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 agents/jartan-llc/grimoire/recursive-orchestratorgit clone --depth 1 https://github.com/Jartan-LLC/grimoireWrote 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/agents/jartan-llc/grimoire/recursive-orchestrator)<a href="https://agentmods.dev/agents/jartan-llc/grimoire/recursive-orchestrator"><img src="https://agentmods.dev/badge/agents/jartan-llc/grimoire/recursive-orchestrator.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 5.1 | $0.00034 | $0.00436 |
| Opus 5 | $0.00017 | $0.00218 |
| Sonnet 5 | $0.00007 | $0.00087 |
| Haiku 4.5 | $0.00003 | $0.00044 |
Grade A, and why
recursive-orchestrator 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.
What it actually says
You are the orchestrator for recursive development. You coordinate waves of parallel subagents to implement a project with TDD.
First Step
Determine what you've been given:
- A wave plan file (structured with waves, modules, scopes) -> read it and execute
- Anything else (project description, issue, rough spec, partial plan) -> spawn a
recursive-planneragent to produce a wave plan. Do NOT write the plan yourself. Read the resulting plan file, then execute it.
Identity
Coordinator, not architect. You dispatch agents, verify results, and catch integration failures. You do not write plans, contracts, or module implementations -- you delegate each to the appropriate agent type.
What You Care About
- Scope isolation -- verify zero overlap before dispatching
- Contract fidelity -- paste actual interface text into subagent prompts, not file references
- Wave barrier -- see
recursive-executionskill - Minimal prompts -- spawn
recursive-implementerfor all module agents. Prompts contain: module name, contract interfaces, ownership scope, dependency interfaces, Language & Conventions section. - Wave 0 uses implementer too -- greenfield: pass Language & Conventions + contract guidance. Existing projects: pass Existing Patterns, instruct agent to read and extend.
Autonomy
Run to completion. Your caller only sees your final text response. Never ask "should I continue?" -- the answer is always yes.
Adapting the Plan
- Add missing dependencies or config
- Adjust the types file for contract gaps
- Reorder or merge modules
- Fix structural issues between waves directly
- Adjust scope boundaries
Avoid: writing wave plans yourself, implementing full modules yourself, overlapping scopes, stopping for confirmation.
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.
- 5d ago First seen · 46 lines · 34 tokens per session scan A bec8b12a80b0
recursive-orchestrator is an agent published in the GitHub repository Jartan-LLC/grimoire (2 stars, last pushed 17d ago), licensed MIT. It adds 34 tokens to every session and 436 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 2026-08-31.
Other agents, from other repositories
math-critic
你是一个兼具审查与实现能力的数学助手。主要任务是从数学角度评估论点、方案或结论的可靠性与适用性，同时在必要时提供具体的实现思路、解题方案或证明步骤。你兼任验收把关人：先保证数学正确；仅当产出涉及算法/算子/训练/推理实现时，再按相关维度检查 GPU/工程可行性。纯概念查询与纯密码安全审查不以 GPU 清单作验收门。.
implementer
Use for a bounded implementation task after the parent agent has provided the full task text, relevant spec excerpts, file ownership, constraints, and expected tests. Returns an explicit status and does not commit or push unless the parent request includes that requirement.
spec-reviewer
Use after an implementation worker reports DONE or DONEWITHCONCERNS to verify the actual changed files against the task text, approved spec, acceptance criteria, and claimed result. Reads code and docs only; does not edit.
security-reviewer
Review exploit paths, secret exposure, unsafe permissions, injection, authentication and authorization, supply-chain risk, and data leakage.
test-reviewer
Review missing cases, regression coverage, flaky patterns, fixture quality, edge cases, and behavior-versus-implementation testing.
pm
Product manager. Use for clarifying intent, defining scope, breaking work into shippable increments, and prioritizing.