♻️ refactor: implement SkillResolver, BaseSystemRoleProvider, and agent document injection pipeline by arvinxx · Pull Request #13315 · lobehub/lobehub

arvinxx · 2026-03-26T16:04:10Z

Summary

Implements two-layer skill resolution architecture mirroring ToolsEngine + ToolResolver
Creates BaseSystemRoleProvider base class to unify system message assembly
Splits AgentDocumentInjector into 5 position-based injectors for correct pipeline placement
Extracts @lobechat/agent-templates package for agent document templates
Adds server-side agent document injection in RuntimeExecutors (was missing)
Auto-enables lobe-agent-documents tool when agent has documents
Adds BOOTSTRAP.md template for first-run agent onboarding

Key Changes

SkillResolver

Add SkillResolver, OperationSkillSet, StepSkillDelta, ActivatedStepSkill types in @lobechat/context-engine
Enhance SkillEngine with enableChecker and generate() method
Wire SkillResolver into RuntimeExecutors call_llm
Replace manual skillMetas assembly in aiAgent with SkillEngine.generate()
Fixes agent-browser content (~1.5k tokens) being injected into non-desktop scenarios

BaseSystemRoleProvider

New base class for all providers that append to system message
Subclasses only implement buildSystemRoleContent() + optional onInjected() for metadata
Migrated: EvalContext, BotPlatformContext, SystemDate, ToolSystemRole, HistorySummary, SkillContext

Agent Document Injection Pipeline

Split monolithic AgentDocumentInjector into 5 injectors:
- AgentDocumentBeforeSystemInjector (before-system) → Phase 2 start
- AgentDocumentSystemAppendInjector (system-append) → Phase 2 end
- AgentDocumentSystemReplaceInjector (system-replace) → Phase 2 end
- AgentDocumentContextInjector (before-first-user) → Phase 3
- AgentDocumentMessageInjector (after-first-user, context-end) → Phase 4
Server-side RuntimeExecutors now fetches and injects agent documents (was missing)
Result: single system message instead of multiple, correct injection ordering

MessagesEngine Pipeline Reorganization

7 phases with clear responsibilities:
1. History Truncation
2. System Message Assembly (all BaseSystemRoleProvider)
3. Context Injection (before first user message)
4. User Message Augmentation
5. Message Transformation
6. Content Processing
7. Cleanup
All providers use enabled field — no spread conditionals in pipeline

@lobechat/agent-templates Package

Extracted from @lobechat/database into standalone package
Contains types (DocumentLoadPosition, DocumentLoadFormat, etc.), templates, and registry
Claw templates use .md file imports for content
Added BOOTSTRAP.md onboarding template
Fixed template positions: AGENTS→before-system, IDENTITY/SOUL→system-append

Auto-enable Agent Documents Tool

Added AgentDocumentsManifest to defaultToolIds
Server createServerAgentToolsEngine enables tool via hasAgentDocuments rule
AiAgentService queries agent documents to determine flag

Fixes LOBE-6410

Test plan

context-engine: 717 tests passing (51 files)
database: 1833 tests passing
SkillResolver + SkillEngine unit tests
AgentDocumentInjector tests updated for split injectors
Verified single system message via agent tracing
Verified agent documents tool auto-enabled and functional
Verify Discord bot no longer receives agent-browser prompt injection
Verify desktop agent-browser skill still works when enabled

🤖 Generated with Claude Code

vercel · 2026-03-26T16:04:40Z

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
lobehub	Ready	Preview, Comment	Mar 27, 2026 2:00am

sourcery-ai

We've reviewed this pull request using the Sourcery rules engine

lobehubbot · 2026-03-26T16:05:08Z

@nekomeowww - This is a backend/agent-runtime refactor implementing SkillResolver architecture. Please take a look.

codecov · 2026-03-26T16:09:15Z

Codecov Report

❌ Patch coverage is 87.12522% with 73 lines in your changes missing coverage. Please review.
✅ Project coverage is 66.83%. Comparing base (926de07) to head (a0d2e2f).
⚠️ Report is 6 commits behind head on canary.

Additional details and impacted files

@@            Coverage Diff             @@
##           canary   #13315      +/-   ##
==========================================
+ Coverage   66.76%   66.83%   +0.07%     
==========================================
  Files        1889     1891       +2     
  Lines      151446   151326     -120     
  Branches    14506    14520      +14     
==========================================
+ Hits       101112   101142      +30     
+ Misses      50222    50065     -157     
- Partials      112      119       +7

Flag	Coverage Δ
app	`58.22% <78.57%> (+0.01%)`	⬆️
database	`96.66% <100.00%> (+0.01%)`	⬆️
packages/agent-runtime	`89.61% <ø> (ø)`
packages/context-engine	`86.47% <88.88%> (+3.24%)`	⬆️
packages/conversation-flow	`92.36% <ø> (ø)`
packages/file-loaders	`87.02% <ø> (ø)`
packages/memory-user-memory	`66.68% <ø> (ø)`
packages/model-bank	`99.85% <ø> (ø)`
packages/model-runtime	`84.48% <ø> (ø)`
packages/prompts	`67.76% <ø> (ø)`
packages/python-interpreter	`92.90% <ø> (ø)`
packages/ssrf-safe-fetch	`0.00% <ø> (ø)`
packages/utils	`90.41% <ø> (ø)`
packages/web-crawler	`88.82% <ø> (ø)`

Flags with carried forward coverage won't be shown. Click here to find out more.

