Skip to main content
Glama

Server Configuration

Describes the environment variables required to run the server.

NameRequiredDescriptionDefault
PORTNoPort for the MCP gateway (default is profile-based: coco=18790, toto=18791, others deterministic in 20000-29999)
DREAM_MODENoDream mode: lite, standard, or pro
MCP_PROFILEYesProfile identifier for memory isolation (e.g., profile-a, coco, toto)
JINA_API_KEYYesJina AI API key for semantic search
DREAM_ENABLEDNoEnable Dream Runner (default false)
MCP_CLIENT_IDNoOAuth client ID for authentication
HERMES_ENABLEDNoEnable Hermes digest runner (default false)
MCP_CLIENT_SECRETNoOAuth client secret for authentication
SUPABASE_BASE_URLYesSupabase project URL from Project Settings → API
MCP_REQUIRE_BEARERNoRequire bearer token for authentication (default false)
MCP_TUNNEL_PROFILENoTunnel profile for Cloudflare (default coco)
DREAM_DIGEST_MCP_URLNoMCP URL for Dream digest
HERMES_DIGEST_ORIGINNoOrigin for Hermes digest
HERMES_DIGEST_MCP_URLNoMCP URL for Hermes digest
DREAM_ENABLE_REPO_SCANNoEnable repo scan (optional override)
DREAM_MCP_BEARER_TOKENNoBearer token for Dream MCP
HERMES_DIGEST_SOURCE_DIRNoSource directory for Hermes digest
DREAM_ENABLE_SOUL_CONTEXTNoEnable soul context (optional override)
MCP_TUNNEL_REQUIRE_BEARERNoRequire bearer for tunnel (default false)
SUPABASE_SERVICE_ROLE_KEYYesSupabase service role key (keep private)
DREAM_ENABLE_ISSUE_SIGNALSNoEnable issue signals (optional override)
HERMES_DIGEST_MCP_BEARER_TOKENNoBearer token for Hermes digest MCP

Capabilities

Features and capabilities supported by this server

CapabilityDetails
tools
{
  "listChanged": false
}

Tools

Functions exposed to the LLM to take actions

NameDescription
insert_memoryA

Store a short-term demo memory into demo.memories (ephemeral context that auto-expires). Use this to capture observations, decisions, session notes, or any context worth remembering temporarily. Memories last 7 days by default unless expires_at is set. For permanent knowledge, use memory_ingest instead.

list_memoriesA

List recent short-term memories from demo.memories in reverse chronological order — like a daily log or activity feed. Use this to review what was recently captured, check for duplicate memories before inserting, or browse recent activity. This is a time-ordered listing tool, not a search tool. For semantic search by meaning, use search_memories instead. For long-term knowledge retrieval, use recall. Returns memory body, source, creation time, and expiration status.

search_memoriesA

Semantic search across short-term memories in demo.memories using Jina embeddings. Finds memories by meaning, not just keywords — ask a natural language question and get the most relevant matches. Use this to recall past decisions, find related context, or check if something was already discussed recently.

reload_source_registryA

Reload the source registry cache from demo.source_registry. Use this after adding or removing entries to refresh which sources are enabled for insert/search filtering. Only effective when running in registry mode (MCP_SOURCE_MODE=registry).

recallA

Semantic recall from long-term memory (demo.marsvault_chunks) using Jina embeddings. Searches across promoted insights, digests, and archived knowledge using vector similarity. Use this to retrieve established patterns, past decisions, documented workflows, or any knowledge that was previously promoted to long-term storage. This is the primary tool for accessing institutional memory.

health_checkA

Run comprehensive demo memory system diagnostics. Returns chunk counts by type and visibility, identifies expiring short-term memories that need promotion, detects timeline coverage gaps, and finds conflicting or redundant long-term chunks. Use this regularly to maintain memory hygiene and before batch promotion.

session_bootA

Initialize a new demo session by loading identity, workflow context, and recent status from long-term memory. Also runs an expiry-focused health snapshot and records a heartbeat sign-in. Call this once at the start of every new conversation to restore continuity and awareness of pending work.

session_closeA

Save a demo session close summary before ending a conversation. The summary is stored as a short-term memory with 7-day retention, providing context for the next session boot. Include what was accomplished, what is still pending, and any important context for continuity.

dream_ingestA

Ingest a Hermes digest (periodic summary report) into demo.marsvault_chunks long-term memory. The content is automatically chunked into semantically meaningful segments and stored with vector embeddings for future recall. Use this for scheduled reports, daily digests, or any structured summary content that should be permanently available.

memory_ingestA

Promote content into permanent long-term memory (demo.marsvault_chunks). Automatically chunks the input text, generates vector embeddings, and stores each segment for semantic recall. Use this to preserve important insights, decisions, patterns, or knowledge that should survive beyond the current session. This is the primary path from ephemeral to permanent memory.

demote_memoryA

Deprecate a long-term memory chunk in demo.marsvault_chunks, marking it as outdated or superseded. The chunk is not deleted — it is flagged so it no longer appears in recall results. Use this when information becomes incorrect, irrelevant, or is replaced by newer knowledge. Optionally link to the replacement chunk for traceability.

soft_forgetA

Expire short-term memories in demo.memories early without permanently deleting them. Marked memories become invisible to search and list operations but remain in the database for audit purposes. Use this to clean up irrelevant or incorrect short-term memories before their natural expiration.

explain_memoryA

Trace the full provenance of a long-term memory chunk in demo.marsvault_chunks — where it came from, how it was created, and what source material it was derived from. Returns the original source memory, session, tool, and timeline information. Use this to understand why a piece of knowledge exists or to verify its reliability.

batch_promoteA

Automatically promote expiring short-term memories from demo.memories to permanent long-term storage (demo.marsvault_chunks). Scans for memories that will expire within the alert window, then chunks and ingests each one with vector embeddings. Use this to prevent valuable context from being lost when short-term memories expire. Supports dry-run mode to preview candidates before committing.

Prompts

Interactive templates invoked by user choice

NameDescription

No prompts

Resources

Contextual data attached and managed by the client

NameDescription

No resources

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/Marsmanleo/MarsNMe'

If you have feedback or need assistance with the MCP directory API, please join our Discord server