A command that assembles and runs a sequence of specialized reasoning roles for complicated or complex tasks.

Use it when the task requires coordinated investigation, architecture decisions, implementation thinking, and review.

It helps break difficult work into stages such as research, analysis, design, and review instead of handling every question in one step.

compose — Command by kouroshez

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 commands/kouroshez/coding-os/compose

Clone the repo

git clone --depth 1 https://github.com/kouroshez/coding-os

Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.

When invoked 486 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.

Model	Per session	Once invoked
Fable 5	$0.00000	$0.00486
Opus 5	$0.00000	$0.00243
Sonnet 5	$0.00000	$0.00097
Haiku 4.5	$0.00000	$0.00049

Measured yesterday against content hash 1919c081a8df, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

compose 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 yesterday.

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.

src/core/commands/compose.md · 24 lines

What it actually says

Compose the 11-semantic-role chain for the current COMPLICATED+ task and dispatch.

Use when the thinking_os gate is COMPLICATED or COMPLEX (Q1) — the chain orchestrates researcher → analyst → architect → … → reviewer/observer per the formula composer at src/core/thinking_os/formula_composer.py. For CLEAR tasks skip; the composer is intentionally heavyweight.

Steps:

Read the active task id ($COS_AGENT_DIR/.task-current) and current gate ($COS_AGENT_DIR/.thinking_os-gate). If $ARGUMENTS is provided as TASK-NNN, use it directly.
Invoke cos_compose_chain(task_id="<TASK-ID>") (the MCP tool registered in src/core/thinking_os/tools/cognition.py). The envelope returns:
- data.chain: ordered RoleActivation list (role · intensity · formula).
- data.signals: the TaskSignals the composer extracted (complexity · domain · risk markers).
State the chain inline so the user sees the orchestration: 🧠 chain: researcher → analyst → … → reviewer.
Execute the chain — either by dispatching each role via cos_dispatch_formula_run (Claude SDK extra installed) or by walking the formulas inline (no SDK extra). Each role's EvidenceBundle is recorded via cos_supervise_record_output.
Update the task's Work Log with chain composed: <role list>.

Acceptance gates:

Skipping cos_compose_chain on a COMPLICATED+ task is a known dead-path: a prompt-based nudge that agents read but don't act on. /compose is the deterministic alternative: one keystroke, one tool call, one chain.
The chain output is persistent — formula_dispatches and persona_selections tables grow by ≥1 row per /compose invocation.

Output format:

🧠 chain: <r1> → <r2> → … → <rN>     (complexity=<Q1>, dimensions=<Q2>)
   signals: <domain>, <risk-markers>
   formulas: <preset-id> · <preset-id> …

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.

yesterday First seen · 24 lines · 0 tokens per session scan A 1919c081a8df

Subscribe to this mod's changes

compose is a command published in the GitHub repository kouroshez/coding-os (6 stars, last pushed yesterday), licensed Apache-2.0. It costs nothing until one of its globs matches a file; then it loads 486 tokens. 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.

Code review Planning Research ai-agents claude-code codex developer-tools knowledge-graph mcp

compose

Getting it into your agent

What it costs to keep this loaded

Grade A, and why

Nothing flagged

What it actually says

What this file has done since we first saw it

Other commands, from other repositories

release-notes

create-issue

export

wh:graph-review

wh:asta-lit

soc2-review