma-separated list of stale files>.
After analysis:
.hyperflow/.checksums (SHA256 of package.json, tsconfig.json, eslint/biome config, etc.).gitignore if .hyperflow/ is not already excludedCreate .hyperflow/memory/ if absent:
.hyperflow/memory/
├── index.md
├── learnings.md
├── decisions.md
├── pitfalls.md
├── patterns.md
├── conventions.md
└── archive/.gitkeepMigration: If ~/.claude/hyperflow-memory.md exists, migrate entries matching the current project path into the appropriate memory files. Tag migrated entries [migrated].
Offer to run scripts/setup-detection.sh --tools all to generate AGENTS.md, Cursor rules, GEMINI.md, and CLAUDE.md.
Flags — --tools <all|agents|cursor|gemini|claude>, --force, --dry-run.
Default — --tools all. Ask once via AskUserQuestion if the user wants to skip any tool.
Print what was created, skipped, and migrated (elegant style, no icons):
Hyperflow init complete
Created .hyperflow/{profile,architecture,conventions,dependencies,testing,git-workflow}.md
Created .hyperflow/.checksums
Created .hyperflow/memory/{index,learnings,decisions,pitfalls,patterns,conventions}.md
Skipped .gitignore entry — already present
Migrated 3 entries from ~/.claude/hyperflow-memory.md
Shims AGENTS.md, .cursor/rules, GEMINI.md, CLAUDE.mdThis skill does not auto-chain. Init is project setup, not feature work. When the user wants to start a feature, they invoke /hyperflow:spec (for ambiguous scope) or /hyperflow:scope (for clear specs).
Full rules in DOCTRINE.md. Output style in output-style.md.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 4,642 | 5,765 | +24% | 1 | 1 | 0% | 245 | 1,194 | +387% | 0 | 0 | — |
case-02 | fail→fail | 6,101 | 6,582 | +8% | 1 | 1 | 0% | 420 | 1,195 | +185% | 0 | 0 | — |
case-03 | fail→fail | 6,446 | 6,940 | +8% | 1 | 1 | 0% | 256 | 1,363 | +432% | 0 | 0 | — |
case-04 | fail→fail | 13,868 | 12,024 | -13% | 1 | 1 | 0% | 2,082 | 2,815 | +35% | 0 | 0 | — |
case-05 | fail→pass | 11,914 | 5,383 | -55% | 1 | 1 | 0% | 1,518 | 1,605 | +6% | 0 | 0 | — |
case-06 | fail→pass | 9,745 | 3,588 | -63% | 1 | 1 | 0% | 1,416 | 1,376 | -3% | 0 | 0 | — |
case-07 | fail→pass | 12,160 | 5,543 | -54% | 1 | 1 | 0% | 1,782 | 1,701 | -5% | 0 | 0 | — |
case-08 | fail→pass | 10,600 | 4,954 | -53% | 1 | 1 | 0% | 1,679 | 1,584 | -6% | 0 | 0 | — |
case-09 | fail→pass | 15,457 | 8,000 | -48% | 1 | 1 | 0% | 2,560 | 1,949 | -24% | 0 | 0 | — |
case-10 | fail→fail | 12,244 | 4,207 | -66% | 1 | 1 | 0% | 1,724 | 1,424 | -17% | 0 | 0 | — |
case-11 | pass→pass | 10,012 | 2,892 | -71% | 1 | 1 | 0% | 1,450 | 1,233 | -15% | 0 | 0 | — |
case-12 | fail→fail | 13,874 | 5,312 | -62% | 1 | 1 | 0% | 2,032 | 1,719 | -15% | 0 | 0 | — |
case-13 | fail→pass | 8,993 | 5,241 | -42% | 1 | 1 | 0% | 1,183 | 1,494 | +26% | 0 | 0 | — |
case-14 | fail→pass | 13,704 | 3,142 | -77% | 1 | 1 | 0% | 1,873 | 1,287 | -31% | 0 | 0 | — |
case-15 | fail→pass | 13,429 | 3,737 | -72% | 1 | 1 | 0% | 2,012 | 1,377 | -32% | 0 | 0 | — |
case-16 | fail→pass | 13,725 | 5,944 | -57% | 1 | 1 | 0% | 1,790 | 1,676 | -6% | 0 | 0 | — |
case-17 | fail→pass | 9,249 | 3,414 | -63% | 1 | 1 | 0% | 1,497 | 1,344 | -10% | 0 | 0 | — |
case-18 | fail→pass | 11,118 | 2,888 | -74% | 1 | 1 | 0% | 1,554 | 1,303 | -16% | 0 | 0 | — |
case-19 | fail→fail | 15,125 | 3,333 | -78% | 1 | 1 | 0% | 2,444 | 1,290 | -47% | 0 | 0 | — |
case-20 | pass→pass | 24,604 | 32,889 | +34% | 1 | 1 | 0% | 3,788 | 5,542 | +46% | 0 | 0 | — |
case-21 | fail→fail | 8,633 | 3,778 | -56% | 1 | 1 | 0% | 1,166 | 1,299 | +11% | 0 | 0 | — |
case-22 | pass→fail | 21,045 | 13,200 | -37% | 1 | 1 | 0% | 3,528 | 1,277 | -64% | 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. 22 cases were attempted, and 18 counted toward the lift figure. The other 4 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 +45 percentage points is the difference between those two pass rates over the 18 comparable cases. 1 case got worse with the skill loaded, and it is 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.