A planning agent that turns an approved research brief into short, checked instructions for separate execution agents. It starts with only the brief, not the research behind it.

Use it to split a software task into agent-sized sub-briefs while carrying along build, test, formatting, copyright, and project-rule checks.

It prevents execution agents from needing the full research history and keeps each assignment focused on the stated requirements.

How trilitech/octez-manager configures Claude Code

Reuse

Borrowing it

Nothing to install: this file belongs to trilitech/octez-manager. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file

curl -O https://raw.githubusercontent.com/trilitech/octez-manager/main/.claude/agents/planner.md

Clone the repo

git clone --depth 1 https://github.com/trilitech/octez-manager

Made for: Claude Code.

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.

README.md

[![agentmods](https://agentmods.dev/badge/agents/trilitech/octez-manager/planner/github.svg)](https://agentmods.dev/agents/trilitech/octez-manager/planner)

Your own site

<a href="https://agentmods.dev/agents/trilitech/octez-manager/planner"><img src="https://agentmods.dev/badge/agents/trilitech/octez-manager/planner/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

Your own site · 80×15

<a href="https://agentmods.dev/agents/trilitech/octez-manager/planner"><img src="https://agentmods.dev/badge/agents/trilitech/octez-manager/planner.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 35 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.

When invoked 1,567 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.

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 5.1	$0.00035	$0.01567
Opus 5	$0.00017	$0.00783
Sonnet 5	$0.00007	$0.00313
Haiku 4.5	$0.00003	$0.00157

Measured 4d ago against content hash 67444990ff04, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-09, from the pricing page.

Security

Grade A, and why

planner 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.

.claude/agents/planner.md · 175 lines

How it starts

The opening of the file, as written. The whole thing — 175 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Planner Agent

Project Context — octez-manager

Quality gates (include in all sub-briefs):

dune build — every commit must compile independently
dune runtest
dune fmt
./scripts/check-copyright.sh (auto-fix: --fix)
make completions — after CLI changes
tools/arch_query — duplication gate

Key project rules to propagate into sub-briefs:

No I/O in render path — use scheduler caches (see src/ui/AGENTS.md)
Every bug fix requires a test that fails without the fix
New files need copyright headers; run --fix after creation
TODO/FIXME must include a GitHub issue reference
Atomic commits: separate refactoring from behavior changes

You receive a validated research brief. Your job is to decompose it into sub-briefs — one per execution agent — and nothing else.

You have no research context. You did not explore the codebase. The brief is your only source of truth. If something is not in the brief, it does not exist for you. Do not speculate beyond it.

Why You Exist

Research consumes large context. Compressed research briefs are small. You operate on the small artifact so execution agents can be spawned with minimal, focused context — not a summary of a long conversation.

Input

The full content of the research brief, pasted inline into your initial prompt by the human.

The brief content is your entire starting context. Do not attempt to read files from disk to reconstruct missing information — if it is not in what you received, it does not exist for you.

Read the brief fully before doing anything else. If it is missing any of the required sections (see Research Brief Format below), do not attempt to fill the gaps. See Ambiguity Escalation below.

Output

One sub-brief per execution agent, written to:

briefs/<task>-<role>.md

Example: briefs/auth-refactor-implementer.md, briefs/auth-refactor-qa.md

Then run the human validation quiz on the full set of sub-briefs before reporting ready for execution.

Read the full file on GitHub · 175 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.

4d ago First seen · 175 lines · 35 tokens per session scan A 67444990ff04

Subscribe to this mod's changes

planner is an agent published in the GitHub repository trilitech/octez-manager (5 stars, last pushed 1mo ago), licensed MIT. It adds 35 tokens to every session and 1,567 once invoked, about $0.0002 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 2026-09-04.

Building agents Planning brief context-compression decomposition

planner

Takes a validated research brief and decomposes it into compressed, verified sub-briefs for each execution agent. Runs with a fresh context — no research history.

mathiasbourgoin/roster · 35 tokens

context-compressor

Compresses top retrieval chunks into citation-rich summary packets that preserve intent while cutting token usage.

Madhan230205/token-reducer · 22 tokens

hybrid-retriever

Runs hybrid retrieval with strict FTS-first policy, BM25 lexical ranking, vector merge, and top 3-5 reranking for token-efficient context selection.

Madhan230205/token-reducer · 38 tokens

noise-chunker

Preprocesses large corpora by removing low-signal noise and creating overlap-aware chunks for retrieval indexing.

Madhan230205/token-reducer · 25 tokens

se-ops-delegate

Delegates software engineering operations to focused subagents using compressed context packets to keep the main conversation lean.

Madhan230205/token-reducer · 26 tokens

symb-intelligence

Deep pattern analysis agent. Invoke when the human wants to understand what their session history reveals about their project, their patterns, or their working style. Goes beyond surface patterns to surface architectural insights, decision drift, and velocity signals. Apply the DuCrest Lock Methodology: find the…

SYMBEYOND/SYMB-Mem · 70 tokens

octez-manager: Agent for Claude Code