feat: Add semantic knowledge graph with sqlite-vec and MCP research infrastructure #823
Conversation
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
…ages Expand the Homebrew formula generation skill from 3 fully-supported languages (Go, Rust, Python) to 24 language enum values with full template pipeline coverage, plus 7 Tier 2 and 7 Tier 3 mapped languages. - Add 15 new install-* schema definitions and allOf dispatch entries - Create 17 mustache partials for language-specific install blocks - Add 25 reference docs with research patterns and common issues - Add 21 example formulas showing expected pipeline output - Add 36 test fixtures and 35 test cases (41 total tests, all passing) - Create meta-checklist for adding future language support - Add build_system_is_* flag preprocessing in formula.helper.ts Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Add human-facing README with pipeline overview, language tiers, and directory layout. Remove TODO.md and settings.json (superseded by repo reorganization). Remove unused add-formula justfile recipe. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Add Homebrew's official MCP server (brew mcp-server) and @gongrzhe/server-json-mcp for JSONPath queries. Replace placeholder configs that had hardcoded /path/to paths. Add jq and node to brewfile as runtime dependencies for the MCP servers. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
…t format Convert flat URL lists to categorized maps with structured entries containing url, desc, and name fields across all 48 MCP registry files. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Add justfile recipe with [confirm] to clone, build, and merge three clone-only MCP servers (json-validate, jsondiffpatch, json-mcp-server) into the plugin's .mcp.json config. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Implement a modular plugin planning and scaffolding pipeline: - Orchestrator `/create-plugin` chains brainstorm → research → roadmap → scaffold → issues - 4 specialized agents: skill-researcher, mcp-researcher, component-assessor, planner - Interactive brainstorming across 7 component categories (skills, commands, agents, styles, hooks, MCP, LSP) - Parallel research using haiku agents, synthesis via sonnet assessor - Roadmap with P0/P1/P2 prioritization and dependency mapping - Scaffold from .template/ with generated plugin.json, sources, mcp config - Beads issue generation with labels and blocking relationships Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Introduce a local MCP server registry cache backed by SQLite with FTS5 full-text search. Split the monolithic plugin-mcp-researcher into three specialized agents (scanner, profiler, researcher) and add a /find-mcp-servers slash command for cache-first server discovery. - Add .data/mcp/registry-cache.sql with schema, triggers, and 39 seeded code-analysis servers (8 fully profiled with tools) - Add mcp-registry-scanner agent (haiku) for lightweight registry discovery - Add mcp-server-profiler agent (sonnet) for deep server enrichment - Rewrite plugin-mcp-researcher for cache-first flow - Add /find-mcp-servers command with FTS5 search, weighted ranking, and interactive next steps - Add justfile recipes for cache management (load, dump, stats, search) - Add cclint config and lint-context.sh for agent/command validation Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Expand the mcp-research stub into a full skill with architecture docs, three discovery/profiling/seeding workflows, web scraping fallback chain, SQL patterns, and troubleshooting guide. Restructure registries.yaml into prioritized tiers for the scanner agent. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Add semantic knowledge graph infrastructure using sqlite-vec for vector storage and Ollama for local embedding generation. Components: - Database schema with entities, chunks, embeddings, and similarity cache - Embedder module with Ollama (primary) and sentence-transformers (fallback) - Markdown chunker with hierarchical levels (file/section/paragraph) - Ingestion pipeline CLI (embed.py) with search, similarity commands - File watcher for automatic re-embedding on changes - Justfile targets: kg-init, kg-ingest, kg-search, kg-similar, etc. Architecture decisions documented in docs/src/adr/: - ADR-001: SQLite as primary store - ADR-002: sqlite-vec for vector index - ADR-003: Ollama embedding models - ADR-004: Recursive CTEs for graph queries - ADR-005: Watchdog file watching - ADR-006: SQL dump for version control - ADR-007: Future graph layer considerations Requires Python 3.14+ (Homebrew) for sqlite extension loading. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Signed-off-by: aRustyDev <36318507+aRustyDev@users.noreply.github.com>
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
Key Components
Knowledge Graph (sqlite-vec)
kg-init,kg-ingest,kg-search,kg-similar, etc.MCP Research Infrastructure
Architecture Decisions
Documented in
docs/src/adr/:Test plan
just initto install dependencies and initialize databasejust kg-statsshows empty databasejust kg-ingestto embed context filesjust kg-search "code analysis"returns relevant resultsjust kg-similar <entity-slug>returns similar entities🤖 Generated with Claude Code