decompose-task

A workflow for splitting a task on a task board into smaller child tasks. It requires checking the stored parent task and the board settings before creating anything.

In plain words
What is it for?
Use it when a configured board task needs a grounded breakdown into native subtasks, with a preview before the subtasks are added.
Why use it?
It helps make large tasks easier to review while avoiding duplicate or incorrectly placed child tasks.

Skill for Claude CodeCodex

Part of the rag-reviewer plugin — 14 skills, 2 hooks shipped together

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/mimfort/rag_for_git/decompose-task
Any agent
npx skills add mimfort/rag_for_git --skill decompose-task
Clone the repo
git clone --depth 1 https://github.com/mimfort/rag_for_git

Made for: Claude Code, Codex.

Or install rag-reviewer, the plugin that ships this one along with the rest of its 14 skills, 2 hooks.

Per session 17 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 2,393 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 $0.00017 $0.02393
Opus 5 $0.00009 $0.01196
Sonnet 5 $0.00003 $0.00479
Haiku 4.5 $0.00002 $0.00239

Measured 3d ago against content hash 8d6adcc54a34, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

decompose-task 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 3d 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.

plugin/skills/decompose-task/SKILL.md · 193 lines

How it starts

The opening of the file, as written. The whole thing — 193 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Decompose Task

Overview

Reply in Russian. Build one grounded, reviewable native-subtask batch; no child may reach the board before the user sees the entire batch. Core principle: preview fidelity and durable idempotency are safety boundaries, not optional ceremony.

Lookup

The stored parent and authoritative native_subtasks capability are required gates.

  1. Inspect the repository task_board key once and resolve exactly one state:
    • If the key is present with a null, empty, or disabled value, board work is explicitly disabled: stop no-write and never call deploy-wide get_board_config.
    • Only if the repository key is absent, use the deploy fallback: call get_board_config() exactly once.
    • If a mapping exists, freeze its generic type, project, and options as board_type, project, and provider_options for the whole flow. Never re-resolve mid-flow. If no mapping resolves, stop without writes.
  2. Read the store first with get_task(parent_key, project=<project>). On miss only, run exactly one scoped sync_board(board=<project or null>, board_type=<type>, provider_options=<options>), then make one retry of the same parent read. If it is still missing, explain the failure and stop.
  3. Call get_board_targets(board_type=<type>, project=<project>, provider_options=<options>). Treat its registry capabilities as authoritative. If it errors or lacks native_subtasks, make a no-write stop. Never infer, add, or spoof a capability from provider identity or prior experience.

Context

Before drafting, attempt all three calls:

  • get_task_context(parent_key, project=<project>) for links, related work, PRs, and touched code;
  • search_tasks(query=<parent intent>, project=<project>) for similar decomposition boundaries;
  • search_codebase(repo=<active repo>, query=<parent intent>) for relevant implementation and real path:line evidence.

Empty successful results are allowed: explicitly report each empty result and continue. A tool error is not an empty result; report it and stop drafting until resolved. Do not draft first and backfill evidence later.

Read the full file on GitHub · 193 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. 3d ago First seen · 193 lines · 17 tokens per session scan A 8d6adcc54a34

Subscribe to this mod's changes

decompose-task is a skill published in the GitHub repository mimfort/rag_for_git (21 stars, last pushed 3d ago), licensed MIT. It adds 17 tokens to every session and 2,393 once invoked, about $0.0001 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 2026-08-30.

Related

Other skills, from other repositories

graft

This repo is indexed by graft/. For ANY task here, whether understanding how something works, finding where code lives, tracing what calls a symbol or what a change breaks, or scoping an edit, get your context from graft before grepping or reading source files.

trailhq/Graft · 56 tokens

roam

Codebase comprehension via roam-code CLI. Use when exploring codebases, planning modifications, debugging failures, assessing PR risk, or checking architecture health. Triggers on: understanding project structure, pre-change safety checks, finding symbols/files, blast radius analysis, affected tests, health scoring…

Cranot/roam-code · 86 tokens

code-graph

This skill should be used when understanding code structure, finding dependencies between functions/classes, tracing call graphs, or exploring code relationships. Trigger phrases include 'code graph', 'call graph', 'who calls', 'what calls', 'find dependencies', 'code structure', 'inheritance', 'find paths'.

FalkorDB/code-graph · 64 tokens

agentmap

Use for TypeScript/JavaScript codebase navigation — symbol lookup, blast radius, reuse checks, and repo orientation. Prefer agentmap before serial grep when exploring imports, exports, or where a symbol lives. Package is @raymondchins/agentmap (not the unrelated npm package agentmap).

raymondchins/agentmap · 64 tokens

m1nd-operator

Use when the user mentions m1nd or when repo investigation, search, review, docs/spec work, or risky change prep should go through m1nd first before grep, glob, or manual file reads. Covers m1nd-first routing, L1GHT and universal document ingestion, risky edit preparation, document-to-code binding, the served owner +…

maxkle1nz/m1nd · 111 tokens

m1nd-first

Use when investigating a repository, searching for implementation, reviewing changes, working from specs/docs, or preparing a risky code change in an environment where m1nd is available. This doctrine makes m1nd the first investigative layer before grep, glob, or manual file reads, except when the task is pure…

maxkle1nz/m1nd · 78 tokens