idea

idea is a command for Claude Code from ihudak/ihudak-claude-plugins. It costs 266 tokens per session (11,449 once invoked), scanned A, original, MIT.

An idea-refinement command that turns one source—such as a prompt, Markdown file, community post, Jira ticket, or existing product requirements document—into a short idea.md file. A product requirements document (PRD) describes what a product change should achieve.

In plain words
What is it for?
Use it to question and refine an idea, optionally checking documentation, earlier work, or selected code repositories for context.
Why use it?
It turns rough or scattered input into a checked, focused starting point for creating a PRD. It does not change code or write to Jira.

Command for Claude Code

Written for Claude Code: allowed-tools in frontmatter. Also seen: mentions subagents.

Installs and runs on its own, but its text points at files inside its plugin — anything it tells you to read at a ${CLAUDE_PLUGIN_ROOT} path is only there once the plugin is installed. Installing the plugin gets both.

Part of the dev-workflows plugin — 2 skills, 28 commands, 38 agents, 3 hooks shipped together

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/ihudak/ihudak-claude-plugins/idea
Clone the repo
git clone --depth 1 https://github.com/ihudak/ihudak-claude-plugins

Made for: Claude Code.

Or install dev-workflows, the plugin that ships this one along with the rest of its 2 skills, 28 commands, 38 agents, 3 hooks.

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.

agentmods badge for idea

README.md
[![agentmods](https://agentmods.dev/badge/commands/ihudak/ihudak-claude-plugins/idea.svg)](https://agentmods.dev/commands/ihudak/ihudak-claude-plugins/idea)
Your own site
<a href="https://agentmods.dev/commands/ihudak/ihudak-claude-plugins/idea"><img src="https://agentmods.dev/badge/commands/ihudak/ihudak-claude-plugins/idea.svg" alt="Measured on agentmods" height="20"></a>
Per session 266 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 11,449 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 5.1 $0.00266 $0.11449
Opus 5 $0.00133 $0.05725
Sonnet 5 $0.00053 $0.02290
Haiku 4.5 $0.00027 $0.01145

Measured today against content hash a03f934f2510, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-05, from the pricing page.

Security

Grade A, and why

idea 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 today.

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/dev-workflows/commands/idea.md · 523 lines

How it starts

The opening of the file, as written. The whole thing — 523 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Refine an idea into idea.md: $ARGUMENTS

/idea is the front door of the PRD-creation flow (PM phase) — upstream of /create-prd and the existing pipeline. It ingests one source, refines it through a grill, and writes a lean one-page idea.md (per ${CLAUDE_PLUGIN_ROOT}/references/idea-format.md) that seeds the Product Requirements Document. It is not a PRD: no code change. Output lands in the PRD folder the key names, on the first write and never relocated. It then vendors what it read into that same folder (Phase 4.5) so the record the specs repo keeps is one whose links resolve for everybody, not only for the operator whose disk the sources came off.

Flags: --deep switches the grill from bounded (≤10 questions) to relentless (until convergence). --no-docs turns off documentation grounding (see Phase 1). --ground-code [<repo>[,<repo>…]] grounds the idea against mounted code (see Phase 2.6) — bare it derives the repo set, with a value it scans exactly those repos. The token after --ground-code is its value only when it contains no whitespace and every comma-separated part matches a top-level directory basename under ${REPOS_PATH:-/workspace}; otherwise the flag is bare and the token is idea text.


Phase 0 — Resolve the address + model routing

  1. The address (mandatory). Parse the first non-flag token and validate it with key-valid (${CLAUDE_PLUGIN_ROOT}/references/addressing.md §1). Absent or malformed → stop: IDEA_NEEDS_KEY: /idea needs a PRD key (^[A-Z][A-Z0-9_]*(-\d+)+$, e.g. ACME-77) — it names the folder this idea will live in. Re-run '/dev-workflows:idea <PRD-KEY> [<prompt>|@<file>]'.

    The key is an argument because there is nowhere keyless to write. idea.md lands in its final folder on the first write — PRD-<KEY>-<slug>/ under $SPECS_PATH/specifications/, resolved with resolve-address and created there when absent (addressing.md §2, §3). It is never relocated afterwards, and /create-prd <KEY> finds it there.

    Validated for shape and checked against nothing, exactly as /brd-intake <BRD-KEY> already asks. Nothing looks a key up, because there is nothing to look it up in.

    Accepted cost: an abandoned idea leaves a folder in specifications/. Reintroducing a staging area to avoid that would restore the relocation step this removes.

  2. Resolve model routing. Invoke the model-routing skill (Skill tool, skill: "dev-workflows:model-routing"), then record:

    model_routing:
      classification: MODERATE          # idea refinement is typically MODERATE
      reason: <one-line>
      current_model: <the model this orchestrator/grill is running under>
      detection_model: <§2.1 Sonnet chain: claude-sonnet-5, fallback claude-sonnet-4-6/4-5>   # idea-reader
      authoring_model: <= current_model>   # the interactive grill + idea.md authoring (session model, not a delegated subagent)
      opus_available: <true if a §2 Opus model resolved, else false>
      notes: <any §2/§2.1 fallback or degradation>
    

    The grill + authoring run inline on current_model (the §2 Opus chain — interactive judgment, not a delegated subagent). idea-reader runs on detection_model. If no Opus resolves, degrade to the best available and record the degradation in notes and the final report — do NOT hard-block (a PM must not be blocked from capturing an idea by a momentary Opus outage). A --ground-code run does not floor the classification at SIGNIFICANT: §1.1's multi-source floor is written for /implement, and §8.3's purpose — the strongest available model on synthesis — is already met here, because the grill and authoring run inline on current_model while the scanners run on detection_model.

Specs-repo preflight. Cite ${CLAUDE_PLUGIN_ROOT}/references/specs-repo-git.md and execute its specs-preflight entry point (§3) inline: flush any leftover session artifacts from an earlier run, retry an artifact commit that failed to push, and settle the branch. Prompt-free and silent when the specs repo is clean and on its default branch. If a guard fires, emit its §5 notice; if it returns specs_git: blocked (§3.3 G0), carry that flag for the whole run — the terminal commit-artifacts step skips on it.

Read the full file on GitHub · 523 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. today Changed · +170 lines · +73 tokens per session a03f934f2510
  2. 5d ago First seen · 353 lines · 193 tokens per session scan A 0869ea33bfe0

Subscribe to this mod's changes

idea is a command published in the GitHub repository ihudak/ihudak-claude-plugins (2 stars, last pushed 3d ago), licensed MIT. It adds 266 tokens to every session and 11,449 once invoked, about $0.0013 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.