Subagent-Driven Development
Execute implementation plans with a fresh implementer subagent per task, a spec+quality review after each, and a whole-branch review at the end
- What
- Execute implementation plans with a fresh implementer subagent per task, a spec+quality review after each, and a whole-branch review at the end
- Cost
- Free
- Needs
- an agent harness that can dispatch subagents (Claude Code style); a git repository; an implementation plan to execute. Companion superpowers skills for the full loop: using-git-worktrees, requesting-code-review, finishing-a-development-branch.
- Install
- Copy the installer prompt below into your Muse — your agent does the rest.
Curated by Skill Harbor — the flagship execution skill of the superpowers suite: run an implementation plan with a fresh subagent per task, a task review (spec compliance AND code quality — both verdicts required) after each one, and one broad whole-branch review at the end. It tracks progress in a git-ignored ledger file (not memory — compaction kills it), pre-flights the plan for task conflicts before dispatching Task 1, picks the cheapest model that fits each role, caps the fix loop at five rounds per task (resume the implementer for rounds 1–3, fresh higher-tier eyes for 4–5, adjudicate at the cap), and keeps going with recorded rulings instead of stalling on human questions — the only stops are irreversible/destructive operations, security-sensitive actions, side effects outside the worktree, or a plan so broken every path is a guess. By @obra, listed here with credit to its creator. Honest caveats: it depends on a harness that can dispatch and message subagents (Claude Code style), on the companion scripts (`sdd-workspace`, `task-brief`, `review-package`) in the repo, and on sibling skills (using-git-worktrees, requesting-code-review, finishing-a-development-branch) for the full loop; it's demanding by design — "continuous execution" means the human is not consulted mid-plan. Skill Harbor never reviews the code, review it yourself before use.
Version:
Quick answer
Subagent-Driven Development is a free Muse skill for executing implementation plans with a fresh implementer subagent per task. After each task comes a spec and quality review, and at the end a whole-branch review. That structure keeps large implementations organized, reviewed at every step and coherent when the branch finally lands.
Install
Copy the install package below, then paste it into Muse🧪 Install ran in our test sandbox · Nov 11, 2026📌 Pinned version: commit bb92a77Step 0 — before installing anything: read this install prompt, summarize in a few lines what this build does and what installing it involves, then ask me to confirm before you proceed. Prerequisites: an agent harness that can dispatch subagents (Claude Code style); a git repository; an implementation plan to execute. Companion superpowers skills for the full loop: using-git-worktrees, requesting-code-review, finishing-a-development-branch. Install "Subagent-Driven Development" for me. It executes implementation plans with a fresh implementer subagent per task, a spec+quality review after each task, a 5-round fix loop, and a final whole-branch review. Repository: https://raw.githubusercontent.com/obra/superpowers/bb92a77741419a4ab5f06e711a283343f1ada0c3/skills/subagent-driven-development/SKILL.md Pinned version: install the code at commit bb92a77741419a4ab5f06e711a283343f1ada0c3 exactly (Skill Harbor tested this version). Do not substitute a newer version. If this commit is unreachable, stop and tell me. 1. Fetch the SKILL.md file (and any helper files) from the repository path into a temporary folder 2. Safety check: review the SKILL.md and scripts for anything suspicious (unexpected network calls, shell commands, credential harvesting). This repo should contain zero secrets in code, credentials only via the secure vault. Verify that holds here; STOP on any red flag and tell me. 3. Install it as a skill: copy SKILL.md and its helper files into the agent's skills directory, in a folder named "obra-subagent-driven-development". 4. Verify with no network calls: frontmatter valid, files in place. 5. Report what was installed, where, and what I still need to do myself. GitHub is optional: if I have a GitHub account or the gh CLI, you may use it; otherwise public access is fine. Never require it unless it's in the prerequisites above. Rules: don't touch anything outside the temp folder and the install target. If anything looks off, stop and ask me.
Saved to your recent installs. Find it anytime on /connect.
🔒 Security, provenance and what this touches
- It can run commands on your Muse’s computer.
What the scan found — excerpts from the install prompt
- runs commands — Prérequis : un harness capable de lancer des sous-agents (style Claude Code) ; un dépôt git ; un plan d'implém
The install prompt below already includes the vetting steps: your agent follows the community checklist before installing anything with executable code. Want more?
Verification certificate
Dated facts about what was checked for this listing, and how. This is not a safety certification.
- Pinned version
- bb92a77 · 2026-10-13
- Sandbox install test
- Install ran in our test sandbox · Nov 11, 2026
- Creator identity
- Not verified
- Fingerprint watch
- Last checked · Nov 19, 2026 : no change detected in the files linked by this install prompt
- Install prompt analysis
- 2026-10-13 : runs commands
Skill Harbor never reviews the code of a listed skill. These are dated facts about the install prompt and our install test, not a safety certification. Review a skill yourself before you install it. How we verify
Questions
How do I install a skill?
Every product page includes a copy-paste install prompt. Paste it into your Muse and it sets the skill up for you — no manual configuration.
Where does my money go?
Straight to the seller. Skill Harbor never processes payments: checkout happens on the seller’s own page, usually Stripe.
What does the ✓ next to a creator’s name mean?
It means we confirmed the identity of the person behind the listing. It says nothing about the code itself — always check a skill before installing it.