Components	Coverage Δ
Store	`66.07% <ø> (ø)`
Services	`49.58% <100.00%> (+0.01%)`	⬆️
Server	`67.38% <75.29%> (+0.02%)`	⬆️
Libs	`51.06% <ø> (ø)`
Utils	`91.01% <ø> (ø)`

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 17c299894f

ℹ️ 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 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

chatgpt-codex-connector · 2026-03-26T16:12:42Z

        identifier: s.identifier,
        name: s.name,


Include DB skill content in operation skill metadata

This refactor now auto-activates any skill whose identifier is in enabledPluginIds, but DB skills are still emitted without content. When a user/market skill is enabled in agentConfig.plugins, SkillResolver will mark it activated; then SkillContextProvider drops it from <available_skills> (because it is activated) and also cannot inject it (because content is missing), so the skill disappears from the prompt and becomes unusable. Add s.content here (or avoid activating skills that have no content).

Useful? React with 👍 / 👎.

Introduces a two-layer skill resolution architecture mirroring ToolsEngine + ToolResolver: - SkillEngine (assembly layer): accepts raw skills + enableChecker, outputs OperationSkillSet - SkillResolver (resolution layer): merges operation + step delta + accumulated activations Key changes: - Add SkillResolver, OperationSkillSet, StepSkillDelta, ActivatedStepSkill types - Enhance SkillEngine with enableChecker and generate() method - Wire SkillResolver into RuntimeExecutors call_llm - Replace manual skillMetas assembly in aiAgent with SkillEngine.generate() - Update client-side skillEngineering to use SkillEngine + enableChecker - Add activatedStepSkills to AgentState for step-level skill accumulation Fixes: agent-browser content injected into non-desktop scenarios (Discord bot) due to missing filterBuiltinSkills call in aiAgent LOBE-6410 Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

… documents server-side - Create @lobechat/agent-templates package with types, templates, and registry - Move DocumentLoadPosition, DocumentLoadFormat, DocumentLoadRule, etc. to new package - Move claw templates (AGENTS, BOOTSTRAP, IDENTITY, SOUL) with .md file imports - Add BOOTSTRAP.md as new onboarding template (priority 1, system-append) - Fix template positions: AGENTS→before-system, IDENTITY/SOUL→system-append - Update database package to re-export from @lobechat/agent-templates - Migrate all consumers to import directly from @lobechat/agent-templates - Add agent documents injection in server-side RuntimeExecutors (was missing) - Support -p CLI flag in devStartupSequence for port configuration Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

…ates Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Stop re-exporting template/templates from database — consumers import directly from @lobechat/agent-templates. Keep types.ts re-exports for internal database code only. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

…pend pattern All providers that append to the system message now inherit from BaseSystemRoleProvider and only implement buildSystemRoleContent(). The base class handles find-or-create and join logic. Migrated providers: - EvalContextSystemInjector - BotPlatformContextInjector - SystemDateProvider - ToolSystemRoleProvider - HistorySummaryProvider - SkillContextProvider Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

…cted hook Add onInjected() callback to BaseSystemRoleProvider so subclasses can update pipeline metadata after successful injection. Also add raw-md plugin to context-engine vitest config for .md imports. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

…ionals in MessagesEngine All providers now accept an `enabled` config field. MessagesEngine pipeline is a flat array with no spread conditionals — each provider is always instantiated and uses `enabled` to skip internally. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Remove numbered prefixes, keep descriptive comments for each provider. Only phase headers use separator blocks. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

…arget Phase 1: History Truncation Phase 2: System Message Assembly (all BaseSystemRoleProvider) Phase 3: Context Injection (before first user message, BaseFirstUserContentProvider) Phase 4: User Message Augmentation (last user message injections) Phase 5: Message Transformation (flatten, template, variables) Phase 6: Content Processing & Cleanup (multimodal, tool calls, cleanup) Moved SkillContext, ToolSystemRole, HistorySummary from Phase 3 to Phase 2 since they append to system message, not user context. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

…jectors - AgentDocumentSystemInjector (Phase 2): before-system, system-append, system-replace - AgentDocumentContextInjector (Phase 3): before-first-user - AgentDocumentMessageInjector (Phase 4): after-first-user, context-end Shared utilities (filterByRules, formatDocument, sortByPriority) extracted to AgentDocumentInjector/shared.ts. Old monolithic injector removed. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

