Manages coordination between multiple dev agents working on a single project feature or goal.

Install

mkdir -p .claude/skills/dev-manager && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/12021" && unzip -o skill.zip -d .claude/skills/dev-manager && rm skill.zip

Installs to .claude/skills/dev-manager

Activation

This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.

Orchestrate multiple dev agents (spawned subagents OR human-driven worktrees) against ONE story-level outcome — hold the e2e vision, decompose into non-overlapping linked tasks with freeze/secrets guardrails baked in, monitor the tasks for movement, and intervene only on collision or drift. Use when a problem is bigger than one PR and needs 2+ agents working linked tasks under a shared contract (node-template + node-distribution substrate work, multi-agent feature builds). Triggers: "manage these devs", "coordinate subagents", "split this into tasks for N agents", "hold the story while agents work the tasks", "dev manager".
631 chars✓ has a “when” triggerlonger than Claude Code's old 250-char listing cap (fine on current versions)
Advanced

Key capabilities

  • →Decompose a story into non-overlapping linked tasks
  • →Inject guardrails into every task
  • →Monitor linked tasks for movement
  • →Intervene on collision or drift between tasks
  • →Route finished tasks to done

How it works

This skill holds the end-to-end vision for a story, decomposes it into linked tasks with guardrails, and monitors task progress. It intervenes only when tasks collide or drift from their contract.

Inputs & outputs

You give it
One story-level outcome
You get back
A status matrix of linked tasks

When to use dev-manager

  • →Coordinate multiple dev agents
  • →Manage complex feature implementation
  • →Split tasks across agents

About this skill

dev-manager

The contract lives in the hub entry dev-manager-contract. This file only triggers it. The loaded session cognition agent-contract owns all normal communication.

Read the full body before acting, using the current node's origin (operator: https://cognidao.org) and the COGNI_NODE_API_KEY saved in .env.cogni:

curl -fsS -H "Authorization: Bearer $COGNI_NODE_API_KEY" \
  "https://cognidao.org/api/v1/knowledge/dev-manager-contract" | jq -er .content

Its lifecycle table is the checklist every task must pass. Also read cicd-e2e-required-sequence the same way; it holds the operator calls for review, flight, merge and promote.

If the fetch fails, retry once. If it still fails, report the failure and do not improvise the lifecycle.

When not to use it

  • →Problems smaller than one PR
  • →When only one agent is working
  • →When a human wants to implement directly

Limitations

  • →No manager-authored, dev-readable, free-text direction field for pre-PR decision points
  • →Work item GET requests do not return body or summary
  • →Claims do not appear in assignees or status

How it compares

This workflow coordinates multiple agents on linked tasks under a shared contract, rather than individual agents working in isolation.

Compared to similar skills

dev-manager side by side with the closest alternatives in the catalog.

SkillInstallsUpdatedSafetyDifficulty
dev-manager (this skill)05moCautionAdvanced
chief-architect611moNo flagsAdvanced
ralplan16moReviewIntermediate
moai-workflow-thinking27moNo flagsIntermediate

Try saying

Example prompts that trigger this skill in your AI assistant.

Search skills

Search the agent skills registry