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.
npx agentmods add commands/joseairosa/recall/decomposegit clone --depth 1 https://github.com/joseairosa/recallWrote 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.
[](https://agentmods.dev/commands/joseairosa/recall/decompose)<a href="https://agentmods.dev/commands/joseairosa/recall/decompose"><img src="https://agentmods.dev/badge/commands/joseairosa/recall/decompose.svg" alt="Measured on agentmods" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00016 | $0.00716 |
| Opus 5 | $0.00008 | $0.00358 |
| Sonnet 5 | $0.00003 | $0.00143 |
| Haiku 4.5 | $0.00002 | $0.00072 |
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 4d 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.
How it starts
The opening of the file, as written. The whole thing — 115 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/decompose Command
Break down an RLM execution chain into subtasks and optionally process them automatically.
Usage
/decompose <chain_id> [strategy] [--auto]
Arguments
| Argument | Required | Description |
|---|---|---|
| chain_id | Yes | The execution chain ID from /load-context |
| strategy | No | Override decomposition strategy |
| --auto | No | Automatically process all subtasks |
Strategies
| Strategy | Description | Best For |
|---|---|---|
| filter | Extract by regex patterns | Log analysis, error finding |
| chunk | Sequential fixed-size pieces | Document reading |
| recursive | Nested decomposition | Complex analysis |
| aggregate | Combine multiple sources | Synthesis tasks |
Examples
# Use recommended strategy
/decompose 01HXYZ12345
# Override with filter strategy
/decompose 01HXYZ12345 filter
# Auto-process all subtasks
/decompose 01HXYZ12345 --auto
# Filter strategy with auto-process
/decompose 01HXYZ12345 filter --auto
Behavior
Manual Mode (default)
- Decompose task into subtasks
- Display subtask list with descriptions
- Wait for user to process each subtask manually
Auto Mode (--auto)
- Decompose task into subtasks
- For each subtask:
- Extract relevant context snippet
- Analyze the snippet
- Record the result
- Report progress after each subtask
- Suggest merging when complete
Output
Task Decomposed
---------------
Chain ID: 01HXYZ12345
Strategy: filter
Subtasks: 5
1. [pending] Find ERROR level messages (query: ERROR|FATAL)
2. [pending] Find WARNING level messages (query: WARN|WARNING)
3. [pending] Find exception stack traces (query: Exception|Traceback)
4. [pending] Find failure indicators (query: failed|failure|crash)
5. [pending] Summarize error patterns (query: ERROR)
Next steps:
- Process each subtask with inject_context_snippet
- Or run /decompose 01HXYZ12345 --auto to auto-process
Auto Mode Progress
Processing subtask 1/5: Find ERROR level messages
- Extracted 1,234 tokens (92% relevance)
- Found 47 error messages
- Result recorded
Processing subtask 2/5: Find WARNING level messages
...
All subtasks complete!
Run /rlm-status 01HXYZ12345 to see summary
Or call merge_results to aggregate findings
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.
- 4d ago First seen · 115 lines · 16 tokens per session scan A ac70a8fdb90a
decompose is a command published in the GitHub repository joseairosa/recall (175 stars, last pushed 4d ago), licensed MIT. It adds 16 tokens to every session and 716 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.
Other commands, from other repositories
smartcompact
Pin important turns before compacting — pins survive compaction via SessionStart hook reinjection.
seance
Summon a fresh Claude chat that channels this session — summarize the task and open a new chat pre-loaded with it.
audit
Audit installed agents/skills - find never-used or stale ones and help prune them.
checkpoint
Summarize completed work to a durable file, then hand the user a steered /compact line.
window
Set the context window Astral measures against (200K / 1M / custom / auto).
help
What Astral does and how to use it.