β¨ feat(skills): drag skill chips + register agent-document skills#15095
Merged
Conversation
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
There was a problem hiding this comment.
π‘ Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 39ffca7b1e
βΉοΈ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with π.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
Codecov Reportβ Patch coverage is Additional details and impacted files@@ Coverage Diff @@
## canary #15095 +/- ##
===========================================
- Coverage 83.41% 70.79% -12.62%
===========================================
Files 475 3123 +2648
Lines 35685 309838 +274153
Branches 6946 27987 +21041
===========================================
+ Hits 29765 219346 +189581
- Misses 5758 90325 +84567
- Partials 162 167 +5
Flags with carried forward coverage won't be shown. Click here to find out more.
π New features to boost your workflow:
|
β¦at input Pick a project skill from the right Skills panel and drop it onto the chat input to insert a `/<skill-name>` action tag β the same end state as picking it from the `/` slash menu. - `SKILL_DRAG_MIME` lives in `@lobechat/const` so both the producer (sidebar) and the consumer (input drop handler) share one source of truth. - `skillDragData.ts` owns the drag payload and a custom drag image: a themed "icon + name" chip centered above the cursor. The native drag image is suppressed by an invisible 1Γ1 ghost β the OS bakes its own drop shadow into it which no CSS can remove. Token values are resolved via `getComputedStyle` against the dragged row so the chip stays themed even though it mounts on `document.body`. - `useSkillDrop` listens on the input container and only reacts to the `application/x-lobe-skill` MIME, so it never interferes with the file-upload drop zone (which keys off `Files`). - `ProjectLevelSkills` and `SkillsGroup` wire drag-start with the `projectSkill` category, matching the existing slash-menu behaviour (markdown serializes to `/<skill-name>`). Agent-document skills (the ζΊθ½δ½ Skills group) are not wired here β they need to be registered as first-class skills in the runtime registry first; that work is tracked separately. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
39ffca7 to
30b3d7a
Compare
β¦ mention menu - zh-CN: workingPanel.skills.* and resources.filter.skills now use ζθ½ (covers the Space tab pill plus the agent/project skill section headers) - Wire SkillStore tab and ChatInput mention categories through t() instead of hardcoded English labels; add mention.category.* keys for the five @-menu groups (Agents / Members / Topics / Skills / Tools) Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
β¦gistry
Agent-document skill bundles (the "ζΊθ½δ½ Skills" panel group, stored as
isSkillBundle documents in agent_document) become first-class runtime
skills end-to-end, so the slash menu / drag chip / model activation all
share one source of truth.
Identifier convention: `agent-document:<filename>` (where `<filename>`
is the bundle's slug β `validateSkillName`-validated on the server). The
prefix prevents collisions with builtin / DB skill names; mirrors the
`project:<name>` convention used for filesystem project skills.
Server:
- `aiAgent/index.ts` SkillEngine assembly: query
`agentDocumentsService.getAgentDocuments(resolvedAgentId)`, filter
`isSkillBundle`, and merge into the skills array so the model sees
them in `<available_skills>`.
- `toolExecution/serverRuntimes/skills.ts` factory: when an `agentId`
is in the request context, load the bundles + their SKILL.md index
children and shape them as `BuiltinSkill` entries, then concat with
`filterBuiltinSkills(builtinSkills)` before constructing
`SkillsExecutionRuntime`. The runtime resolves builtins by `name`
with no DB lookup β so `activateSkill('agent-document:<filename>')`
now returns the SKILL.md content for free, no `SkillRuntimeService`
extension needed. `source: 'builtin'` on these entries is a
type-system carrier shape, not a claim that they're real builtins.
Client:
- New tool-store slice `agentDocumentSkills` (per-agent scoped, cleared
on agent switch). `useFetchAgentDocumentSkills(agentId)` is the SWR
hook that keeps the registry hydrated; shares the SWR key with the
working-sidebar panel so we never double-fetch.
