Conversation
…aster orchestrator
…stats, and IPC handlers
…cture and flow telemetry
…cade subsystem and flow telemetry
…t unlinked file references into clickable file links
…URIs in TemplateEngine
Add dedicated SQLite TelemetryDB, disaggregate token details, record real tool execution timestamps, and surface raw provider errors.
…TelemetryDB Direct telemetry log retrieval to TelemetryDB and auto-clear thread telemetry on conversation deletion.
…s, and cleanup tools Add Telemetry DB status row, explicit caller badges (System vs LLM), token breakdown chips, and target log cleanup controls.
- Route task queries to workspace_task_summary with >0.80 confidence - Restrict graph search for task queries unless explicitly requested - Add fallback sequence for empty task queries & filter evidence < 0.25 - Log plannerDecision and retrievalQuality telemetry metrics - Cache static prompt core and add multi-tier provider fallback - Add pipelineRegression test suite and update VitePress docs
…lity - Fix ReferenceError in ContextOrchestrator for outputStr & s2SpanId - Add conversational_followup router for context reuse queries - Enforce min relevance threshold (0.10) and early exit in _deriveNextSteps - Emit Stage 4 and estimated tokens on LLM execution errors - Synchronize planner confidence and expose prompt breakdown metrics - Update Groq provider model mapping to supported models - Add unit & integration regression tests (134/134 passing)
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Source of Truth: Set .md files as authoritative for personas (resources/prompts/personas/.md and appData/personas/.md).
Lookup Order: PersonaManager.getPersona() inspects .md files via PromptLoader before falling back to DB.
Custom Personas: createCustomPersona() writes formatted .md files on creation.
Database Refactor: Stripped prompt column from PersonaDB table. SQLite stores index metadata only; _hydratePersonaRow() loads frontmatter/body from .md files at runtime.
Auto Migration: Added automatic ALTER TABLE personas DROP COLUMN prompt migration in PersonaDB._createTables() for backwards compatibility with existing on-disk databases.
Default ID Alignment: Aligned default persona ID to 'general' across ConversationStore and ContextEngine.
Static Core Policies: Added response-policy.md, conversation-policy.md, and formatting-policy.md to PromptPipeline._getStaticCore().
Tool Calling Discipline: Added tool parameter pre-extraction rules to planning-policy.md.
Evidence Sanitation & Truncation: Stripped capability error strings from retrieved evidence and added newline-aware evidence budget truncation.
Dev Hot-Reloading: Added clearPromptCache() to PromptPipeline and PromptLoader.
Stream Fallback Duplication Fix: Emitted onChunk({ type: 'replace', ... }) during fallback retries to prevent duplicate UI text rendering.
Groq Compatibility: Omitted toolChoice (undefined) when tool maps are empty to prevent Groq HTTP 400 errors on toolChoice: 'none'.
Conversational Task Intent Routing: IntentAnalyzer infers needsTasks on conversational follow-ups (e.g. "Which task first") when prior turns discussed tasks. Targeted task questions bypass deterministic summary formatting to allow LLM synthesis.
Grounding Citation Link Cleanup: Added regex replacement in GroundingEngine.cleanMarkdownLinkEscaping() to normalize nested double links ([label](file://...)).