orchestrate
Skill3 starsChanged 2 months agoMITView on GitHub
You are the responsible agent. Loading this skill means following the delivery procedure here, not launching another orchestrator. Keep the reasoning and responsibility established with the user through implementation and delivery. The default flow is: For multiple selected Tickets, dispatch-tickets provides Mission supervision and starts or resumes each Ticket's owner. An owner launched by that dispatcher follows this same procedure; it does not add another coordinator. Preserve existing ownership when resuming work. A context that performed a Prompt Audit of…
161 other projects use a version of this.
Adopted by
Most recently taken up by
- alexratmanpl/business-agent-skillstheir own version, first seen Nov 8, 2026
- gtonkov/claude-codetheir own version, first seen Nov 8, 2026
- edilsonvilarinho/usage-monitortheir own version, first seen Nov 8, 2026
- nathanonn/agent-skillstheir own version, first seen Nov 8, 2026
- peterpme/skillstheir own version, first seen Nov 8, 2026
- GDriem/agent-skillstheir own version, first seen Nov 8, 2026
Used with
Projects with it also use define-requirements, orchestrate-plan 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.
sson0-erSkilldefine-requirements
Work with the user, one question at a time, to complete the requirements document (docs/claudex/<slug>/01-requirements.md). Run by the orchestrator (parent). Stays out of design and implementation.
Also in 3 projects that use this
EternallLightSkillorchestrate-plan
Use when executing a locked, reviewed PLAN.md by delegating the implementation to visible worker agents in Orca, herdr, or cmux ONE PHASE at a time — splitting the plan into context-window-sized phases, giving every phase a pristine worker context, and (when the plan touches INDEPENDENT repos) running one worker per repo IN PARALLEL. Triggers on "run the plan with a worker", "execute the plan in the other pane", "split the plan into phases and build it with workers", "build the backend and frontend in parallel panes". Not for edits you can finish inline; outside Orca, herdr, or cmux use tgc-skills:orchestrate.
Also in 3 projects that use this
sson0-erSkillrecord-evidence
How to record external facts (docs, command output, codebase findings) as evidence and design choices as decisions. Used by the researcher, designer and reviewer agents.
Also in 3 projects that use this
PinkneyJiangSkillimprove-architecture
扫描代码库中的架构摩擦点,识别浅模块并挖掘深模块演进机会。生成可视化交互式 HTML 诊断报告,并在用户确认后闭环实施深度重构。
Also in 4 projects that use this
kjnezSkillticket
Work on a JIRA/Linear ticket end-to-end: read ticket details, explore codebase, create a branch, implement with TDD, run quality checks, update ticket status, and create a PR. Use when the user provides a ticket ID to implement. Triggers on requests like "/ticket PROJ-123", "work on this ticket", "implement PROJ-123".
Also in 4 projects that use this
xfstudioSkillreverse-engineer
Expert reverse engineer specializing in binary analysis,
Also in 3 projects that use this
Other versions161 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 161
- Significant-Gravitas/AutoGPT★ 188ksame name
- rohitg00/pro-workflow★ 2.9ksame name
- apmantza/pi-lens★ 521same name
- boshu2/agentops★ 453same name
- quantumaikr/quant.cpp★ 403same name
- Noizefield/audio-plugin-coder★ 334same name
- kagenti/kagenti★ 303same name
- varun29ankuS/shodh-memory★ 294same name
- provencher/codex-skills★ 271same name
- first-fluke/fullstack-starter★ 226same name
151 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.
In this repository31 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- codebase-designSkill
- code-reviewSkill
- diagnosing-bugsSkill
- dispatch-ticketsSkill
- grill-with-docsSkill
- implementSkill
- improve-codebase-architectureSkill
- prototypeSkill
- researchSkill
- resolving-merge-conflictsSkill
- setup-omskillsSkill
- tddSkill
- to-specSkill
- to-ticketsSkill
- triageSkill
- wayfinderSkill
- cavemanSkill
- designSkill
- excalidrawSkill
- grill-meSkill
- handoffSkill
- model-routingSkill
- prompt-comprehension-auditsSkill
- teachSkill
- tmux-workerSkill
- voiceSkill
- wormholeSkill
- write-a-skillSkill
- writing-great-skillsSkill
The file
We split this skill into the rules it gives an agent
7 rules. 0 are common, 7 are distinctive, and 0 are very specific to this project.
0
Common
Most of the 144 edited versions we read keep these.
7
Distinctive
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 28
Before reading any relative link, run
- Distinctive: line 35
Read repository instructions, affected code/tests, and the accepted request.
- Distinctive: line 73
Check candidate identity before consequential changes, inspect the complete diff, and run the relevant checks.
- Distinctive: line 87
Follow code-review's review policy for the complete candidate.
- Distinctive: line 97
After asynchronous acceptance, retain the review identity and release the turn.
- Distinctive: line 114
Verify final candidate identity, status, complete diff, acceptance criteria, and required checks.
- Distinctive: line 149
After discussing #42, the user says, “Fix it and push.
“Common” is measured: kept by at least 6 in 10 of the 144 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
Orchestrate
You are the responsible agent. Loading this skill means following the delivery procedure here, not launching another orchestrator. Keep the reasoning and responsibility established with the user through implementation and delivery. The default flow is:
current owner: investigate → implement and test → independent reviewer
→ adjudicate and correct → verify and deliverFor multiple selected Tickets, dispatch-tickets provides Mission supervision and starts or resumes each Ticket's owner. An owner launched by that dispatcher follows this same procedure; it does not add another coordinator. Preserve existing ownership when resuming work. A context that performed a Prompt Audit of an implementation contract hands that implementation to a fresh owner under the audit's isolation rules.