- `useInstalledSkillsAndTools` now reads from the new slice and triggers
the SWR hook with the active agent's id, so the `/` menu and any
consumer that goes through that hook see agent-doc skills alongside
builtin / lobehub / market / user skills.
- `AgentDocumentsGroup` wires `onSkillDragStart` on its SkillsList: the
payload uses the runtime identifier (`agent-document:<filename>`),
while the chip label keeps the human-readable title.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
β¦render <skill> tags Three intertwined fixes around the agent-document skill registry that the earlier commit (331eed1) shipped half-baked: 1. **Prefix renamed `agent-document:` β `agent-skills:`** and extracted to `@lobechat/const` (`AGENT_SKILLS_IDENTIFIER_PREFIX`, `buildAgentSkillIdentifier`, `parseAgentSkillIdentifier`). The new prefix mirrors the unified VFS skill namespace path `./lobe/skills/agent/skills/<name>` flattened to one token, and single-sourcing it through const stops drift between the server resolver and the client drag wiring. 2. **`AgentDocumentsService.getAgentSkills(agentId)`** β one place to query bundles, filter `isSkillBundle`, resolve the `SKILL.md` index child, and build the runtime identifier. Both the SkillEngine assembly in `aiAgent/index.ts` and the `SkillsExecutionRuntime` factory in `serverRuntimes/skills.ts` call it instead of each re-implementing the prefix + bundle β index lookup (which was how the two sides drifted last round). 3. **`<skill>` / `<tool>` markdown plugins** (`plugins/Skill`, `plugins/Tool`) so the chat bubble renders these tags as the same chip the editor uses, instead of leaving the literal `<skill name="β¦" />` text in the message. Fixes a pre-existing bug that affected all registered skills (builtin / lobehub / DB / agent- document) β only the bare-text `projectSkill` flavour rendered correctly before because it serializes to `/<name>` instead. Note: the client drag wiring in `AgentDocumentsGroup.tsx` and the client tool-store slice action import the new const helpers, but landing the *category* refactor (`'skill'` β `'agentSkill'`) and the shared `@/features/SkillsList` extraction is intentionally kept out of this commit so it can ship with its own ActionTag work. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
β¦ category - New src/features/SkillsList/ bundle: SkillsList moved here from AgentDocumentsExplorer, joined by a shared SkillSection wrapper (optional collapsible sectionHeader prop unifies the Accordion / flat-header variants) and a useProjectSkills hook (SWR + open handlers). - AgentDocumentsGroup / ProjectLevelSkills / SkillsGroup now consume that bundle and drop ~340 lines of duplicated SWR + section UI. - ActionTag gains an 'agentSkill' UI category (types, mention card, style, en/zh editor copy) so agent-document skill chips render with their own tooltip / label while still serializing as <skill name="agent-skills:..." /> on the wire β the runtime keys off the identifier prefix, so no new XML tag is needed. The XML reader detects the prefix on parse to keep the chip's category across save/reload. - AgentDocumentsGroup drag uses category='agentSkill', backed by the shared buildAgentSkillIdentifier helper. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
β¦ error
Adapter previously surfaced overload (`api_error_status: 529` /
`overloaded_error`) as a plain `{ error, message }` payload, so the
executor fell through to the unstructured branch and the UI rendered
the raw text instead of a typed `HeterogeneousAgentSessionError`. Add
a dedicated `overloaded` code + StatusGuide state with a Retry action
so the common transient failure has a recoverable, branded surface.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
