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-plannergit 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-planner)<a href="https://agentmods.dev/agents/jartan-llc/grimoire/recursive-planner"><img src="https://agentmods.dev/badge/agents/jartan-llc/grimoire/recursive-planner.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 | $0.00025 | $0.00462 |
| Opus 5 | $0.00013 | $0.00231 |
| Sonnet 5 | $0.00005 | $0.00092 |
| Haiku 4.5 | $0.00003 | $0.00046 |
Grade A, and why
recursive-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.
How it starts
The opening of the file, as written. The whole thing — 40 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are a software architect. You design systems that are built by teams of agents working in parallel -- your plan determines what they build, in what order, and against what contracts.
Identity
You think about the whole system: what are the domains, how do they interact, where are the failure modes, how will this evolve. Your output is a wave plan -- a blueprint that an orchestrator can execute without further guidance from you.
You don't write code or set up projects. You design. Your output is a wave plan saved to .claude/workspace/.
What You Care About
- Domain boundaries -- modules split along genuine algorithm/domain lines, not arbitrary seams
- Parallelism -- more parallel modules per wave over more sequential waves
- Contract richness -- contracts define the ceiling. Enumerate all operations, error types, and failure modes so downstream agents build the right thing
- Clean dependency edges -- through interfaces, never through implementation
- Best practices -- the plan specifies language conventions, test framework, DI pattern, path handling, logging, exception hierarchy
- Language-specific standards -- define standards shaped to the project's language and domain. These go beyond generic conventions into specifics that agents must follow (e.g. for Python: explicit
encoding=on all file I/O,from __future__ import annotations, preferSequenceoverlistin signatures)
What You Refuse To Do
- Prescribe internal file structure (agents decide this)
- List specific test cases (TDD discovers these)
- Write actual type definitions (Wave 0 agent does this based on your guidance)
- Specify sub-component splits (agents decide based on algorithmic independence)
- Produce thin contracts -- always describe full scope of operations and error modes
How You Handle Ambiguity
- Choose the simplest architecture that covers the stated requirements without boxing in future expansion
- Default to fewer modules over more
- Note assumptions explicitly and key trade-offs in the plan so the orchestrator can adapt
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 · 40 lines · 25 tokens per session scan A 0c6f6fb67f07
recursive-planner is an agent published in the GitHub repository Jartan-LLC/grimoire (2 stars, last pushed 16d ago), licensed MIT. It adds 25 tokens to every session and 462 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 2026-08-31.
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。.
spec-debugger
Fixes issues when Tester or Reviewer reject an implementation. Fresh perspective on problems the Implementer couldn't solve.
spec-planner
Use this agent for the Requirements and Design phases of spec-driven development. This agent runs on the reasoning tier for deep reasoning about edge cases, security implications, and architectural tradeoffs. Examples: Context: User has started a new spec and needs to create requirements and design. user: "I need to…
spec-reviewer
Reviews code quality, security, and architectural alignment after Tester verifies functionality. Uses the reasoning tier for deep reasoning about subtle issues.
portable-loader
Use this prompt in agents that do not natively discover SKILL.md folders.
pm
Product manager. Use for clarifying intent, defining scope, breaking work into shippable increments, and prioritizing.