Agent Teams Workflow

VerifiedSafe

Orchestrates a team of AI agents for autonomous parallel development. The lead plans, assigns file ownership, spawns teammates, and reviews; teammates follow TDD, own specific files, and communicate blockers. Helps scale large tasks by dividing work without file conflicts.

Sby Skills Guide Bot
DevelopmentAdvanced
6/2/2026
Claude Code
#team-workflow#autonomous-development#parallel-development#git-workflow

Recommended for

Automated summary

Provides a structured workflow for agent teams to work autonomously in parallel on a single feature branch, with a team lead orchestrating planning, spawning teammates, and reviewing integration.

Strengths

  • Clear separation of file ownership to avoid conflicts
  • Automated failure recovery with retry and logging
  • Integration of sub-agents for testing, code review, and e2e validation
  • Standardized git branch strategy for parallel development

Limitations

  • Requires careful task breakdown to ensure no overlapping file ownership
  • Relies on manual plan approval before execution
  • Sub-agents may introduce overhead if not properly configured
When to use it

Use this workflow when you have a large, multi-file feature that can be broken into independent tasks and multiple agents are available.

When not to use it

Avoid this workflow for small, single-file changes or when you need tight coordination on overlapping files.

Security analysis

Safe
Quality score85/100

The skill provides operational guidance for AI agent teams, with no executable commands, no destructive instructions, and no data exfiltration risks.

No concerns found

How this analysis is made

Examples

Start as team lead
/workflow-team lead
Start as teammate
/workflow-team teammate
Spawn example (from lead phase 2)
Create an agent team with 2 teammates: 1. "FeatureA" — Read feature-a-plan.md. Own files: src/featureA/**/*. 2. "FeatureB" — Read feature-b-plan.md. Own files: src/featureB/**/*. Use Opus for all teammates. Require plan approval.

name: workflow-team description: | Workflow for Agent Teams autonomous parallel development. Provides team lead orchestration guide and teammate execution protocol. Covers file ownership, communication, and merge strategy. argument-hint: "[lead|teammate]"

Agent Teams Workflow

Autonomous parallel development workflow for Agent Teams. Invoke with: /workflow-team lead or /workflow-team teammate


For Team Lead (/workflow-team lead)

Phase 1: Plan

| Step | Action | |------|--------| | 1 | Enter PlanMode | | 2 | Read CLAUDE.md, docs/PROJECT-STRUCTURE.md, docs/ROADMAP.md | | 3 | Analyze task scope, identify required files and dependencies | | 4 | Break work into tasks with clear file ownership (no overlapping files) | | 5 | Create detailed step-by-step plan with task breakdown | | 6 | Exit PlanMode → wait for plan approval |

After plan approval, create tasks via TaskCreate and spawn teammates immediately. No separate confirmation needed.

Phase 2: Execute (after user approval)

| Step | Action | |------|--------| | 1 | Switch to development branch, create a working branch for the team | | 2 | Spawn teammates with detailed prompts (see Spawn Example below) | | 3 | Use Delegate Mode (Shift+Tab) — do NOT implement yourself | | 4 | Monitor teammate progress, unblock as needed |

Enable Plan Approval for complex/risky tasks. All teammates work on the same feature branch (file ownership prevents conflicts).

Spawn Example

Create an agent team with N teammates:
1. "{name}" — Read {task-file-path}. Own files: {file-list}.
2. "{name}" — Read {task-file-path}. Own files: {file-list}.
Use Opus for all teammates. Require plan approval.

Phase 3: Review & Merge (after all teammates complete)

| Step | Action | |------|--------| | 1 | Run the project's test command (see CLAUDE.md Commands) to verify integration | | 2 | Run code-reviewer sub-agent on all changed files | | 3 | Run e2e-tester sub-agent to validate user flows | | 4 | Read report in /docs/reports/code-review/ → fix ALL issues where status ≠ "complete" | | 5 | Merge working branch → development | | 6 | Update ROADMAP.md and PROJECT-STRUCTURE.md |

Merge Strategy

main
 └── development
      └── {working-branch}  ← single branch, all teammates work here
           ├── teammate-A commits (owns: file-list-A)
           ├── teammate-B commits (owns: file-list-B)
           └── teammate-C commits (owns: file-list-C)

After all teammates done → Phase 3

Git Conventions

See workflow-commits.md


For Teammates (/workflow-team teammate)

Execution Steps

| Step | Action | |------|--------| | 1 | Read CLAUDE.md, docs/PROJECT-STRUCTURE.md, assigned task file | | 2 | Run unit-test-writer sub-agent (Red Phase). NEVER analyze patterns or write test code yourself — always delegate to the unit-test-writer subagent. | | 3 | Implement code to pass tests (Green Phase) → run the project's test command (see CLAUDE.md Commands) | | 4 | Run the project's coverage command (see CLAUDE.md Commands) | | 5 | Commit per workflow-commits.md | | 6 | Message lead: files changed, test results, remaining issues |

