lead-producer-oss
CLAUDE.md2 starsChanged 3 months agoMITView on GitHub
74 total skills: 1 coordinator + 73 specialist skills (46 roles, 13 teams, 14 workflows). This file is Claude Code host guidance. Canonical skill content lives in .claude/skills/. Codex uses the same skill files through linked installs and follows .codex/INSTALL.md for host-specific setup. Do not load skills speculatively. Only the Lead Producer is invoked directly, and it loads only the skills needed for the current task. Prohibited actions: - Do not read, scan, or invoke any SKILL.md file except lead-producer…
In this repository74 other files it gives its agents.
- lead-producerSkill
- role-analytics-engineerSkill
- role-backend-engineerSkill
- role-behavioral-economistSkill
- role-brand-strategistSkill
- role-code-reduction-engineerSkill
- role-community-developerSkill
- role-context-managerSkill
- role-cpp-engineerSkill
- role-ctoSkill
- role-database-engineerSkill
- role-data-engineerSkill
- role-devops-engineerSkill
- role-economistSkill
- role-economy-designerSkill
- role-education-researcherSkill
- role-elementary-math-teacherSkill
- role-elementary-teacherSkill
- role-frontend-engineerSkill
- role-game-balance-designerSkill
- role-game-designerSkill
- role-godot-engineerSkill
- role-go-engineerSkill
- role-icelandic-linguistSkill
- role-icelandic-teacherSkill
- role-java-engineerSkill
- role-kotlin-engineerSkill
- role-level-designerSkill
- role-liveops-engineerSkill
- role-move-sui-developerSkill
- role-narrative-designerSkill
- role-open-source-engineerSkill
- role-principal-software-engineerSkill
- role-product-managerSkill
- role-product-marketing-managerSkill
- role-python-engineerSkill
- role-qa-engineerSkill
- role-railway-deploymentSkill
- role-rust-engineerSkill
- role-scalability-engineerSkill
- role-security-engineerSkill
- role-software-architectSkill
- role-technical-product-managerSkill
- role-technical-writerSkill
- role-typescript-engineerSkill
- role-ui-ux-designerSkill
- role-unity-engineerSkill
- team-architecture-reviewSkill
- team-blue-teamSkill
- team-brand-teamSkill
- team-dev-teamSkill
- team-documentationSkill
- team-economy-teamSkill
- team-elementary-educationSkill
- team-frontend-teamSkill
- team-infrastructureSkill
- team-move-teamSkill
- team-open-sourceSkill
- team-product-teamSkill
- team-red-teamSkill
- The first 60 shown.
The file
Show its rules and text
We split this CLAUDE.md into the rules it gives an agent
16 rules. About 0 are common, 10 are distinctive, and 6 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈10
Distinctive · estimate
This author's own choices. Worth a look.
6
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 11
Do not load skills speculatively: Only the Lead Producer is invoked directly, and it loads only the skills needed for the current task.
- Very specific: line 14
Do not read, scan, or invoke any SKILL.md file except lead-producer unless Lead Producer explicitly requests it for the current task.
- Distinctive: line 15
Do not pre-read skills to "understand scope" or "prepare for routing."
- Distinctive: line 16
Do not bypass Lead Producer by invoking skills directly, even if the routing table shows an obvious match.
- Distinctive: line 17
Do not load context modules unless the user or Lead Producer explicitly requests them.
- Distinctive: line 18
Do not list or infer skill availability from file structure.
- Very specific: line 19
If the user names a specialist, play, or alias directly, still send the request through LP.
- Distinctive: line 23
Wait for LP's complete response.
- Very specific: line 24
If LP says Route Now: <skill-id>[, <skill-id> ...], load ONLY those exact skills.
- Very specific: line 25
If LP says Suggested Play: <skill-id>, do NOT load it yet.
- Distinctive: line 26
If the user follows up after a suggested play, names a specialist directly, or uses a legacy alias, send that follow-up back through LP.
- Distinctive: line 27
If a loaded skill identifies the need for an additional skill, it escalates back to LP, not directly to the next skill.
- Very specific: line 35
Use these after linking .claude into a project:
- Very specific: line 73
No field -> active (the default.
- Distinctive: line 90
When Claude needs routing semantics, follow LP's returned skill ids or the explicit opt-in aliases in the invocation protocol above.
- Distinctive: line 100
Do not load context modules by topic inference.
≈ “Common” is an estimate: at least 1 in 5 of the 27,641 CLAUDE.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Lead Producer Pack - Claude Code Host Guide
74 total skills: 1 coordinator + 73 specialist skills (46 roles, 13 teams, 14 workflows).
This file is Claude Code host guidance. Canonical skill content lives in .claude/skills/.
Codex uses the same skill files through linked installs and follows .codex/INSTALL.md for
host-specific setup.
Loading Rules
Do not load skills speculatively. Only the Lead Producer is invoked directly, and it loads only the skills needed for the current task.
Prohibited actions:
- Do not read, scan, or invoke any SKILL.md file except
lead-producerunless Lead Producer explicitly requests it for the current task. - Do not pre-read skills to "understand scope" or "prepare for routing."
- Do not bypass Lead Producer by invoking skills directly, even if the routing table shows an obvious match.
- Do not load context modules unless the user or Lead Producer explicitly requests them.
- Do not list or infer skill availability from file structure.
