Open-Brain
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| ADMIN_KEY | No | Optional: admin key for POST /admin/provision. | |
| SUPABASE_URL | Yes | Your Supabase project URL (https://xxx.supabase.co). | |
| MCP_ACCESS_KEY | Yes | Admin access key for privileged endpoints (brain provisioning). | |
| OPENROUTER_API_KEY | Yes | OpenRouter API key for embeddings + LLM calls. | |
| TELEGRAM_BOT_TOKEN | No | Optional: Telegram bot token (capture path, not used by MCP server itself). | |
| TELEGRAM_SECRET_TOKEN | No | Optional: Telegram webhook secret. | |
| TELEGRAM_ALLOWED_CHAT_ID | No | Optional: Allowlisted Telegram chat ID. | |
| SUPABASE_SERVICE_ROLE_KEY | Yes | Supabase service role key — server-side, bypasses RLS. |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": false
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| search_thoughtsB | Hybrid search (BM25 + vector with RRF). Supports source filter, 1-hop graph expansion, quality gating. |
| deep_searchA | Multi-hop retrieval with graph traversal + LLM gap-filling sub-queries. Best for bridging topics and multi-session synthesis. |
| list_thoughtsB | Browse thoughts with salience-ordered filters (type, theme, topic, person, activity, days, min_quality). |
| thought_statsC | Aggregate counts with type/theme breakdown, top topics, people, and activity. |
| capture_thoughtB | Store a thought with auto-embedding, auto-linking, entity extraction, and semantic dedup merge. |
| delete_thoughtB | Permanently delete a thought (cascades connections). |
| update_thoughtB | Rewrite a thought's content (re-embeds, re-extracts metadata). |
| weekly_reviewC | LLM synthesis of recent themes, open loops, and suggested next steps. |
| migration_guideB | Import runbook for external platforms (Notion, Obsidian, Readwise, etc.). |
| get_connectionsC | Graph traversal from a thought via typed links (extends, contradicts, is-evidence-for, etc.). |
| analyzeC | Graph analysis: hubs, density, sources, co_occurrence, themes, synthesis_candidates. |
| dedup_reviewC | Duplicate candidate pairs with similarity zone histogram. |
| refresh_salienceB | Recompute salience scores (recency × access × links × merges × source). |
| list_entitiesB | Browse extracted entities (person, project, tool, organization) by frequency. |
| serendipity_digestC | Resurface forgotten high-quality thoughts across rediscovery/orphan/underrepresented/echo slots. |
| pipelineC | Pipeline monitoring: health, runs, merges. |
| review_staleC | Review flagged stale thoughts (list/approve/reject). |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 17 tools
Most tools have distinct purposes, though some overlap exists. For example, 'deep_search' and 'search_thoughts' both handle search but with different approaches (multi-hop vs. hybrid), which could cause confusion. However, descriptions clarify their specific use cases, and other tools like 'capture_thought' and 'delete_thought' are clearly differentiated.
Naming conventions are mixed, with some tools using verb_noun patterns (e.g., 'capture_thought', 'delete_thought') and others using noun-only or compound terms (e.g., 'analyze', 'deep_search', 'serendipity_digest'). This inconsistency reduces predictability, but the names remain generally readable and descriptive of their functions.
With 17 tools, the count is slightly high but reasonable for a knowledge management system covering analysis, storage, retrieval, and maintenance. It avoids being overwhelming (e.g., not 25+), and each tool appears to serve a specific role in the domain, though some consolidation might improve focus.
The tool set provides comprehensive coverage for a thought management system, including CRUD operations (capture, update, delete), retrieval (search, list), analysis (analyze, stats), maintenance (refresh, review), and synthesis (weekly_review). There are no obvious gaps; tools support the full lifecycle from input to insights.