Teammate Rules

  • ONLY modify files assigned to you
  • NEVER touch files owned by another teammate
  • Shared files (barrel index.ts, routes.ts): message lead before modifying
  • New files: create freely within your task scope
  • Do NOT create branches — work on the feature branch created by lead

Failure Recovery (Autonomous)

IF any step fails:
  1. Log to docs/reports/failures/{teammate-name}-{timestamp}.md
  2. Retry SAME approach (1 attempt)
  3. Retry DIFFERENT approach (1 attempt)
  4. After 3 failures:
     → Message lead: "Blocked on [issue]. Attempted [approaches]."
     → Pick up next available task
     → DO NOT STOP

Communication

| Event | Action | |-------|--------| | Task complete | Message lead with summary | | Blocked by another task | Message lead, pick up next task | | Found issue in shared code | Message lead (don't fix directly) | | Need design decision | Message lead with options + recommendation |


Cost Notes

  • Use opus model for teammates
  • Teammates: NO code-reviewer — TDD cycle is the quality gate, lead handles all review in Phase 3
  • Lead runs code-reviewer + e2e-tester as the single review gate post-merge
  • Minimize sub-agent calls per teammate
  • Avoid broadcast messages — message lead directly

From the file to a first result

The download contains exactly the content displayed on this page. This copy may differ from the latest version in the source repository.

View the source and supporting files

This download contains SKILL.md only. Any scripts, references, templates and dependencies are not included. Get the required files from the source, preserving their directory structure, before continuing.

1. Download the file

Download SKILL.md

From the root of a test project, save the file as .skill-downloads/agent-teams-workflow-mm78nmrf/SKILL.md, or use the command below. This folder is for inspecting the skill before installation.

Command for macOS/Linux or WSL: requires a POSIX shell and curl. On Windows without WSL, use the download button and create the folder manually.

mkdir -p '.skill-downloads/agent-teams-workflow-mm78nmrf' && curl --fail --show-error --location 'https://skills-guides.com/api/skills/agent-teams-workflow-mm78nmrf/download' --output '.skill-downloads/agent-teams-workflow-mm78nmrf/SKILL.md.part' && mv '.skill-downloads/agent-teams-workflow-mm78nmrf/SKILL.md.part' '.skill-downloads/agent-teams-workflow-mm78nmrf/SKILL.md'

2. Check prerequisites

Read SKILL.md before using it. If it references scripts or external services, inspect them and prepare the required access in your test environment.

No recognized prerequisites section in this file. This does not mean it is self-contained: libraries, tools and access requirements may be mentioned elsewhere in the content or source repository.

The platforms on this page are catalog indications. The paths below follow tool documentation; they do not prove this skill has been run in those tools. No minimum tool version is confirmed for this listing.

3. Load the skill

After inspection, copy the complete folder into your project using one of the commands below, then open a new tool session. The command stops if a folder with the same name already exists: compare the files before replacing it.

Claude Code
test -f '.skill-downloads/agent-teams-workflow-mm78nmrf/SKILL.md' && test -s '.skill-downloads/agent-teams-workflow-mm78nmrf/SKILL.md' && mkdir -p '.claude/skills' && test ! -e '.claude/skills/workflow-team' && cp -R '.skill-downloads/agent-teams-workflow-mm78nmrf' '.claude/skills/workflow-team'

OpenSkills

OpenSkills must already be installed. The local install/read procedure is checked with OpenSkills 1.5.0, which requires Node.js ≥ 20.6.0. Catalog slugs are not resolved automatically.

Official OpenSkills installation
openskills install './.skill-downloads/agent-teams-workflow-mm78nmrf'

openskills read 'agent-teams-workflow-mm78nmrf'

Reading should display the installed file content and its base directory. It does not run the skill or install its dependencies.

For an agent that reads AGENTS.md, then sync the index. This command changes AGENTS.md: review the diff before using it.

openskills sync

Manual method or another tool: attach SKILL.md to a conversation, or open it in an agent that can read files. Include any required supporting resources. No universal folder is recognized by every tool.

4. Try a first task

Use a short, synthetic input: an API endpoint, a small document or an interface brief. In a new conversation, attach the file or give the agent its path, then adapt this request:

Read .skill-downloads/agent-teams-workflow-mm78nmrf/SKILL.md and apply its instructions to this example: [describe a test input]. Identify any missing files, dependencies or access. Produce the result requested by the skill without inventing data or running unreviewed scripts.

Expected result: output matching the skill instructions, or a precise list of missing prerequisites. A read confirmation alone does not validate a first use. If an external resource is unavailable, that path remains untested.

Related skills