setup
Skill2 starsChanged 11 months agoMITView on GitHub
Run once after installing turbocharge. Audits global config for conflicts — duplicate agents, competing skills, stale rules — and offers to fix them. Also validates turbocharge plugin health.
We split this skill into the rules it gives an agent
19 rules. 0 are common, 7 are distinctive, and 12 are very specific to this project.
0
Common
Most of the 166 edited versions we read keep these.
7
Distinctive
This author's own choices. Worth a look.
12
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 53
Read ~/.claude/rules/common/agents.md (if exists):
- Very specific: line 58
Action: If agents.md doesn't point to turbocharge, offer to replace it with the turbocharge-aware version:
- Very specific: line 67
Run claude plugin list --json to enumerate installed plugins and their enable state, then read ~/.claude/settings.json enabledPlugins:
- Very specific: line 73
Run claude plugin details turbocharge to confirm the plugin is installed, enabled, and to see its standing token cost.
- Very specific: line 79
All 10 skills have SKILL.md files
- Distinctive: line 80
All 5 agents have .md files
- Very specific: line 106
Why this matters: Every global rule file is loaded into context on every turn (~4 bytes/token).
- Distinctive: line 110
Run this phase after the conflict audit completes and before the final report.
- Distinctive: line 125
Do not ask the user about anything detectable.
- Very specific: line 129
Ask at most these five, each with [skip] producing the template default:
- Very specific: line 145
If CLAUDE.md exists, AGENTS.md also exists, and CLAUDE.md has no @AGENTS.md line → offer to add @AGENTS.md as its first line.
- Distinctive: line 146
If CLAUDE.md exists and contains a block with the same marker → replace between markers, leave surrounding content untouched.
- Distinctive: line 147
If CLAUDE.md exists and does not contain that block → append to end of file.
- Very specific: line 148
If CLAUDE.md does not exist but AGENTS.md does → create CLAUDE.md with @AGENTS.md as its first line, then append the rendered blocks below …
- Very specific: line 149
If neither CLAUDE.md nor AGENTS.md exists → suggest /init first, do not create a bare CLAUDE.md from the template alone.
- Distinctive: line 153
Always show the diff and ask for confirmation before writing.
- Very specific: line 157
After writing, if CLAUDE.md exceeds 180 lines, warn the user and suggest extracting personal rules to ~/.claude/CLAUDE.md.
- Very specific: line 161
If ATLAS.md does not exist, offer: /turbocharge:atlas to generate the navigation index.
- Distinctive: line 198
Delete files without listing them first and getting confirmation
“Common” is measured: kept by at least 6 in 10 of the 166 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Setup Turbocharge
One-time audit and cleanup to make turbocharge your single orchestration system.
Announce: "Running turbocharge setup — auditing your config for conflicts."
The Goal
Turbocharge replaces ad-hoc agents, scattered skills, and custom commands with one pipeline. This setup finds and removes the overlap so Claude doesn't get confused about which system to use.
Audit Steps
1. Check for Competing Agents
Scan ~/.claude/agents/ for agent definitions that overlap with turbocharge agents:
| Turbocharge Agent | Conflicts With |
| builder | tdd-guide, implementer |
| planner | planner, architect |
| code-reviewer | code-reviewer |
| researcher | explorer, investigator |
| task-reviewer | spec-reviewer, quality-reviewer |
Also check for: session-wrapper, session-wrap, build-error-resolver (covered by turbocharge:debug)
Action: List conflicts and offer to delete them. Explain what turbocharge agent replaces each one.
2. Check for Duplicate Skills/Commands
Scan .claude/commands/ (project level) for commands that overlap with turbocharge skills:
| Turbocharge Skill | Conflicts With |
| wrap | session-wrap, wrap-up, session-wrapper |
| story | story-author, user-story, story-writer |
| plan | task-breakdown, plan-tasks, task-planner |
| review | code-review |
| build | implement, implement-story |
| brainstorm | brainstorm (if non-turbocharge) |
| debug | debug, troubleshoot |
Action: List conflicts with file paths. Offer to delete duplicates. Flag any that have unique functionality turbocharge doesn't cover (recommend keeping those).
3. Check agents.md Rule
Read ~/.claude/rules/common/agents.md (if exists):
- Does it reference turbocharge as the primary system? Good.
- Does it list phantom agents (agents not in
~/.claude/agents/)? Bad — offer to rewrite. - Does it reference both turbocharge AND other agent systems? Bad — offer to consolidate.
