Subagent Driven Development
Run an approved implementation plan or a batch of unrelated bugs by dispatching a fresh subagent per task with mandatory review gates between steps.
Overview
Subagent-driven development is an agent skill most often used in Build (also Ship/review) that executes plans by assigning each task to a fresh subagent with review gates between steps.
Install
npx skills add https://github.com/neolabhq/context-engineering-kit --skill subagent-driven-developmentWhat is this skill?
- Fresh subagent per task or issue to avoid context pollution in the parent session
- Mandatory code or output review after each task or batch to catch regressions early
- Sequential mode for tightly coupled, order-dependent tasks
- Parallel mode for unrelated files, subsystems, or bugs with no shared state
- Stays in the same session—no human-in-the-loop between every micro-step
- Sequential and parallel execution modes
- Trigger: 3+ independent issues without shared state
Adoption & trust: 557 installs on skills.sh; 1.1k GitHub stars; 3/3 security scanners passed (skills.sh audits).
What problem does it solve?
You have an implementation plan or several unrelated bugs, but one long agent session mixes context and skips review until everything is already merged in your head.
Who is it for?
Solo builders running multi-task implementation plans in Claude Code or similar agents who want Superpowers-style quality gates without pausing for human review after every line.
Skip if: Single cohesive refactors that need one shared mental model across files, or work where subagents cannot share the repo state your plan assumes.
When should I use this skill?
Executing implementation plans with independent tasks in the current session, or facing 3+ independent issues investigatable without shared state or dependencies.
What do I get? / Deliverables
Each task or issue is completed in an isolated subagent pass with reviewed output before the next dispatch, so you iterate faster with fewer context-driven mistakes.
- Completed tasks per plan item
- Review notes or approvals between dispatches
- Consolidated session outcome without parent-context overload
Recommended Skills
Journey fit
Spans multiple journey phases - primary shelf plus alternate fits below.
Plan execution and task orchestration live in Build/PM—the canonical shelf for turning specs into shipped work via agents. Subagent-driven development is explicitly for executing implementation plans and independent issue queues, which is project-management-style agent workflow rather than a single integration hook.
Where it fits
Walk task-by-task through a written plan, dispatching a subagent per checkbox and reviewing diffs before the next task.
Fix three unrelated API bugs in parallel subagents when each touches different packages with no ordering dependency.
Run a review pass after each subagent batch before merging, treating review as the gate between dispatches.
Triage a backlog of independent production issues by spinning an investigator subagent per ticket.
How it compares
Use instead of one monolithic agent thread for long plans—quality gates per task, not a single end-of-day review.
Common Questions / FAQ
Who is subagent-driven development for?
Indie and solo developers using agentic coding tools who execute structured plans or triage multiple independent issues in one session.
When should I use subagent-driven development?
During Build when executing an implementation plan with separable tasks; during Ship/review when batching unrelated bug investigations; whenever you face three or more independent issues without shared dependencies.
Is subagent-driven development safe to install?
It is procedural guidance for dispatching subagents and reviewing their work—review the Security Audits panel on this Prism page before installing from the upstream repo.
Workflow Chain
Requires first: writing plans
Then invoke: requesting code review
SKILL.md
READMESKILL.md - Subagent Driven Development
# Subagent-Driven Development Create and execute plan by dispatching fresh subagent per task or issue, with code and output review after each or batch of tasks. **Core principle:** Fresh subagent per task + review between or after tasks = high quality, fast iteration. Executing Plans through agents: - Same session (no context switch) - Fresh subagent per task (no context pollution) - Code review after each or batch of task (catch issues early) - Faster iteration (no human-in-loop between tasks) ## Supported types of execution ### Sequential Execution When you have a tasks or issues that are related to each other, and they need to be executed in order, investigating or modifying them sequentially is the best way to go. Dispatch one agent per task or issue. Let it work sequentially. Review the output and code after each task or issue. **When to use:** - Tasks are tightly coupled - Tasks should be executed in order ### Parallel Execution When you have multiple unrelated tasks or issues (different files, different subsystems, different bugs), investigatin or modifying them sequentially wastes time. Each task or investigation is independent and can happen in parallel. Dispatch one agent per independent problem domain. Let them work concurrently. **When to use:** - Tasks are mostly independent - Overral review can be done after all tasks are completed ## Sequential Execution Process ### 1. Load Plan Read plan file, create TodoWrite with all tasks. ### 2. Execute Task with Subagent For each task: **Dispatch fresh subagent:** ``` Task tool (general-purpose): description: "Implement Task N: [task name]" prompt: | You are implementing Task N from [plan-file]. Read that task carefully. Your job is to: 1. Implement exactly what the task specifies 2. Write tests (following TDD if task says to) 3. Verify implementation works 4. Commit your work 5. Report back Work from: [directory] Report: What you implemented, what you tested, test results, files changed, any issues ``` **Subagent reports back** with summary of work. ### 3. Review Subagent's Work **Dispatch code-reviewer subagent:** ``` Task tool (superpowers:code-reviewer): Use template at requesting-code-review/code-reviewer.md WHAT_WAS_IMPLEMENTED: [from subagent's report] PLAN_OR_REQUIREMENTS: Task N from [plan-file] BASE_SHA: [commit before task] HEAD_SHA: [current commit] DESCRIPTION: [task summary] ``` **Code reviewer returns:** Strengths, Issues (Critical/Important/Minor), Assessment ### 4. Apply Review Feedback **If issues found:** - Fix Critical issues immediately - Fix Important issues before next task - Note Minor issues **Dispatch follow-up subagent if needed:** ``` "Fix issues from code review: [list issues]" ``` ### 5. Mark Complete, Next Task - Mark task as completed in TodoWrite - Move to next task - Repeat steps 2-5 ### 6. Final Review After all tasks complete, dispatch final code-reviewer: - Reviews entire implementation - Checks all plan requirements met - Validates overall architecture ### 7. Complete Development After final review passes: - Announce: "I'm using the finishing-a-development-branch skill to complete this work." - **REQUIRED SUB-SKILL:** Use superpowers:finishing-a-development-branch - Follow that skill to verify tests, present options, execute choice ### Example Workflow ``` You: I'm using Subagent-Driven Development to execute this plan. [Load plan, create TodoWrite] Task 1: Hook installation script [Dispatch implementation subagent] Subagent: Implemented install-hook with tests, 5/5 passing [Get git SHAs,