decompose

A planning command that turns written project specifications into a dependency-ordered backlog of GitHub Issues. A backlog is a prioritized list of work, and dependencies show which tasks must come first.

In plain words
What is it for?
Use it to plan work for selected parts of a repository, review the proposed tasks in a dry run, and create them only after approval.
Why use it?
It converts broad specifications into concrete tasks with acceptance checks while considering existing issues and project conventions.

Skill for Claude CodeCodex

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/blakemartz/delivery-loop/decompose
Any agent
npx skills add blakemartz/delivery-loop --skill decompose
Clone the repo
git clone --depth 1 https://github.com/blakemartz/delivery-loop

Made for: Claude Code, Codex.

Per session 56 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,103 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00056 $0.01103
Opus 5 $0.00028 $0.00551
Sonnet 5 $0.00011 $0.00221
Haiku 4.5 $0.00006 $0.00110

Measured 2d ago against content hash 9275d6c3fa75, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

decompose scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

plugins/delivery-loop/skills/decompose/SKILL.md · 49 lines

How it starts

The opening of the file, as written. The whole thing — 49 lines — stays where its author put it; the contents beside it link to each section on GitHub.

decompose — specs → backlog

You are the decomposer role from the delivery spec (${CLAUDE_PLUGIN_ROOT}/docs/agentic_delivery_spec.md, §3.2, §7 method).

Config — read .claude/delivery.conf at the repo root first (defaults in parens): SPEC_SOURCES (docs/*_spec.md) — the files you decompose; STYLEGUIDES_DIR (styleguides/) — conventions; MODULES — the allowed module: set; GATE_CMD (bash scripts/check.sh) — the acceptance-criteria gate.

Inputs

  1. The spec sources — every file matched by SPEC_SOURCES. These are the primary source of tasks and their acceptance criteria. If a spec ends with a "decompose-ready cutline" (a dependency-ordered task list with module / size / command-checkable seams), translate it — don't re-derive it.
  2. The conventions in STYLEGUIDES_DIR — read the ones relevant to what you're decomposing; tasks must fit them. If the directory is missing or holds no ratified guides, say so and suggest /delivery-loop:author-styleguides — then proceed against the stack's idiomatic best practice (spec §2.1).
  3. The existing backlog: gh issue list --label task --state all --limit 500 --json number,title,labels — never propose a duplicate of an open or closed task.
  4. Any product/context docs the specs reference — consult for a task's Context section (the why); never source acceptance criteria from prose (Hard rule 1).
  5. Optional scope argument narrowing what to decompose (default: the next unstarted slice of work in the specs).

Method

  • Epics ≈ major spec sections or modules. Tasks are children of exactly one epic.
  • Dependency edges follow the natural build order: scaffold → models → repositories → services → endpoints → wiring/integration (adapt to the stack).
  • Every task gets: title, exactly one module, size, command-shaped acceptance criteria, and its Depends-on: edges.
  • A task that introduces a tool, test suite, or build step must include extending the gate (GATE_CMD / scripts/check.sh) to cover it as one of its acceptance criteria. This is how the gate self-maintains — it grows with the repo instead of going stale.

Read the full file on GitHub · 49 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 49 lines · 56 tokens per session scan A 9275d6c3fa75

Subscribe to this mod's changes

decompose is a skill published in the GitHub repository blakemartz/delivery-loop (4 stars, last pushed 1mo ago), licensed MIT. It adds 56 tokens to every session and 1,103 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other skills, from other repositories

create-issue

Draft and submit a GitHub issue from a user idea or bug description, with bilingual body and correct labels.

QwenLM/qwen-code · 26 tokens

github-issue-triage

Issue triage and lifecycle management agent for ZeroClaw. Use this skill whenever the user wants to: triage open issues, close stale/duplicate/fixed issues, apply labels, run a backlog sweep, enforce the current issue stale policy, or handle a specific issue. Trigger on: 'triage issues', 'issue triage', 'sweep…

zeroclaw-labs/zeroclaw · 140 tokens

coordinate

Coordinate a small team of Qwen Code teammates with enforced read-only workers, an optional worktree-pinned writer, shared tasks, peer messages, and existing Agent View tabs. Invoke explicitly with /coordinate.

QwenLM/qwen-code · 43 tokens

trello

Manages Trello boards, lists, and cards via the Trello REST API. Use when the user wants to create cards, move tasks between lists, list boards, add comments, archive cards, or check what is on a Trello board. Handles authentication, pagination, and rate-limit awareness for all Trello REST endpoints.

elizaOS/eliza · 69 tokens

team-workflow

Design the team's operating rhythm — task management, collaboration rituals, and tooling. Use when the day-to-day cadence needs structure. For a time-boxed sprint, use design-sprint-plan.

Owl-Listener/designer-skills · 43 tokens

attio-interaction

Generic Attio CRM REST API v2 recipes for querying records, upserting companies/people/deals, writing notes/tasks/comments, managing lists, and handling webhooks.

desplega-ai/agent-swarm · 39 tokens