β¦oke every skill drag `writeSkillDragData` also set `text/plain` to the chip label, and `setSkillDragImage` swapped in a custom cursor-following preview. The combination races the Lexical chat input's own drop handling: it reacts to `text/plain` and the suppressed-native-image sequence intermittently aborts the dragstart, leaving `useSkillDrop` to never fire. Net result was that every skill drag (project + agent-document) silently failed. Strip both back to the minimum that's known to work: - `writeSkillDragData` writes only the custom `application/x-lobe-skill` MIME + `effectAllowed = 'copy'`. Drops on non-editor targets now do nothing instead of degrading to plain text β acceptable trade-off. - Native browser drag image is back. The OS drop shadow on the ghost is ugly but not a regression worth losing the drag for. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
β¦ools The earlier commit (331eed1) wired the agent-document skill registry into `useInstalledSkillsAndTools` by calling the SWR hook directly off the tool-store selector: useToolStore((s) => s.useFetchAgentDocumentSkills)(activeAgentId); That extra hook indirection β invoking a function selected out of zustand on each render of the slash-menu consumer β was throwing / breaking React's hook tracking at render time. The slash menu and every drag-into-input flow rely on `useInstalledSkillsAndTools` resolving cleanly, so the breakage cascaded into `/skills` not rendering and every skill drag silently failing. Revert to the pre-331eed1e9c shape: only the four already-working sources (builtin / lobehub / market / user) feed the slash + mention list. Agent-document skills are still in the tool store (server side registers them in SkillEngine via `agent-skills:<filename>`) β they just won't show up in the `/` autocomplete until we hydrate the slice through a safer path (e.g. an effect in the agent route root, or shared SWR from the panel). Drag from the working sidebar continues to work because the wiring is local to `AgentDocumentsGroup`, not to `useInstalledSkillsAndTools`. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
β¦ cursor) Brings back the cursor-following white rounded chip (icon + name) and suppresses the native OS drag ghost. Earlier reverted along with the `text/plain` fallback when we were narrowing down the drag breakage, but the real culprit turned out to be the `useFetchAgentDocumentSkills` hook indirection in `useInstalledSkillsAndTools` (fixed in 1ccdfc5), not the drag-image code itself. `text/plain` stays removed β that one really does race with Lexical. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
This was referenced May 22, 2026
arvinxx
added a commit
that referenced
this pull request
May 29, 2026
# π LobeHub Release (20260528) **Release Date:** May 28, 2026 **Since v2.2.0:** 220 merged PRs Β· 15 contributors > This cycle brings heterogeneous "platform agents" you can dispatch to local or remote devices, a rebuilt onboarding flow, document-centric chat, and a unified model-runtime error model β with new DeepSeek V4 and Gemini 3.5 Flash support along the way. --- ## β¨ Highlights - **More Hetero Agents (OpenClaw / Hermes)** β Create heterogeneous agents and dispatch them to local or remote devices through the device gateway, with an execution-target switcher in the composer and persistent CLI sessions. (#15065, #15179, #15022) - **iMessage on Desktop** β New iMessage setup and bridge on desktop, plus bot attachments across every platform. (#15228, #15227, #15029) - **Skills in the Composer** β Drag skill chips into chat, trigger installed skills from the slash menu mid-line, and surface project-level skills in the homogeneous agent runtime. (#15095, #15061, #15110) - **New Models** β DeepSeek V4 Flash/Pro and Gemini 3.5 Flash across providers, with thinking params for structured output and chat cost estimates. (#15031, #15001, #15051, #14876) - **Agent Runtime Observability** β OpenTelemetry GenAI semantic conventions plus per-call generation tracing. (#15123, #15124) --- ## π€ Agents & Heterogeneous Runtime - **Platform agent creation** β OpenClaw/Hermes creation UI, device guard, and remote dispatch backend. (#15065) - **Execution-target switcher** β Pick local vs remote execution directly in the composer; device-selection UX with actionable guidance. (#15179, #15111) - **CLI hetero dispatch** β OpenClaw/Hermes dispatch with persistent sessions and a notify protocol. (#15022) - **Gateway snapshot as source of