…jectors - AgentDocumentBeforeSystemInjector: prepends as separate system message (before-system) - AgentDocumentSystemAppendInjector: appends to system message (system-append) - AgentDocumentSystemReplaceInjector: replaces entire system message (system-replace) Each has distinct semantics and correct pipeline placement: - BeforeSystem → before SystemRoleInjector - SystemAppend → after HistorySummary (end of Phase 2) - SystemReplace → last in Phase 2 (destructive) Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

- Add AgentDocumentsManifest to defaultToolIds - Add hasAgentDocuments rule in server createServerAgentToolsEngine - Query agent documents in AiAgentService.execAgent to determine flag - Pattern matches KnowledgeBase auto-enable via enableChecker rules Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

lobehubbot · 2026-03-27T02:10:17Z

❤️ Great PR @arvinxx ❤️

The growth of project is inseparable from user feedback and contribution, thanks for your contribution! If you are interesting with the lobehub developer community, please join our discord and then dm @arvinxx or @canisminor1990. They will invite you to our private developer channel. We are talking about the lobe-chat development or sharing ai newsletter around the world.

sourcery-ai Bot reviewed Mar 26, 2026

View reviewed changes

vercel Bot deployed to Preview March 26, 2026 16:08 View deployment

chatgpt-codex-connector Bot reviewed Mar 26, 2026

View reviewed changes

vercel Bot had a problem deploying to Preview March 26, 2026 16:44 Failure

arvinxx and others added 3 commits March 27, 2026 00:52

🐛 fix: correct import statement for non-type exports from agent-templ…

eab8756

…ates Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

arvinxx force-pushed the refactor/skill-resolver branch from 8572407 to eab8756 Compare March 26, 2026 16:52

arvinxx and others added 2 commits March 27, 2026 01:01

📦 build: add @lobechat/agent-templates to root dependencies

b6ff5b0

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

vercel Bot deployed to Preview March 26, 2026 17:25 View deployment

arvinxx and others added 9 commits March 27, 2026 01:32

✨ feat: add enabled field to AgentDocumentInjector config

4a6804d

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

💄 style: clean up MessagesEngine pipeline comments

40bdeaa

Remove numbered prefixes, keep descriptive comments for each provider. Only phase headers use separator blocks. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

💄 style: split Phase 6 into Content Processing (6) and Cleanup (7)

3194479

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

vercel Bot deployed to Preview March 26, 2026 18:17 View deployment

arvinxx and others added 3 commits March 27, 2026 02:37

🔨 chore: add agent documents status to execAgent operation log

5bfbaed

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

update content

1a96ac9

vercel Bot had a problem deploying to Preview March 26, 2026 18:56 Failure

fix tests

adb3173

vercel Bot deployed to Preview March 27, 2026 01:31 View deployment

🐛 fix: add raw-md plugin to database vitest configs

a0d2e2f

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

vercel Bot deployed to Preview March 27, 2026 02:00 View deployment

arvinxx changed the title ~~♻️ refactor: implement SkillResolver to replace ad-hoc skill assembly~~ ♻️ refactor: implement SkillResolver, BaseSystemRoleProvider, and agent document injection pipeline Mar 27, 2026

arvinxx merged commit 4576059 into canary Mar 27, 2026
34 checks passed

arvinxx deleted the refactor/skill-resolver branch March 27, 2026 02:10

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

♻️ refactor: implement SkillResolver, BaseSystemRoleProvider, and agent document injection pipeline#13315

♻️ refactor: implement SkillResolver, BaseSystemRoleProvider, and agent document injection pipeline#13315
arvinxx merged 19 commits into
canaryfrom
refactor/skill-resolver

arvinxx commented Mar 26, 2026 •

edited

Loading

Uh oh!

vercel Bot commented Mar 26, 2026 •

edited

Loading

Uh oh!

sourcery-ai Bot left a comment

Uh oh!

lobehubbot commented Mar 26, 2026

Uh oh!

codecov Bot commented Mar 26, 2026 •

edited

Loading

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

chatgpt-codex-connector Bot Mar 26, 2026

Uh oh!

Uh oh!

lobehubbot commented Mar 27, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

arvinxx commented Mar 26, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Key Changes

SkillResolver

BaseSystemRoleProvider

Agent Document Injection Pipeline

MessagesEngine Pipeline Reorganization

@lobechat/agent-templates Package

Auto-enable Agent Documents Tool

Test plan

Uh oh!

vercel Bot commented Mar 26, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

sourcery-ai Bot left a comment

Choose a reason for hiding this comment

Uh oh!

lobehubbot commented Mar 26, 2026

Uh oh!

codecov Bot commented Mar 26, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector Bot Mar 26, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

lobehubbot commented Mar 27, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

arvinxx commented Mar 26, 2026 •

edited

Loading

vercel Bot commented Mar 26, 2026 •

edited

Loading

codecov Bot commented Mar 26, 2026 •

edited

Loading