v3.3.0

Product landscape — a registry of competitors, adjacent players and benchmarks, and six optional research connectors. New skill product-landscape scans the user's machine, discovers products by category and market, categorises and characterises them with sources, maps the category, and starts the same-flow research across any number of products the user picks. MINOR: new skill, new storage folder, new vault type, new optional connectors.

Added

• skills/product-landscape v0.1.0 — modes scan / discover / add / update / characterize / map / research; shared registry ~/.grow-pm/landscape/ (registry.yaml, products/<slug>.md with kind, category, markets, platforms, walkable surfaces, size signals, characterization, per-product role links, sources, status, last_researched, walkthroughs[]); candidates confirmed in batches of 10–15; research proposals ranked with no cap; any product the plugin touches is registered as status: auto; example examples/shopping-category-scan.md.
• scripts/landscape_scan.sh — one JSON line per candidate: Mac apps, iPhone apps installed from the Mac App Store (/Applications/*.app/Wrapper), adb packages; Chrome/Safari bookmark domains only with --bookmarks (after a per-scan yes); never browser history.
• Six optional connectors in .mcp.json — lazyweb, similarweb, mobbin, semrush, tavily, apify (URL-matched; the user's own accounts; no skill requires them); rows in integration-strategy.md 1a and a paragraph on when each matters (tavily for hosts without web search, apify for Google Play data).
• templates/built-in/research/landscape-v1.md (26 seed templates); vault type landscape → Research/landscape/ (36 types); storage folder landscape/.
• plugin-configurator Landscape setup (Extended add-on, deferred id landscape): bookmarks consent, product category, import of product.competitors into the registry; schema section ## Landscape.
• Trigger-evals Group O; test cases for scan, discover, map, research and the no-registry regression; lint vocabulary for roles and source types.

Changed

• product-research 0.10.6 — competitor list from the registry when it exists, new competitors registered as auto, post-study offer; flow-walkthrough 0.2.1 — compare candidates from the registry, walked products registered, acted-on references registered as inspiration; cjm-research 0.7.6 and brainstorm-features 0.10.3 — landscape offers, landscape:<slug> evidence; plugin-configurator 2.9.5; local-context-protocol.md documents product.competitors as the registry seed.

Not in this version (next)

Periodic re-scans and change alerts, price monitoring, review-sentiment mining across stores, automatic app installs; competitor account creation — never.

Backwards compatibility

Full. Without a registry every skill behaves as before (product.competitors); the six connectors are optional and unconnected by default.