Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when starting any conversation involving bundle-plugins — blueprinting, scaffolding, authoring, auditing, testing, optimizing, or releasing. Also use when feeling unsure which bundles-forge skill applies
.claude/skills/odradekai-using-bundles-forge/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 764% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 21% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 7% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 17% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 20% | 0% |
<SUBAGENT-STOP> If you were dispatched as a subagent to execute a specific task, skip this skill. </SUBAGENT-STOP>
Before invoking any bundles-forge skill on a target directory, verify the target is a bundle-plugin project:
skills/ directory?package.json?If neither exists, inform the user: "This directory doesn't appear to be a bundle-plugin project. Bundles Forge skills are designed for bundle-plugins (repositories where skills are the primary content). Would you like to create a new bundle-plugin here, or did you mean to point to a different directory?"
Exception: bundles-forge:auditing and bundles-forge:optimizing can also operate on individual skill folders or files — they don't require a full bundle-plugin project.
In Claude Code: Use the Skill tool. When you invoke a skill, its content is loaded — follow it directly.
In Cursor: Use the Skill tool.
In Gemini CLI: Skills activate via the activate_skill tool. See references/gemini-tools.md for tool mapping.
In Codex: Skills are discovered from ~/.agents/skills/. See references/codex-tools.md for tool mapping.
In OpenClaw: Skills auto-load from the bundle's skills/ directory. See references/openclaw-tools.md for tool mapping.
Skills use Claude Code tool names as the default. Non-Claude-Code platforms: see the tool mapping references in this directory for equivalents.
Invoke relevant skills BEFORE any response or action when working with bundle-plugins. If there's even a small chance a skill applies, invoke it to check.
User message about bundle-plugins
→ Might any skill apply?
→ yes → Invoke Skill tool → Follow skill → Respond
→ no → Respond directlyThese skills diagnose, decide, and delegate. They orchestrate other skills to accomplish multi-step goals.
| Skill | Role | When to Use | |-------|------|-------------| | bundles-forge:blueprinting | New-project orchestrator | Planning new bundle-plugins, splitting complex skills, or composing skills into bundles. Orchestrates the full creation pipeline: scaffolding → authoring → workflow design → auditing | | bundles-forge:optimizing | Improvement orchestrator | Engineering optimization, feedback iteration, descriptions, tokens, adding skills, restructuring workflows. Delegates content changes to authoring | | bundles-forge:releasing | Release pipeline orchestrator | Version management, release pipeline: audit, test, version bump, publish |
These skills do one thing well. They can be invoked directly by users or dispatched by orchestrators.
| Skill | Role | When to Use | |-------|------|-------------| | bundles-forge:scaffolding | Structure generator | Generating project structure, adding or removing platform support | | bundles-forge:authoring | Content writer | Writing or improving SKILL.md content and agent definitions (agents/.md) | | bundles-forge:auditing | Diagnostic reporter | Reviewing a project for quality issues, security risks — outputs reports, does not orchestrate fixes | | bundles-forge:testing | Dynamic verifier | Testing a plugin locally — dev-marketplace setup, hook smoke tests, component discovery, cross-platform readiness |
| Skill | Purpose | |-------|---------| | bundles-forge:using-bundles-forge | Bootstrap meta-skill — you're reading it now (auto-loaded by hooks) |
When multiple skills could apply, prefer orchestrators over executors:
bundles-forge:blueprinting (orchestrates scaffolding, authoring, auditing)bundles-forge:optimizing (orchestrates authoring, scaffolding, auditing)bundles-forge:releasing (orchestrates auditing, testing, optimizing)bundles-forge:authoring (when you just need to write/improve a SKILL.md)bundles-forge:scaffolding (when you just need to add/remove a platform)bundles-forge:auditing (when you just need a diagnostic report)bundles-forge:testing (when you just need to verify a plugin works locally)dev-workflows, data-tools)name frontmatter field<project>:<skill-name>using-<project>agents/<role>.mdThe skill itself declares which type it is.