Action: If agents.md doesn't point to turbocharge, offer to replace it with the turbocharge-aware version:
# Agent Orchestration
## Primary System: Turbocharge Plugin
All orchestration goes through the turbocharge plugin...4. Check for Plugin Conflicts
Run claude plugin list --json to enumerate installed plugins and their enable state, then read ~/.claude/settings.json enabledPlugins:
- Are there disabled plugins cluttering the config? Offer to remove.
- Scan the installed-plugins list for competing orchestration plugins (e.g. superpowers, everything-claude-code) enabled alongside turbocharge — not just duplicate definitions in
~/.claude/agents/. Warn about conflicts and explain the overlap.
5. Validate Turbocharge Plugin Health
Run claude plugin details turbocharge to confirm the plugin is installed, enabled, and to see its standing token cost. Then run the validation script if available:
./scripts/validate.shOr manually check:
- All 10 skills have SKILL.md files
- All 5 agents have .md files
- hooks/hooks.json exists
- .claude-plugin/plugin.json exists and has correct version
6. Check Global Rules Alignment
Scan ~/.claude/rules/common/ for rules that conflict with turbocharge's iron laws:
- Testing rules should include "never ask permission to run tests"
- Development workflow should include "verify understanding before coding"
- Coding style should include a completion gate
- Reviews should require comprehensive scope by default
Action: List missing rules and offer to add them.
7. Check for Global Rules / Turbocharge Overlap
Scan ~/.claude/rules/common/ for files that duplicate what turbocharge skills already enforce:
| Global Rule File | Overlaps With | Recommendation |
agents.md | turbocharge pipeline (SessionStart bootstrap) | Keep only the "Primary System: Turbocharge Plugin" header + pipeline table. Remove agent dispatch details — turbocharge skills handle dispatch. |
development-workflow.md | turbocharge:plan, turbocharge:build | Remove TDD steps and plan-first sections — builder.md and planner.md enforce these. Keep only git workflow and research steps. |
testing.md | builder.md TDD mandate | Remove TDD workflow section — builder.md is the single source. Keep coverage targets and test types. |
Action: For each overlap found, show the user what's duplicated and offer to trim. Don't delete files — trim the overlapping sections and keep unique content.
Why this matters: Every global rule file is loaded into context on every turn (~4 bytes/token). Duplicate instructions between rules and turbocharge skills waste ~2,000 tokens per session and can cause conflicting guidance.
CLAUDE.md Phase
Run this phase after the conflict audit completes and before the final report.
1. Auto-Detect (no questions)
Probe the project root for these files and extract values:
| Signal | Inspect | Extract |
package.json | scripts.test, scripts.lint, top-level name | test command, lint command, project name |
pyproject.toml | [tool.poetry], [project], [tool.pytest.ini_options] | language (Python), test framework |
Cargo.toml | [package].name, [dev-dependencies] | language (Rust), test runner |
go.mod | module line | language (Go), module path |
pnpm-lock.yaml / yarn.lock / package-lock.json | presence | package manager |
| Primary entry | src/index.*, main.*, cmd/*/main.go | entry-point file |
Do not ask the user about anything detectable.
2. Interview (≤5 questions, each skippable)
Ask at most these five, each with [skip] producing the template default:
- Test discipline — Strict TDD / Tests-alongside / Tests-when-reasonable
- File-header convention — ABOUTME / JSDoc-style / None
- Naming style — camelCase / snake_case / mixed-by-language
- Debug protocol strictness — Always 4-phase / Non-trivial only
- Project-specific domain terms — free text; empty skips the block
The interview must be completable in under 90 seconds.
3. Render and Append
Read ${CLAUDE_PLUGIN_ROOT}/templates/CLAUDE-turbocharge.md. Substitute answers (test command, naming style, file-header convention, domain terms). Each block is delimited by <!-- turbocharge:NAME --> and <!-- /turbocharge:NAME -->. Versions 3.0–3.1 wrote the same blocks under <!-- turboflow:NAME --> / <!-- /turboflow:NAME -->: treat those as the same block — replace it in place and rewrite its markers to turbocharge; never append a duplicate. If both a turbocharge:NAME and a turboflow:NAME block exist (3.x setup appended a copy), update the turbocharge block and remove the turboflow duplicate.
CLAUDE.md here means ./CLAUDE.md, or ./.claude/CLAUDE.md when the project keeps it there — edit the one that exists, never create a second. Imports resolve relative to the file that contains them, so inside .claude/CLAUDE.md the import is @../AGENTS.md: write that form wherever the bullets below say @AGENTS.md, and treat either form as the import already being present.