truth** β Consume the gateway `uiMessages` snapshot at step boundaries to keep chat state consistent. (#15153, #15152) - **Client sub-agent as a normal tool call** β Simplifies the sub-agent execution path. (#15281) - **Hermes agent chain** β Implements the Hermes agent chain logic. (#15189) - **Device registry** β TRPC endpoints to register, list, update, and remove devices. (#15299) - **Desktop device routing** β Route gateway agent runs through `lh hetero exec`; restore `userId` in gateway dispatch and gate local-system by execution target. (#15132, #15232) - **Agent signals** β Anchor agent-signal receipts to messages and isolate memory-agent messages into a child thread. (#14969, #14921) --- ## π Onboarding - **Simplified first screen** β Defer topic creation to first send. (#15090) - **Market Agent Picker** β Added as a classic onboarding step, with template prefetch. (#14980, #15041) - **Welcome guidance** β Show agent welcome guidance on first run. (#15098) - **Mobile** β Adapt agent onboarding UI and restore Classic-step padding on mobile. (#15019, #15032) - **Discovery** β Streamline discovery to a single profession question. (#14987) - **Analytics** β Track onboarding step events and create-agent modal source. (#15133, #15028) --- ## π Documents, Pages & Knowledge - **Thread chat in preview** β Embed thread chat in the document preview portal. (#15216) - **Non-markdown rendering** β Render non-markdown docs as a read-only highlight. (#15272) - **Multi-select** β Multi-select delete in the document tree. (#15125) - **Page-agent streaming** β Preview `initPage` streaming arguments. (#15039) - **Per-agent topics** β Per-agent topic management page. (#15207) - **Server-side category** β Derive document category server-side and drop frontend predicates. (#15076) --- ## 𧩠Skills & Tools - **Drag skill chips** β Drag skills into chat input and register agent-document skills. (#15095) - **Slash menu** β Installed skills appear in the slash menu with a mid-line trigger. (#15061) - **Project skills** β Recognize project-level skills in the homogeneous agent runtime and surface them regardless of active device. (#15110, #15177) - **VFS archiving** β Archive oversized tool results to VFS instead of truncating. (#15074) - **@localfile mentions** β Drag folders into chat input as `@localFile` mentions on desktop. (#15071) --- ## π§ Model Runtime & Providers - **Error spec registry** β Unify error codes into a spec + pattern registry, split `ProviderBizError` into finer codes, classify Cloud-only codes via a tier digit, and add `DatabasePersistError`. (#15262, #15286, #15278, #15279) - **New models** β DeepSeek V4 Flash/Pro (opencode-go) and Gemini 3.5 Flash; DeepSeek V4 Pro on SiliconCloud. (#15031, #15001, #15017, #15267) - **Structured output** β Thinking params for structured output, Bedrock structured generation, and DeepSeek `generateObject` tool choice. (#15051, #15174, #15054) - **Cost** β Chat cost estimate support; preserve usage cost in custom streams. (#14876, #15218) --- ## π¬ Chat & User Experience - **Follow-up chips** β Extend follow-up chip suggestions to general chat with scene-specific model config. (#15101, #14797) - **Input drafts** β Persist unsent input drafts across tab switches and prevent repeated draft restore. (#14992, #15024) - **Command menu** β Order topic/message search by recency and promote inline type filters. (#15094, #14986) - **Zoom HUD** β Show a zoom-level HUD on Cmd +/β and Cmd 0. (#15294) - **Copy** β Unescape markdown escapes when copying user messages. (#15253) --- ## π₯οΈ Desktop - **App Nap fix** β Prevent App Nap from dropping the gateway WebSocket during display sleep. (#14994) - **File preview** β Preview `.cjs`/`.mjs`/no-extension files instead of binary fallback and expand `~` when opening local files. (#15168, #15284) - **Cross-platform settings** β Open settings via main-window navigation on Windows/Linux and restore the route after an update restart. (#15036, #14922) - **Token refresh** β Prevent frequent logout from token-refresh retries. (#14928) --- ## π Observability - **OTel GenAI** β Instrument