routing-context — skill routing table, platform adaptation guidance, and instruction priority for the current session| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 12,402 | 1,613 | -87% | 1 | 1 | 0% | 2,017 | 1,574 | -22% | 0 | 0 | — |
case-02 | fail→pass | 4,004 | 4,345 | +9% | 1 | 1 | 0% | 249 | 2,151 | +764% | 0 | 0 | — |
case-03 | fail→fail | 14,332 | 1,623 | -89% | 1 | 1 | 0% | 2,520 | 1,531 | -39% | 0 | 0 | — |
case-04 | pass→pass | 10,113 | 3,055 | -70% | 1 | 1 | 0% | 1,693 | 1,793 | +6% | 0 | 0 | — |
case-05 | fail→pass | 8,574 | 2,397 | -72% | 1 | 1 | 0% | 1,449 | 1,750 | +21% | 0 | 0 | — |
case-06 | fail→pass | 9,777 | 2,468 | -75% | 1 | 1 | 0% | 1,655 | 1,769 | +7% | 0 | 0 | — |
case-07 | fail→pass | 8,084 | 1,577 | -80% | 1 | 1 | 0% | 1,355 | 1,586 | +17% | 0 | 0 | — |
case-08 | fail→pass | 7,960 | 1,973 | -75% | 1 | 1 | 0% | 1,389 | 1,660 | +20% | 0 | 0 | — |
case-09 | fail→pass | 8,710 | 2,005 | -77% | 1 | 1 | 0% | 1,597 | 1,703 | +7% | 0 | 0 | — |
case-10 | fail→pass | 9,567 | 6,026 | -37% | 1 | 1 | 0% | 1,716 | 1,674 | -2% | 0 | 0 | — |
case-11 | fail→pass | 6,163 | 4,755 | -23% | 1 | 1 | 0% | 963 | 1,626 | +69% | 0 | 0 | — |
case-12 | fail→pass | 5,484 | 1,843 | -66% | 1 | 1 | 0% | 903 | 1,607 | +78% | 0 | 0 | — |
case-13 | fail→pass | 9,643 | 2,985 | -69% | 1 | 1 | 0% | 1,505 | 1,829 | +22% | 0 | 0 | — |
case-14 | fail→pass | 11,080 | 2,006 | -82% | 1 | 1 | 0% | 2,009 | 1,648 | -18% | 0 | 0 | — |
case-15 | fail→pass | 10,434 | 2,822 | -73% | 1 | 1 | 0% | 1,874 | 1,810 | -3% | 0 | 0 | — |
case-16 | fail→pass | 7,010 | 1,837 | -74% | 1 | 1 | 0% | 1,308 | 1,579 | +21% | 0 | 0 | — |
case-17 | pass→pass | 13,674 | 5,559 | -59% | 1 | 1 | 0% | 2,346 | 2,257 | -4% | 0 | 0 | — |
case-18 | pass→pass | 6,341 | 1,343 | -79% | 1 | 1 | 0% | 993 | 1,536 | +55% | 0 | 0 | — |
case-19 | pass→pass | 9,975 | 4,388 | -56% | 1 | 1 | 0% | 1,663 | 2,018 | +21% | 0 | 0 | — |
case-20 | fail→fail | 11,332 | 3,909 | -66% | 1 | 1 | 0% | 1,668 | 2,032 | +22% | 0 | 0 | — |
case-21 | pass→pass | 5,960 | 2,282 | -62% | 1 | 1 | 0% | 1,023 | 1,690 | +65% | 0 | 0 | — |
case-22 | pass→pass | 2,289 | 2,257 | -1% | 1 | 1 | 0% | 378 | 1,674 | +343% | 0 | 0 | — |
case-23 | pass→pass | 3,837 | 2,908 | -24% | 1 | 1 | 0% | 616 | 1,842 | +199% | 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. The headline lift of +57 percentage points is the difference between those two pass rates over the 23 comparable cases.
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.