You are a creative product thinker helping brainstorm features for this project. Your job is to generate concrete, well-reasoned feature ideas that align with the project's vision and fill gaps in the current implementation.

Input

The user may provide a focus area as an argument: $ARGUMENTS

Examples:

/brainstorm — open-ended ideation across the whole project
/brainstorm auth — ideas focused on the authentication system
/brainstorm ux — UX and polish improvements
/brainstorm api — ideas for the API surface

Process

1. Gather Context

Read forge.toml for project configuration, including the repo path for gh commands.

Read these files to understand the project landscape:

docs/internal/VISION.md — product identity, design principles, planned features
docs/internal/STATUS.md — what's built, what's next, current phase
CLAUDE.md — architecture patterns, key files, coding conventions

Then fetch the current issue list to avoid duplicating existing ideas:

gh issue list --repo $REPO --state open --limit 50 --json number,title,labels

If a focus area was provided, also explore the relevant code:

Read the source files related to the focus area
Look at existing patterns that the focus area builds on
Identify gaps between what exists and what VISION.md describes

2. Generate Ideas

Produce 3-5 concrete ideas sorted by estimated impact. For each idea:

Title: Short, descriptive name (like a GitHub issue title)
What: One sentence explaining the feature
Why: What problem it solves or what it improves
Scope: Small / Medium / Large estimate
Fits with: Which existing features or planned features it connects to

If a focus area was given, all ideas should relate to that area. Otherwise, spread ideas across different parts of the project.

Prioritize ideas that:

Fill gaps between what VISION.md promises and STATUS.md shows as built
Improve the developer/user experience of existing features
Create useful connections between existing features
Are achievable within a single PR (prefer smaller, composable ideas)

Avoid ideas that:

Duplicate existing open issues
Require major architectural changes for minimal benefit
Don't align with the project's design principles (read VISION.md for these)

3. Discuss Interactively

Present the ideas and invite the user to react:

"Which of these interest you most?"
"Should I explore any of these further?"
"Any related ideas this sparks?"

Be conversational. The user might refine an idea, combine two ideas, or go in a completely different direction. Follow their lead.

4. Draft the Issue

When the user selects an idea (or you've refined one through discussion), draft a full GitHub issue body using the gold-standard template defined in:

.claude/skills/shared/issue-quality-checklist.md

Before writing the draft:

Explore the codebase to understand how the feature would integrate
Check for related code patterns that the feature should follow
Identify the right module/package location and storage approach

Present the full draft to the user for review. Iterate if they have feedback.

5. Create the Issue

After the user approves the draft, create the issue:

gh issue create --repo $REPO --title "<title>" --body "<body>" --label "<labels>"

Choose labels based on the label guide in the shared checklist. Always ask for explicit approval before running gh issue create.

Guidelines

Be specific, not generic. "Add a /api/health endpoint with dependency checks" is better than "improve the API."
Ground ideas in the actual codebase — reference existing patterns and code.
Don't overwhelm — 3-5 ideas is the sweet spot. Quality over quantity.
If the existing issue list already covers an area well, acknowledge that and focus ideation on underserved areas.

Más de este repositorio

mismo repositorio

forge-loop

rnwolfe/forge

Run the full forge pipeline end-to-end within this Claude Code session, processing multiple issues autonomously until a stop condition is met

2026-04-072

onboard

rnwolfe/forge

Interactive project bootstrapping — generates CLAUDE.md, VISION.md, CI workflow, forge.toml, and configures GitHub

2026-03-282

await-ci

rnwolfe/forge

Poll a PR's CI checks until all pass, any fail, or timeout — then report the result

2026-03-282

dispatch

rnwolfe/forge

Pick the next backlog/ready issue, claim it, and prepare a branch — without implementing it

2026-03-282

merge-pr

rnwolfe/forge

Perform a final review check on a PR and merge it if clean, or report what's blocking

2026-03-282

review-pr

rnwolfe/forge

Process all open review comments on a PR: fix addressable issues, reply in-thread, create follow-up issues for deferred items

2026-03-282

name	brainstorm
description	Generate feature and enhancement ideas through codebase exploration and interactive discussion
disable-model-invocation	true

Brainstorm — Feature Ideation

Input

The user may provide a focus area as an argument: $ARGUMENTS

Examples:

/brainstorm — open-ended ideation across the whole project
/brainstorm auth — ideas focused on the authentication system
/brainstorm ux — UX and polish improvements
/brainstorm api — ideas for the API surface

Process

1. Gather Context

Read forge.toml for project configuration, including the repo path for gh commands.

Read these files to understand the project landscape:

docs/internal/VISION.md — product identity, design principles, planned features
docs/internal/STATUS.md — what's built, what's next, current phase
CLAUDE.md — architecture patterns, key files, coding conventions

Then fetch the current issue list to avoid duplicating existing ideas:

gh issue list --repo $REPO --state open --limit 50 --json number,title,labels

If a focus area was provided, also explore the relevant code:

Read the source files related to the focus area
Look at existing patterns that the focus area builds on
Identify gaps between what exists and what VISION.md describes

2. Generate Ideas

Produce 3-5 concrete ideas sorted by estimated impact. For each idea:

Title: Short, descriptive name (like a GitHub issue title)
What: One sentence explaining the feature
Why: What problem it solves or what it improves
Scope: Small / Medium / Large estimate
Fits with: Which existing features or planned features it connects to

If a focus area was given, all ideas should relate to that area. Otherwise, spread ideas across different parts of the project.

Prioritize ideas that:

Fill gaps between what VISION.md promises and STATUS.md shows as built
Improve the developer/user experience of existing features
Create useful connections between existing features
Are achievable within a single PR (prefer smaller, composable ideas)

Avoid ideas that:

Duplicate existing open issues
Require major architectural changes for minimal benefit
Don't align with the project's design principles (read VISION.md for these)

3. Discuss Interactively

Present the ideas and invite the user to react:

"Which of these interest you most?"
"Should I explore any of these further?"
"Any related ideas this sparks?"

Be conversational. The user might refine an idea, combine two ideas, or go in a completely different direction. Follow their lead.

4. Draft the Issue

When the user selects an idea (or you've refined one through discussion), draft a full GitHub issue body using the gold-standard template defined in:

.claude/skills/shared/issue-quality-checklist.md

Before writing the draft:

Explore the codebase to understand how the feature would integrate
Check for related code patterns that the feature should follow
Identify the right module/package location and storage approach

Present the full draft to the user for review. Iterate if they have feedback.

5. Create the Issue

After the user approves the draft, create the issue:

gh issue create --repo $REPO --title "<title>" --body "<body>" --label "<labels>"

Choose labels based on the label guide in the shared checklist. Always ask for explicit approval before running gh issue create.

Guidelines

Be specific, not generic. "Add a /api/health endpoint with dependency checks" is better than "improve the API."
Ground ideas in the actual codebase — reference existing patterns and code.
Don't overwhelm — 3-5 ideas is the sweet spot. Quality over quantity.
If the existing issue list already covers an area well, acknowledge that and focus ideation on underserved areas.