Skip to main content
Glama
claudedowling

librechat-mnemonic

by claudedowling
OverviewSchemaRelated ServersScoreDiscussions
TypeScript
Remote

Server Configuration

Describes the environment variables required to run the server.

NameRequiredDescriptionDefault
HOSTNoListen address0.0.0.0
PORTNoListen port8710
MCP_PATHNoWhere to serve the MCP endpoint/mcp
LOG_LEVELNopino levelinfo
UPSTREAMSYesJSON array of upstream provider configs
OLLAMA_URLNoURL of the Ollama instance for embeddings
DISABLE_GITNoDisable git in mnemonic
EMBED_MODELNoEmbedding model name
MCP_ENABLEDNoServe the MCP endpointtrue
MNEMONIC_TAGNoTag added to everything this service writeslibrechat
MNEMONIC_URLNoRequired when MNEMONIC_MODE=remote
EXTRACT_MODELNoModel name for extraction
MNEMONIC_MODENospawn runs the bundled mnemonic over stdio; remote connects to a streamable-http instancespawn
EMBED_PROVIDERNoEmbedding provider for mnemonic
GEMINI_API_KEYNoGemini API key for embeddings
OPENAI_API_KEYNoOpenAI API key for embeddings
EXTRACT_API_KEYNoBearer token for extraction model
EXTRACT_BASE_URLNoOpenAI-compatible base URL, including /v1
MNEMONIC_COMMANDNoExecutable used in spawn mode
MNEMONIC_HEADERSNoJSON headers for the remote instance, e.g. auth{}
MEMORY_WRITE_MODENollm extracts automatically, explicit only on 'remember that …', off disables writingllm
EXTRACT_TIMEOUT_MSNoExtraction is detached; a timeout drops the write, never the reply30000
LIBRECHAT_MONGO_DBNoOverride the database name
MEMORY_PROJECTLESSNooff disables memory entirely in chats not assigned to a projectglobal
LIBRECHAT_MONGO_URIYesConnection string for LibreChat's MongoDB
MEMORY_MAX_PER_TURNNoCap on memories written per exchange3
MNEMONIC_TIMEOUT_MSNoPer-call timeout20000
MNEMONIC_VAULT_PATHNoVault directory, passed through as VAULT_PATH/vault
MNEMONIC_WRITE_SCOPENoglobal stores in the main vault stamped with the project; project writes a project vaultglobal
LIBRECHAT_USER_HEADERNoHeader carrying {{LIBRECHAT_USER_ID}}x-librechat-user-id
MEMORY_COMMAND_PREFIXNoChange if it clashes with something/memory
MEMORY_RECALL_ENABLEDNoSet false to write memories without injecting themtrue
MNEMONIC_PROJECT_ROOTNoWhere per-project directories live/projects
MNEMONIC_RECALL_LIMITNoMemories retrieved per turn6
MNEMONIC_RECALL_SCOPENoproject isolates, all boosts the current project, global returns the whole main vaultall
MEMORY_DEFAULT_ENABLEDNoWhether memory is on for chats with no explicit settingtrue
MEMORY_DEDUPE_THRESHOLDNoRecall score above which a candidate is treated as already known0.82
MNEMONIC_MIN_SIMILARITYNoSimilarity floor passed to recall0.3
MEMORY_MAX_CONTEXT_CHARSNoBudget for the injected block4000
MEMORY_QUERY_MESSAGE_COUNTNoUser turns used to build the recall query3
LIBRECHAT_CONVERSATION_HEADERNoHeader carrying {{LIBRECHAT_BODY_CONVERSATIONID}}x-librechat-conversation-id

Capabilities

Server capabilities have not been inspected yet.

Tools

Functions exposed to the LLM to take actions

NameDescription

No tools

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/claudedowling/librechat-mnemonic'

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