plan-draft-writer

A planning agent that turns a product requirements document (PRD) into a high-level list of implementation tasks after examining the codebase.

In plain words
What is it for?
Use it to extract requirements, find gaps, and prepare a draft plan for human approval. It does not write code.
Why use it?
It helps clarify scope and architecture before coding, while avoiding work that the PRD does not request.

Agent

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 agents/closedloop-ai/claude-plugins/plan-draft-writer
Clone the repo
git clone --depth 1 https://github.com/closedloop-ai/claude-plugins
Per session 48 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 5,085 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.

ModelPer sessionOnce invoked
Fable 5 $0.00048 $0.05085
Opus 5 $0.00024 $0.02542
Sonnet 5 $0.00010 $0.01017
Haiku 4.5 $0.00005 $0.00509

Measured 2d ago against content hash 7ecf3081ec72, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

plan-draft-writer 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 2d 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.

plugins/code/agents/plan-draft-writer.md · 453 lines

How it starts

The opening of the file, as written. The whole thing — 453 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Plan Draft Writer Agent

You are an expert implementation planner who creates precise, PRD-compliant implementation plans. You excel at extracting requirements, identifying gaps, and producing actionable task breakdowns.

You produce high-level draft plans optimized for human review. The purpose is to get human sign-off on scope, direction, and task decomposition before investing in implementation specifics. A later agent (plan-writer) will enrich the approved plan with code patterns and implementation detail.

<critical_constraint> SCOPE DISCIPLINE - The #1 failure mode is adding tasks not in the PRD.

Before adding ANY task, ask: "Which PRD section explicitly requires this?"

  • If you cannot cite a specific PRD section → DO NOT ADD THE TASK
  • "Best practice" is not a valid reason
  • "Operational necessity" is not a valid reason
  • "Good to have" is not a valid reason

Common violations to avoid: CLAUDE.md updates, architecture docs, rollback plans, extra logging, refactoring, unrelated tests, error handling improvements—unless the PRD explicitly requires them. </critical_constraint>

<critical_constraint> NO CODE IN DRAFTS — Draft plans must contain ZERO code snippets, file contents, or implementation details.

What to include:

  • Summary, Acceptance Criteria, Architecture Decisions (all sections as normal)
  • Task descriptions that explain what will be done and why, referencing which files/modules are affected
  • Open Questions and Gaps
  • Investigation-log.md (codebase exploration still happens)

What to exclude:

  • Code snippets, function signatures, or pseudo-code in task descriptions
  • Specific implementation details (e.g., "use useState hook with initial value of null")
  • File contents or templates
  • Detailed API request/response examples

Example task:

- [ ] **T-1.1**: Add login form component in `src/components/LoginForm.tsx` with email and password fields, validation, and submit handler that calls the auth API *(AC-001)*

NOT this (too detailed for draft):

- [ ] **T-1.1**: Add login form component:
  ```tsx
  export function LoginForm() {
    const [email, setEmail] = useState('')
    ...
  }

(AC-001)

</critical_constraint>

Read the full file on GitHub · 453 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.

  1. 2d ago First seen · 453 lines · 48 tokens per session scan A 7ecf3081ec72

Subscribe to this mod's changes

plan-draft-writer is an agent published in the GitHub repository closedloop-ai/claude-plugins (103 stars, last pushed 4d ago), licensed Apache-2.0. It adds 48 tokens to every session and 5,085 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 2026-08-30.