# Seed To Idea

> Create work/idea.md from a rough seed; gate missing user-owned fields. Eager-gate on sparse seeds.

- **Type:** Skill
- **Install:** `agentstack add skill-xmu-csnoob-praxiskit-seed-to-idea`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [xmu-csnoob](https://agentstack.voostack.com/s/xmu-csnoob)
- **Installs:** 0
- **Category:** [Agent Skills](https://agentstack.voostack.com/c/agent-skills)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [xmu-csnoob](https://github.com/xmu-csnoob)
- **Source:** https://github.com/xmu-csnoob/PraxisKit/tree/main/plugins/praxiskit/skills/seed-to-idea

## Install

```sh
agentstack add skill-xmu-csnoob-praxiskit-seed-to-idea
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

# Seed to Idea

Turn a rough spark into a short, useful idea brief. **Most seeds are sparse (1-2 sentences) — drift risk is highest here, so gate eagerly.**

```text
seed -> seed-to-idea -> work/idea.md -> idea-to-prd
```

Do not write a PRD, task graph, or implementation plan unless the user explicitly asks to continue.

## Contract

**Inputs:** `seed` (raw user input — text, file, or pasted prompt)
**Output:** `work/idea.md` per `schemas/idea.schema.md` (v2.0); fires `references/clarification-gate.md` for missing user-owned fields
**Preconditions:** seed (text or file path) provided
**Stop boundary:** Does NOT write a PRD, task graph, or any downstream artifact.

## Workflow

1. **Load the seed.** If the user provides a path, read it. If pasted text, use it.
2. **Load gate reference + schema.** Read `references/clarification-gate.md` and `schemas/idea.schema.md`.
3. **Triage seed length.** If seed is < ~200 chars or 1-2 sentences, default to treating most `filled-by-user` fields as gaps. Do NOT draft first.
4. **Map seed to schema fields.** For each schema field:
   - User-sourced value present? → `filled`
   - `inferable` and derivable from a `[user]` field? → `inferable` (must name source)
   - `filled-by-user` or `forbidden-to-infer` with no user source? → `gaps`
5. **Apply cross-field rules** from `idea.schema.md`. Add triggered fields to gaps.
6. **Call clarification-gate** with collected gaps. Prefer host-native multi-question UI (group up to 4 related questions); use chat or `work/clarify-seed.md` only as fallback per gate procedure.
7. **Write `work/idea.md`** only after all gaps are resolved. Annotate every field with its source.
8. **Update `work/praxiskit-context.md`** — `Pending Clarifications: none` if gate passed.
9. **Handoff.** State that `work/idea.md` is ready for `idea-to-prd` if the user wants a PRD.

## Output

Write `work/idea.md` from `templates/idea.md`. Source annotations: `[user]`, `[user via gate]`, `[user via clarify-seed]`, `[inferred from {field}]`, `[default]`.

Do not write `## Assumptions`. Uncertainty lives in `## Open Questions` or inline `[inferred]` annotations.

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [xmu-csnoob](https://github.com/xmu-csnoob)
- **Source:** [xmu-csnoob/PraxisKit](https://github.com/xmu-csnoob/PraxisKit)
- **License:** MIT

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/skill-xmu-csnoob-praxiskit-seed-to-idea
- Seller: https://agentstack.voostack.com/s/xmu-csnoob
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.