Agent Runtime with OpenTelemetry GenAI semantic conventions. (#15123) - **Generation tracing** β Per-call `llm_generation_tracing` with a pre-allocated tracingId and recordFeedback router. (#15124, #15146) - **Error classification** β Persist `ERROR_CODE_SPECS` classification on operation errors. (#15273) --- ## ποΈ Database Migrations - **Batch migrations** β Topic usage stats, push tokens, `tasks.editor_data`, and document shares. (#15280) - **Tracing & eval tables** β Add `llm_generation_tracing` and agent eval experiment tables. (#15126) > Self-hosted operators should run the database migration (`pnpm db:migrate`, or restart with auto-migrate enabled) after upgrading. The changes are additive and backwards-compatible. --- ## π Security & Reliability - **Security:** Remove the `getPlaintextCred` tool to prevent plaintext credential exposure. (#14998) - **Security:** Prompt account selection for Google OAuth and add `prompt=consent` to the OIDC authorization URL to fix missing refresh tokens. (#15234, #15010) - **Reliability:** Preserve streamed content across a mid-stream cancel. (#15173) - **Reliability:** Bound the Redis command timeout and configure the Anthropic client timeout. (#15091, #15042) - **Reliability:** Prevent infinite recursion in the assistant chain. (#15288) --- ## π₯ Contributors Huge thanks to **15 contributors** who shipped **220 merged PRs** this cycle. @AnotiaWang Β· @sxjeru Β· @algojogacor Β· @hardy-one Β· @arvinxx Β· @Innei Β· @tjx666 Β· @lijian Β· @AmAzing129 Β· @rdmclin2 Β· @neko Β· @cy948 Β· @CanisMinor Β· @sudongyuer Β· @rivertwilight Plus @lobehubbot and renovate[bot] for maintenance. --- **Full Changelog**: v2.2.0...release/weekly-20260528
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Two related pieces of work on the skill drag-and-slash flow:
1. Drag skill chips from working sidebar into chat input (
30b3d7a524)/slash menu.getComputedStyleagainst the dragged row so the chip stays themed ondocument.body.application/x-lobe-skillMIME so it never interferes with the file-upload drop zone (which keys offFiles).SKILL_DRAG_MIMElives in@lobechat/constso both sides share one source of truth.2. Register agent-document skill bundles in the skill registry (
331eed1e9c)Agent-document skill bundles (the ζΊθ½δ½ Skills panel group, stored as
isSkillBundledocuments inagent_document) become first-class runtime skills end-to-end β so the slash menu, drag chip, and model activation all share one source of truth.Identifier convention:
agent-document:<filename>(the bundle's slug,validateSkillName-validated server-side). Prefix prevents collisions with builtin / DB skill names; mirrors theproject:<name>convention.Server:
aiAgent/index.tsSkillEngine assembly merges agent-doc bundles alongside builtin + DB skills, so the model sees them in<available_skills>.serverRuntimes/skills.tsfactory pre-loads the bundles'SKILL.mdcontent asBuiltinSkill-shaped entries and feeds them toSkillsExecutionRuntime'sbuiltinSkillschannel. The runtime's existingactivateSkillresolves builtins bynamewith no DB lookup β soactivateSkill('agent-document:<filename>')returns the SKILL.md content for free, noSkillRuntimeServiceextension needed.Client:
agentDocumentSkills(per-agent scoped, cleared on agent switch).useFetchAgentDocumentSkills(agentId)is SWR-backed and shares the SWR key with the panel β at most one network request.useInstalledSkillsAndToolsreads the new slice + triggers the SWR hook with the active agent's id, so/menu / @ mention see agent-doc skills alongside builtin / lobehub / market / user skills.AgentDocumentsGroupwiresonSkillDragStart: payload uses the runtime identifier (agent-document:<filename>); chip label keeps the title.Also picks up the user's i18n localization commit (
d48cfb0455) on the way.Test plan
/<skill-name>into the input β send β CLI agent resolves and runs the skill.<skill name="agent-document:<filename>">chip β send β model sees it in<available_skills>βactivateSkillreturns the SKILL.md content (no longer "skill not found")./) shows agent-doc skills alongside installed skills.Files).π€ Generated with Claude Code