- If CLAUDE.md exists,
AGENTS.mdalso exists, and CLAUDE.md has no@AGENTS.mdline → offer to add@AGENTS.mdas its first line. Claude Code ignores AGENTS.md whenever a CLAUDE.md exists, so without the import its instructions never load. Leave the file unchanged if the user declines. - If CLAUDE.md exists and contains a block with the same marker → replace between markers, leave surrounding content untouched.
- If CLAUDE.md exists and does not contain that block → append to end of file.
- If CLAUDE.md does not exist but
AGENTS.mddoes → create CLAUDE.md with@AGENTS.mdas its first line, then append the rendered blocks below it. Claude Code reads AGENTS.md only while no CLAUDE.md exists, so this import is what keeps AGENTS.md loading once CLAUDE.md is created. - If neither CLAUDE.md nor AGENTS.md exists → suggest
/initfirst, do not create a bare CLAUDE.md from the template alone.
4. Show Diff, Confirm, Write
Always show the diff and ask for confirmation before writing. Never modify CLAUDE.md silently.
5. Size Guard
After writing, if CLAUDE.md exceeds 180 lines, warn the user and suggest extracting personal rules to ~/.claude/CLAUDE.md.
6. Chain Forward
If ATLAS.md does not exist, offer: /turbocharge:atlas to generate the navigation index. This closes the bootstrap loop: /init → /turbocharge:setup → /turbocharge:atlas.
Report Format
TURBOCHARGE SETUP AUDIT
=======================
Plugin Health: ✅ OK (10 skills, 5 agents, hooks configured)
Conflicts Found:
🔴 ~/.claude/agents/session-wrapper-agent.md → covered by turbocharge:wrap
🔴 .claude/commands/story-author.md → covered by turbocharge:story
🟡 agents.md references phantom agents, not turbocharge
Recommendations:
1. Delete ~/.claude/agents/session-wrapper-agent.md
2. Delete .claude/commands/story-author.md
3. Rewrite agents.md to reference turbocharge
No Conflicts:
✅ .claude/commands/epic-author.md — unique, keep
✅ .claude/commands/consistency-review.md — unique, keep
Apply fixes? [list specific changes, ask for confirmation]After Setup
Offer to chain:
/turbocharge:brainstormif user has an idea to explore/turbocharge:planif user has requirements ready- Nothing if user just wanted the audit
Red Flags
Do NOT:
- Delete files without listing them first and getting confirmation
- Modify CLAUDE.md without showing the diff
- Assume any file is "safe to delete" — always explain what replaces it
- Skip the audit steps to "save time"
Other versions187 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 187
- anthropics/knowledge-work-plugins★ 34ksame name
- garrytan/gbrain★ 31ksame name
- gavrielc/nanoclaw★ 31ksame name
- nanocoai/nanoclaw★ 31ksame name
- qwibitai/nanoclaw★ 31ksame name
- Q00/ouroboros★ 6.3ksame name
- pytorch/executorch★ 5.1ksame name
- agenticnotetaking/arscontexta★ 3.5ksame name
- aws/agent-toolkit-for-aws★ 3.4ksame name
- spotify/portal-ai-plugins★ 2.5ksame name
177 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
Used alongsideProjects with it also use add-discord, add-slack and 10 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
MicrosoftSkilladd-discord
Add Discord bot channel integration via Chat SDK.
Also in 11 projects that use this
MicrosoftSkilladd-slack
Add Slack channel integration via Chat SDK.
Also in 11 projects that use this
juancruzmunozalbeloSkilladd-voice-transcription
Add voice message transcription to NanoClaw using OpenAI's Whisper API. Automatically transcribes WhatsApp voice notes so the agent can read and respond to them.
Also in 5 projects that use this
LambenthanSkilldaily-arxiv
每日从 arXiv 拉取新论文,过滤相关性,auto-ingest 高优先级论文,检测 SOTA 更新
Also in 5 projects that use this
LambenthanSkillexp-eval
实验判决门:Review LLM 独立评判实验结果 → 4 种判决路径 → 自动更新 claims confidence、ideas status、graph edges
Also in 5 projects that use this
- +7 more from microsoft
In this repository10 other files it gives its agents.
- .claude/CLAUDE.mdCLAUDE.md
- atlasSkill
- brainstormSkill
- buildSkill
- debugSkill
- planSkill
- reviewSkill
- shipSkill
- storySkill
- wrapSkill
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
Reviews can't be read right now.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

