subagents-orchestration-guide
Guides subagent coordination through implementation workflows. Use when orchestrating multiple agents, managing workflow phases, or determining autonomous execution mode.
npx skills add shinpr/claude-code-workflows --skill subagents-orchestration-guide --agent claude-code
Same command for any agent — swap --agent for codex, cursor, copilot.
Weekly change comes from our own snapshots, not the repository page — it measures attention, not adoption.
What it does
Guides subagent coordination through implementation workflows. Use when orchestrating multiple agents, managing workflow phases, or determining autonomous execution mode. The role of the orchestrator is to own workflow decisions, routing, progress management, user interaction, investigation and validation, and explicit mechanical operations using any available tool. Specialists handle explicitly assigned investigation and semantic deliverable creation or modification and are invoked before producing or changing code, tests, configuration, documents, task files, or other artifacts.
Workflow Subagent Context — Mandatory
Specifies that the smallest valid Agent prompt should reduce handoffs to exactly the exhaustive input-contract fields, preserving each value’s meaning and applying only the declared serialization. This supersedes general prompt self-containment to avoid conflicting context.
First Action Rule
On a new full-cycle task, pass user requirements to requirement-analyzer and judge convergence and Structural Scale with its signals. Build and judge the convergence record in the orchestrator with the requirement-convergence skill. Run its hearing protocol at the requirements stop point. Re-invoke requirement-analyzer only if an answer changes the repository analysis target or scope evidence. ADR qualification occurs only after credible options from codebase-analyzer and confirmed scope.
Requirement Change Detection During Flow
Treat new or changed behaviors as requirement changes, re-run requirement-analyzer, identify affected artifacts or task boundaries, and resume from the earliest invalidated gate while preserving valid outputs.
Orchestration Principles
Outcome Stewardship
Direct workflow toward the smallest sufficient set of deliverables, evaluating proposals against this boundary. Preserve specialist evidence ownership and approved artifacts as semantic sources.
Delegation Boundary: What vs How
Pass governing requirement source and specialist’s expected action. Each specialist determines execution method from repository evidence. If unable to determine method, escalate to the user.
Review Resolution
Apply references/review-resolution.md to actionable deliverable-review findings; orchestrator routes work, specialist creates or changes deliverables.
Task Assignment with Responsibility Separation
Lists responsibilities for task-executor and quality-fixer; frontend/back-end variants exist for those roles.
How to Call Subagents
Execution Method
Each subagent invocation is a fresh Agent tool call with fields: subagent_type, description, and prompt serialized from the input contract. A SendMessage resume reuses context; otherwise isolation is preserved.
Orchestrator Execution Boundary
Tool choice does not define responsibility; the orchestrator may use any tool for mechanical operations, while specialists perform semantic deliverable creation or modification.
Prompt Construction Rule
Pass the active workflow’s input contract fields in their declared forms; the prompt consists of those fields and values; artifact paths and unchanged outputs carry their semantics.
Handling Requirement Changes
Use create mode for initial documents; for requirement-driven revisions, invoke the owning document specialist in update mode and append history per role (work-planner, technical-designer/prd-creator, document-reviewer).
Basic Flow: Planning and Implementation
Details various planning sequences by scale (Small, Medium, Large) and steps to execute through requirement-analyzer, codebase-analyzer, ADRs, Design Doc, UI analysis, and further workflow steps.
Autonomous Execution Mode
Outlines pre-execution gate, autonomous execution summary for Medium/Large, and a 4-step per-task cycle within batch approval. Includes post-implementation verification criteria for code-verifier and security-reviewer, and escalation rules for blocked or escalation-needed states.
Task Management: 4-Step Cycle
Begins describing a per-task cycle, with steps elaborated beyond the truncated input.
# Subagents Orchestration Guide ## Role: The Orchestrator The orchestrator owns workflow decisions, routing, progress management, user interaction, the investigation and validation needed for those decisions, and explicitly assigned mechanical operations, using any available tool. Named specialists own explicitly assigned investigation and semantic deliverable creation or modification; invoke them before producing or changing code, tests, configuration, documents, task files, or other artifacts. ### Workflow Subagent Context — Mandatory This workflow's specialists are already self-contained through their agent definitions, loaded skills, and referenced artifacts. The smallest valid Agent prompt is the most reliable: reduce each handoff to exactly the exhaustive input-contract fields. Preserve each value's meaning from its authoritative source and apply only the serialization declared for that field. This rule supersedes general-purpose prompt self-containment because added context competes with the specialist's loaded process and can prevent coherent completion. ### First Action Rule When receiving a new full-cycle task, pass user requirements directly to requirement-analyzer. Use
- Role: The Orchestrator
- Workflow Subagent Context — Mandatory
- First Action Rule
- Requirement Change Detection During Flow
- Orchestration Principles
- Outcome Stewardship
- Delegation Boundary: What vs How
- Review Resolution
- Task Assignment with Responsibility Separation
- Constraints Between Subagents
- Explicit Stop Points
- Scale Determination and Document Requirements
- How to Call Subagents
- Execution Method
What does the subagents-orchestration-guide skill do?
Guides subagent coordination through implementation workflows. Use when orchestrating multiple agents, managing workflow phases, or determining autonomous execution mode.
How do I install it?
Run `npx skills add shinpr/claude-code-workflows --skill subagents-orchestration-guide --agent claude-code` — it drops the skill into your project so the agent can pick it up. Swap the --agent value for codex, cursor or copilot if you use one of those.
Where does this skill come from?
From shinpr/claude-code-workflows, a repository with 664 stars. We read it straight from the repository tree rather than a submitted listing, so what you see here is what is actually published.
Is a popular skill a good skill?
Not necessarily. Stars measure attention, not adoption — a repository can trend for a week and be abandoned. That is why we show the weekly change from our own snapshots next to the total, instead of a single flattering number.