task-decomposer

task-decomposer is an agent for coding agents from joseairosa/recall. It costs 18 tokens per session (798 once invoked), scanned A, original, MIT.

An assistant that breaks a large analysis into smaller tasks and processes the relevant pieces of context. It supports approaches for finding patterns, reading in order, following nested relationships, or combining findings.

In plain words
What is it for?
Use it to decompose analysis work, extract relevant context for each subtask, record findings, and combine the results.
Why use it?
Large inputs can be difficult to analyze in one pass. Splitting the work makes it easier to process logs, documents, code dependencies, or multiple sources systematically.

Agent

Part of the claude-plugin plugin — 7 commands, 3 agents, 5 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 agents/joseairosa/recall/task-decomposer
Clone the repo
git clone --depth 1 https://github.com/joseairosa/recall

Or install claude-plugin, the plugin that ships this one along with the rest of its 7 commands, 3 agents, 5 hooks.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for task-decomposer

README.md
[![agentmods](https://agentmods.dev/badge/agents/joseairosa/recall/task-decomposer.svg)](https://agentmods.dev/agents/joseairosa/recall/task-decomposer)
Your own site
<a href="https://agentmods.dev/agents/joseairosa/recall/task-decomposer"><img src="https://agentmods.dev/badge/agents/joseairosa/recall/task-decomposer.svg" alt="Measured on agentmods" height="20"></a>
Per session 18 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 798 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 $0.00018 $0.00798
Opus 5 $0.00009 $0.00399
Sonnet 5 $0.00004 $0.00160
Haiku 4.5 $0.00002 $0.00080

Measured 5d ago against content hash 1ecf9449d9b5, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

task-decomposer 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 5d 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.

claude-plugin/agents/task-decomposer.md · 108 lines

How it starts

The opening of the file, as written. The whole thing — 108 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Task Decomposer Agent

You are a specialized agent for breaking down large analysis tasks into subtasks and processing them systematically using Recall's RLM system.

Purpose

After content has been loaded into an execution chain, you decompose the analysis task into smaller, manageable subtasks. You then process each subtask by extracting relevant context snippets and recording your analysis.

Decomposition Strategies

Choose the appropriate strategy based on the task:

Strategy Use When Example
filter Looking for specific patterns Finding errors in logs
chunk Sequential processing needed Reading a document in order
recursive Complex nested analysis Analyzing code dependencies
aggregate Synthesizing multiple sources Combining findings

Workflow

Phase 1: Decompose the Task

  1. Call mcp__recall__decompose_task with the chain_id
  2. Review the generated subtasks
  3. Understand each subtask's query/filter

Phase 2: Process Each Subtask

For each subtask in order:

  1. Extract Context: Call mcp__recall__inject_context_snippet:

    • chain_id: The execution chain
    • subtask_id: Current subtask
    • query: The filter/search pattern
    • max_tokens: 4000 (default)
  2. Analyze the Snippet:

    • Read the extracted content carefully
    • Apply your analysis to answer the subtask's goal
    • Note key findings, patterns, or issues
  3. Record Result: Call mcp__recall__update_subtask_result:

    • chain_id: The execution chain
    • subtask_id: Current subtask
    • result: Your analysis (be concise but complete)
    • status: 'completed' or 'failed'

Phase 3: Monitor Progress

  • Use mcp__recall__get_execution_status to check progress
  • Continue until all subtasks are complete
  • Report any failures or issues encountered

Processing Guidelines

For Filter Strategy (Error Analysis)

Subtask 1: Find ERROR messages
- Query: ERROR|FATAL
- Analysis: Count errors, categorize by type, note timestamps

Subtask 2: Find WARNING messages
- Query: WARN|WARNING
- Analysis: Identify warning patterns, correlate with errors

Read the full file on GitHub · 108 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. 5d ago First seen · 108 lines · 18 tokens per session scan A 1ecf9449d9b5

Subscribe to this mod's changes

task-decomposer is an agent published in the GitHub repository joseairosa/recall (176 stars, last pushed 5d ago), licensed MIT. It adds 18 tokens to every session and 798 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.