Before reading any relative link, run
python3 -c 'from pathlib import Path; import sys; print((Path(sys.argv[1]).resolve().parent / sys.argv[2]).resolve())' '<loaded-file-path>' '<relative-link>'
with this file's actual path and the link, then read the printed path. Pass
resolved skill paths to fresh agents.
1. Establish the work and stopping point
Read repository instructions, affected code/tests, and the accepted request. For tracked work, read the complete Ticket, relevant comments, configured tracker operations, governing sources, and dependency results. Follow setup-omskills if required configuration is missing. An untracked request needs no Ticket or tracker setup merely to use this skill.
Reuse established scope, decisions, verification requirements, and delivery boundary. The user's presence changes how unresolved Questions are handled, not who owns the work. “I'm leaving; continue through X” keeps this conversation responsible through X. Record the decisions and recoverable state needed to continue; absence alone adds no Mission, readiness, or Prompt Audit gate. Honor an audit or other gate when the accepted task or repository requires it.
Inspect the live repository and candidate ownership. Use the existing checkout when appropriate for direct work. Each Mission Ticket owns an exclusive branch and worktree. Follow WORKTREES.md when creating, reusing, or cleaning temporary artifacts. Record the candidate path, branch, starting HEAD, and fixed review base; preserve unrelated work. For integration, verify each predecessor's exact produced commit and delivery evidence before combining it.
This step is complete when scope, candidate, capabilities, and stopping point permit work, or the concrete missing prerequisite is identified.
2. Implement and verify
Investigate, implement, test, and debug while retaining technical ownership. Use diagnosing-bugs for reproduction-first bug work and tdd for test-first changes. Bounded assistance and explicit user delegation choices can support the work without transferring the owner's decisions or introducing a coordinator layer.
Apply user steering directly to the work and relevant helpers. Update affected durable authority when the contract changes materially. Resolve routine choices from accepted sources; consult an available maintainer when missing input blocks progress. While the maintainer is away, continue what the established authority resolves and preserve a concrete blocker for what it does not.
Check candidate identity before consequential changes, inspect the complete diff, and run the relevant checks. Failing checks and correctable defects are work to finish, not reasons to hand responsibility elsewhere or request routine approval. For every Mission Ticket, including integration, commit the complete candidate and record the exact full base and review SHAs. Review the whole base-to-review range, including committed predecessor inputs. Direct work may instead use complete WIP review when that state contains the entire candidate; otherwise use the complete committed range under the review contract.
This step is complete when the candidate covers the accepted requirements and is ready for the applicable review, or a genuine blocker has a recoverable handoff.
3. Review and resolve findings
Follow code-review's review policy for the complete candidate. Purely editorial documentation may be self-reviewed; code, behavior, and governing-authority changes require one fresh independent adversarial pass. For that pass, supply the resolved skill path, concise contract, repository instructions, candidate identity and complete range/state, and verification results. Use the environment's isolated subagents or tmux-worker; read model-routing before a model-selectable launch and honor active harness authorization and inheritance.
For delegated review, follow the transport's actual result-delivery contract. After asynchronous acceptance, retain the review identity and release the turn; resume from its matching completed result without requiring user approval. Unknown acceptance requires evidence recovery before another launch. Recover the full findings and completed outcome; partial or interrupted review is not completed review.
Adjudicate every finding, fix confirmed in-scope problems, and verify corrections in this conversation. Review is a one-way findings handoff, not an approval loop: continue resolving the work without sending corrections back for reviewer approval. Record the reviewed and final states honestly, including later changes.
This step is complete when the applicable review is complete, every finding has an evidence-based disposition, and confirmed in-scope corrections are verified. Record whether review was independent or permitted editorial self-review.
4. Deliver to the authorized boundary
Verify final candidate identity, status, complete diff, acceptance criteria, and required checks; reuse results only where relevant inputs are unchanged. Record reviewed/final states, findings/dispositions, checks, and limitations in the appropriate delivery evidence. Follow the requested delivery method and repository conventions; stop before publication if that is the agreed boundary.
- For a branch artifact, push and record its remote reference and exact full commit; preserve it for its integration consumer.
- For target delivery, use the agreed direct-push or PR method. A PR is optional unless required. When used, squash-merge, verify target/content equivalence, and record source-to-squash mapping.
- For integration, review and verify the complete combined candidate and record exact predecessor-to-result mappings before dependents proceed.
Complete applicable tracker obligations and eligible cleanup under WORKTREES.md. Preserve protected artifacts with reasons. If an operation fails, check what actually completed before recovery; publication alone does not prove the whole requested delivery is complete.
Continue through the user's stopping point without inserting approvals between implementation, review, corrections, and delivery. Progress messages need not pause work. Report there, on an explicit stop, or when a genuine blocker prevents further authorized progress. Transport-required turn release is continuation, not a request for the user to restart the task.
Return concise evidence and any remaining obligation to the user or dispatcher. For incomplete work, retain candidate state, conversation, checks, hypothesis, and next action; prefer the same owner on resumption. Settle prior writing activity before any replacement. Protect private operational notes and secrets.
This step is complete when the agreed boundary is verified and required cleanup is complete, or the precise blocker/recovery state has been reported.
Example
After discussing #42, the user says, “Fix it and push; I'll be away.” Keep #42 here, reproduce the bug, implement and test the fix, and obtain an independent review. Resolve its findings, verify, push, and clean eligible temporary artifacts before reporting. Neither the user's departure nor loading this skill creates another coordinator. For a dispatched Ticket, return the same delivery evidence to the supervising dispatcher.
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.

