Feature discovery

Turns "what should we build next?" into a grounded, ranked roadmap by fanning out many agents instead of one agent guessing. The whole run is bundled as a deterministic Workflow script so every invocation follows the same pipeline rather than being re-improvised.

When this fits

Reach for it when the user wants ideas that are researched, deduped, specced, and pressure-tested, not a quick off-the-cuff list. It is deliberately heavyweight (roughly 35-40 agents at full depth), so for a casual "give me three ideas" just answer directly; use this when they want the exhaustive version.

How it works

Five phases (see scripts/feature-discovery.workflow.js):

1. Ground - one agent maps the current product by reading the repo (entry points, routes, data models, content, services, config, docs); in parallel a planner proposes competitor segments, then one analyst per segment researches it via web search. The product is always mapped, even in internal scope, so ideation never re-proposes what exists.
2. Ideate - one agent per product lens (discoverability, core value, UX, monetization, engagement, trust, IA), each grounded in the inventory + competitor findings, each forbidden from proposing anything that already exists.
3. Shortlist - a single curator merges duplicates, drops the trivial and the already-built, ranks by value-to-effort, and picks the top 8-12.
4. Spec & validate - a pipeline: spec each feature, then an adversarial skeptic scores novelty, real value, feasibility in the current architecture, competitor precedent, and maintenance burden.
5. Synthesize - one agent writes the final Markdown report: exec summary, gaps (internal + vs competitors), a ranked roadmap table, full specs for the build/maybe features, dropped ideas with reasons, and a quick-wins-vs-bigger- bets split.

Running it

Invoke the Workflow tool pointed at the bundled script. Do not paste the script inline; use its path so the persisted version stays the source of truth:

Workflow({
  scriptPath: ".claude/skills/feature-discovery/scripts/feature-discovery.workflow.js",
  args: { product: "one-line description of your product", scope: "mixed", depth: "exhaustive" }
})

The workflow runs in the background and returns { report, meta, counts } on completion, where report is the finished Markdown.

Parameters (args)

If the user does not specify, ask for product, scope, and depth (or default to a mapped product + mixed + exhaustive, the intended full run) before launching.

Presenting the result

• Lead with the executive summary and the ranked roadmap table in chat, then offer to render the full report as an Artifact for readability.
• Report the run counts (raw ideas / shortlisted / specced) so the user can see the funnel.
• Treat the report as a working artifact, not a committed file, unless the user asks to save it.
• If error is returned (product-mapping-failed or empty-shortlist), say so plainly and offer to rerun rather than presenting a half-report.

Extending

The lens list (ALL_LENSES) lives at the top of the script; add or reword entries to change ideation coverage. Competitor segments are planned dynamically from the product, so no market list is hardcoded. The phase wiring and schemas do not need to change.