- If the user names a specialist, play, or alias directly, still send the request through LP.
lead-producer/SKILL.mdowns which names and opt-ins are treated as routing signals.
Lead Producer invocation protocol:
- In Claude Code, invoke
/lead-producerwith the user's full request. - Wait for LP's complete response. If LP asks a clarifying question, answer it; do not route to skills.
- If LP says
Route Now: <skill-id>[, <skill-id> ...], load ONLY those exact skills. If LP describes a domain but does not name a skill, ask LP to specify. - If LP says
Suggested Play: <skill-id>, do NOT load it yet. Wait for user opt-in. - If the user follows up after a suggested play, names a specialist directly, or uses a legacy alias, send that follow-up back through LP. Do not interpret undocumented routing phrases in this host guide.
- If a loaded skill identifies the need for an additional skill, it escalates back to LP, not directly to the next skill.
Lead Producer response modes:
- Route Now - LP has named one or more skill ids to load immediately.
- Suggested Play - LP recommends a deeper workflow by ID; the user must opt in before it loads.
Manual Smoke Test Prompts
Use these after linking .claude into a project:
/lead-producer Review this onboarding guide for clarity and accuracy.Manual smoke expectation: first lines includeRoute Now: team-documentationandSuggested Play: none/lead-producer This codebase is inherited and messy. Figure out whether we should do a focused architecture spike or a broader discovery pass first.Manual smoke expectation: first lines includeRoute Now: noneandSuggested Play: workflow-project-discovery/lead-producer This payout rollback plan is high stakes. Run the specialist hardening play and repeat until 9.Manual smoke expectation: first lines includeRoute Now: workflow-specialist-hardeningandSuggested Play: none/lead-producer We're about to migrate all player wallets to the new contract in one irreversible cutover next week. Pressure-test what could go wrong before we commit.Manual smoke expectation: first lines includeRoute Now: noneandSuggested Play: workflow-premortem/lead-producer I want to add a new specialist skill to this pack for accessibility audits.Manual smoke expectation: first lines includeRoute Now: workflow-author-skillandSuggested Play: none/lead-producer We're ending for today; summarize this session so another agent can pick it up tomorrow.Manual smoke expectation: first lines includeRoute Now: workflow-session-handoffandSuggested Play: none/lead-producer I have a vague idea to "make the rewards feel better" but haven't thought it through. Grill me before we build anything.Manual smoke expectation: first lines includeRoute Now: workflow-requirements-grillandSuggested Play: none/lead-producer Review the questline for this new zone — does the story land and stay consistent with our lore?Manual smoke expectation: first lines includeRoute Now: role-narrative-designerandSuggested Play: none/lead-producer Lay out the encounter flow for the new dungeon — which mechanics appear and in what order.Manual smoke expectation: first lines includeRoute Now: role-level-designerandSuggested Play: none/lead-producer Plan the go-to-market and positioning for our launch next month.Manual smoke expectation: first lines includeRoute Now: role-product-marketing-managerandSuggested Play: none/lead-producer Review my Godot scene architecture and GDScript for performance and pattern issues.Manual smoke expectation: first lines includeRoute Now: role-godot-engineerandSuggested Play: none/lead-producer Review this TypeScript service for type safety and async correctness.Manual smoke expectation: first lines includeRoute Now: role-typescript-engineerandSuggested Play: none/lead-producer Review my Unity MonoBehaviours and prefab structure for GC pressure and lifecycle issues.Manual smoke expectation: first lines includeRoute Now: role-unity-engineerandSuggested Play: none/lead-producer This query is slow and the migration touches our biggest table — review the schema, indexes, and rollout plan.Manual smoke expectation: first lines includeRoute Now: role-database-engineerandSuggested Play: none/lead-producer Review these Icelandic math exercises for 8-year-olds: do they teach the right thing, and is the Icelandic correct?Manual smoke expectation: first lines includeRoute Now: team-elementary-educationandSuggested Play: none/lead-producer Check that all the Icelandic strings in this UI are spelled and inflected correctly.Manual smoke expectation: first lines includeRoute Now: role-icelandic-linguistandSuggested Play: none
Skill Lifecycle
Skills carry an optional status frontmatter field:
- No field -> active (the default; do not add
status: active). status: draft-> unfinished, not ready to route.status: deprecated-> retired; include a one-line "replaced by<skill-id>" in the body.
Lead Producer must not route to draft or deprecated skills. Use workflow-author-skill to add or formalize skills so routing, counts, and docs stay consistent.
Hierarchy
User -> Claude -> Lead Producer -> specialist roles/teams/workflows
Claude's only job is to invoke Lead Producer, execute Route Now decisions, or wait on Suggested Play recommendations until the user opts in. Even if the user asks to skip LP and invoke a skill directly, route through LP first.
Routing Canon
Runtime routing truth lives in .claude/skills/lead-producer/SKILL.md. This host guide
intentionally does not duplicate LP's full route table.
When Claude needs routing semantics, follow LP's returned skill ids or the explicit opt-in aliases in the invocation protocol above. Do not expand the contract from memory or infer undocumented routes from file names.
Project Context Modules
This pack supports external context module packs. Context modules are loaded ONLY when:
- The user explicitly names them.
- Lead Producer explicitly names them in a routing decision.
Do not load context modules by topic inference. Context modules are reference material only; they do not make design decisions.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
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.

