| name | compare-ideas |
| description | Compare multiple captured ideas head-to-head across explicit dimensions — commercial prospect, speed to first revenue, capital intensity, fit-to-user, defensibility, social impact — and produce a ranked recommendation. Use when the user has 2+ ideas in the workspace and wants help choosing where to focus. |
Compare Ideas
Ranked head-to-head comparison across the user's idea portfolio.
Steps
- Read
<workspace_path>/ideas/ — list available idea slugs. Confirm with the user which ideas to include in the comparison (default: all ideas with at least one lens analysis).
- For each idea, gather the available lens outputs. If an idea has only
idea.md and no analysis, warn the user that the comparison will be shallow for that one.
- Score each idea on a consistent rubric (1–10):
- Commercial prospect (TAM, defensibility, pricing power)
- Speed to first revenue (weeks to first paying customer)
- Capital efficiency (how far user can take it on bootstrap budget)
- Fit to user (per user-context.md — domain expertise, capital appetite, risk profile, values)
- Defensibility / moat
- Social impact alignment
- Solo-buildable (can the user, plus realistic help, ship it)
- Compute a weighted overall (default equal weights — let the user override).
- Build a comparison matrix table.
- Write a verdict paragraph: which idea to pursue first, which to defer, which to drop. Be opinionated.
Output
<workspace_path>/comparisons/YYYY-MM-DD-HHMM.md (workspace-level, not per-idea) with frontmatter:
---
type: idea-comparison
generated: YYYY-MM-DD
ideas: [<slug1>, <slug2>, ...]
ranked: [<slug>, <slug>, ...]
recommended: <slug>
---
Notes
- If the user has done deep analysis on some ideas and not others, weight the comparison conservatively for the under-analysed ones (mark scores as low-confidence).
- Surface non-numeric considerations the rubric misses — e.g. "Idea A is the rational choice but you've told me you don't want to spend three years in B2B sales".