Skip to main content

ac-ideate

Run Auto-Claude AI-powered ideation for any project directory. Generates low-hanging fruit, high-value features, and UI/UX ideas using the Auto-Claude backend. Use when brainstorming new specs, features, or architectural directions for a project.

Jump to install

Source facts

Repository
ryanmaclean/skills
Last source activity
April 15, 2026 at 03:58
Detected SKILL.md language
English
Stars
1
Forks
0

Install options

The review-first prompt is selected by default. You can switch to a direct command or download a local copy.

Review the source files

Read SKILL.md and any companion files shown by SkillsMP before deciding whether to install.

Showing SKILL.md

SKILL.md
Source instructions · Read-only preview
name
ac-ideate
description
Run Auto-Claude AI-powered ideation for any project directory. Generates low-hanging fruit, high-value features, and UI/UX ideas using the Auto-Claude backend. Use when brainstorming new specs, features, or architectural directions for a project.
# ac-ideate — Auto-Claude Ideation Runner Runs the Auto-Claude ideation pipeline against a project directory and writes spec ideas to `<project>/.auto-claude/ideation/`. ## Binary location ``` /Applications/Auto-Claude.app/Contents/Resources/backend/runners/ideation_runner.py ``` Always set `PYTHONPATH` to the backend root: ``` PYTHONPATH=/Applications/Auto-Claude.app/Contents/Resources/backend ``` ## Basic usage ```bash # Ideate on current directory PYTHONPATH=/Applications/Auto-Claude.app/Contents/Resources/backend \ python3 /Applications/Auto-Claude.app/Contents/Resources/backend/runners/ideation_runner.py \ --project /path/to/project # With options PYTHONPATH=/Applications/Auto-Claude.app/Contents/Resources/backend \ python3 /Applications/Auto-Claude.app/Contents/Resources/backend/runners/ideation_runner.py \ --project /path/to/project \ --types low_hanging_fruit,high_value_features \ --thinking-level high \ --max-ideas 8 \ --refresh ``` ## All flags | Flag | Default | Description | |------|---------|-------------| | `--project PATH` | cwd | Project directory to analyze | | `--output PATH` | `<project>/.auto-claude/ideation` | Where to write results | | `--types LIST` | all | Comma-separated: `code_improvements`, `ui_ux_improvements`, `documentation_gaps`, `security_hardening`, `performance_optimizations`, `code_quality` | | `--thinking-level` | `medium` | `low` / `medium` / `high` | | `--max-ideas N` | 5 | Max ideas per type | | `--model` | `sonnet` | `haiku`, `sonnet`, `opus`, or full model ID | | `--refresh` | false | Force regenerate even if results exist | | `--append` | false | Add to existing session instead of replacing | | `--no-roadmap` | false | Skip roadmap context | | `--no-kanban` | false | Skip kanban context | | `--fast-mode` | false | Faster Opus 4.6 output | ## Open a Zellij tab for ideation ```bash # New tab — runs ideation then drops to shell zellij --session $ZELLIJ_SESSION_NAME action new-tab --name "ideation" \ -- sh -c 'cd /path/to/project && \ PYTHONPATH=/Applications/Auto-Claude.app/Contents/Resources/backend \ python3 /Applications/Auto-Claude.app/Contents/Resources/backend/runners/ideation_runner.py \ --project . --thinking-level high --max-ideas 8 --refresh; exec $SHELL' ``` `$ZELLIJ_SESSION_NAME` is set automatically inside any Zellij session. If running outside, replace with the session name from `zellij list-sessions`. ## Typical workflow 1. **New project** — run with `--types low_hanging_fruit,high_value_features --refresh` 2. **Read results** — `ls <project>/.auto-claude/ideation/` 3. **Promote to specs** — copy ideas into `.auto-claude/specs/<NNN>-<slug>/spec.md` 4. **Dispatch** — `./dispatch.nu run --wave N --yes` or via Auto-Claude UI ## Projects | Project | Path | |---------|------| | dub (EFI bootloader) | `$HOME/efi` | | bop (jobcard system) | `$HOME/bop` | | zam (unikernel VM reader) | `$HOME/zam` | ## Notes - Ideation reads `PLAN.md`, existing specs, and `CLAUDE.md` for context automatically. - `--refresh` is safe to run repeatedly — it overwrites the previous session. - Does **not** require an API key — uses Auto-Claude's OAuth session (Claude Code subscription). - Inside an agent session, `CLAUDECODE` env var is already set; no need to unset it for this runner.
View on GitHub