7 Commits

Author SHA1 Message Date
Matt Van Horn
b6a79003ad
feat(design-system): export the generated design system as design tokens (#577) 2026-11-16 05:07:39 +07:00
Matt Van Horn
efd9822135
feat(design-system): write a self-contained preview page for the generated system (#576) 2026-11-16 02:05:27 +07:00
wangyuehua
bf41cbc3ce
fix: align AutoClaw validation baselines (#550)
Co-authored-by: wangyuehua <wangyuehua0429@users.noreply.github.com>
2026-10-31 23:07:44 +07:00
Elvin Novruzov
076a2b3a31
feat(cli): add Kimi Code platform (--ai kimi) (#538)
* feat(cli): add Kimi Code platform (--ai kimi)

Kimi Code discovers skills as SKILL.md folders (name + description frontmatter) in .kimi-code/skills/ (project) and ~/.kimi-code/skills/ (user). Add a kimi platform template, register the type, detect an existing .kimi-code folder, and document it in the README.

Tests: add kimi to the dry-run and script-path e2e specs. check:assets, typecheck and those specs pass; init --ai kimi was run in an empty repo. Kimi Code itself was not run.

Refs #455.

* test(cli): add kimi install and detection specs

Cover the rendered SKILL.md frontmatter, a real install into .kimi-code/skills (skill, data, scripts and sub-skills), and detection of a .kimi-code folder. Refs #455.

* fix(scripts): expect 24 platform configs in validate-agent-guide

Adding the kimi platform template raises the config count from 23 to 24. Keep the check strict and update the expected count. Refs #455.

* test(cli): expect 24 platform configs in agent-query-contract

The kimi platform raises the platform config count from 23 to 24. Refs #455.
2026-10-22 23:08:14 +07:00
XiaOldMan
7246805992
feat: add OpenClaw platform support (#527)
* feat: add OpenClaw platform support

Add OpenClaw as an install target so the skill can be installed with
`uipro init --ai openclaw` and removed via `uipro uninstall`.

OpenClaw discovers AgentSkills-format skills from `.agents/skills` in the
agent workspace and from `~/.agents/skills` (personal library on the default
state dir), so the platform config reuses the `.agents` layout already used
by the universal/codex targets with its own display name, detection entry,
and docs.

- new platform config src/ui-ux-pro-max/templates/platforms/openclaw.json
  (synced to cli/assets via sync:assets)
- register 'openclaw' in AIType, AI_TYPES, AI_FOLDERS, AI_TO_PLATFORM
- detect the shared .agents workspace directory and describe the platform
- keep the .agents-family suggestion on codex when all four .agents
  consumers are detected (length 2-3 -> 2-4)
- cover openclaw in the dry-run, template-script-paths, and
  agent-query-contract e2e specs
- document OpenClaw in README and cli/README

* chore: bump version to 2.15.0 in skill metadata files

Aligns skill.json, marketplace.json, and plugin.json with the latest
v2.15.0 release; main was still at 2.13.0. CLI package version stays on
its own npm release track.

* fix: cover OpenClaw in offline validator and detection regression

- bump scripts/validate-agent-guide.py PLATFORM_COUNT 21 -> 23 (main
  already had 22 configs with a stale count of 21; OpenClaw widens the
  gap to 23, so update the gate and keep it meaningful)
- add e2e regression: a sole .agents directory detects all four
  consumers (antigravity, codex, universal, openclaw) and still
  suggests codex rather than all
2026-10-12 18:08:03 +07:00
Neal Meyer
477bcb28c9
fix: stop Cursor skill pointing at empty Quick Reference (#509)
* fix: stop Cursor skill pointing at empty Quick Reference

Platforms that set quickReference to false omitted the inlined
reference but still told the agent to use "the Quick Reference
sections above." Resolve QR/NOQR variants so Cursor (and other
non-inlined platforms) follow the existing search flow instead.
Claude and Codex keep the inlined reference wording.

Co-authored-by: Neal Meyer <ngmeyer@users.noreply.github.com>

* fix: accept 21 platform configs after ZCode

ZCode landed on main without updating the agent-guide validator
or the platform-count e2e assertion. This branch did not add a
platform; the stale expected count of 20 was failing CI.

Co-authored-by: Neal Meyer <ngmeyer@users.noreply.github.com>

* fix: refresh stale relevance runtime fingerprint

verify:data would fail next on evaluate:relevance because main's
threshold manifest still hashed the pre-#500 runtime. Metrics are
unchanged; only the fingerprint now matches current scripts/data.

Co-authored-by: Neal Meyer <ngmeyer@users.noreply.github.com>

---------

Co-authored-by: Cursor Agent <cursoragent@cursor.com>
Co-authored-by: Neal Meyer <ngmeyer@users.noreply.github.com>
2026-10-03 23:05:26 +07:00
Viet Tran
a38d04c3d5
feat(search): overhaul relevance and curated design data
Overhaul BM25 relevance, reasoning and data-quality contracts; refresh UI styles and framework guidance; add resilient text, chip, badge and micro-interaction guidance; strengthen release, provenance and catalog refresh gates; update bilingual documentation.
2026-08-14 00:08:23 +07:00