Install Arca MCP

Pick your client, copy the command, done.

claude mcp add -e "ARCA_APP_AUTH_KEY=${ARCA_APP_AUTH_KEY}" -e "ARCA_GOOGLE_API_KEY=${ARCA_GOOGLE_API_KEY}" arca-mcp -- python -m app

Required:ARCA_APP_AUTH_KEYARCA_GOOGLE_API_KEY+ 2 optional

Environment Variables

Set these before running Arca MCP.

VariableDescriptionRequired

ARCA_APP_AUTH_KEYBearer token for MCP authenticationYes

ARCA_GOOGLE_API_KEYGoogle API key for Gemini embeddingsYes

ARCA_APP_HOSTServer bind addressNo

ARCA_APP_PORTServer portNo

Once configured, Arca MCP gives your AI agent access to:

memory/addStore content in memory with a vector embedding.

contentbucketconnected_nodesrelationship_types

memory/getRetrieve memories via semantic similarity search.

querybuckettop_k

memory/deleteDelete a specific memory by its UUID.

memory_id

memory/clearClear all memories in a bucket.

bucket

memory/list_bucketsList all buckets in the current namespace.

memory/connectCreate a directed edge between two memory nodes.

source_idtarget_idrelationship_type

memory/disconnectRemove one or all directed edges between two nodes.

source_idtarget_id

After setup, try these prompts with your AI agent:

→Store this project summary in my 'work' bucket for future reference.

→Search my memory for any notes related to the Q4 marketing strategy.

→List all available memory buckets to see how my data is organized.

→Connect the memory node for 'Project Alpha' to 'Project Beta' with the relationship 'depends_on'.

Conare · memory for coding agents

Save the docs, env vars, and workflow around Arca MCP in Conare so Claude Code, Codex, and Cursor remember it next time.