eve-mcp

One Memory.

CommunityOnlineNo sign-inGlobalFreeRead-only

What it can do

  • Memory Store: Store a piece of knowledge into memory. Args: text: The exact text content to remember (required for semantic/episodic/auto, optional for preference store) source: The source origin of t
  • Memory Search: Search memory using a natural language query. Args: query: The natural language question or keywords source_agent: Agent/tool calling this API (audit trail). limit: Max number of result
  • Memory Get Preferences: Get stored preferences for a context. Use at session start to load behavioral preferences. Args: source_agent: Agent/tool calling this API (audit trail). Default "unknown". con

What data it sees

Do you need an account

No: the server works without sign-in

One Memory. Every Agent. Eve gives teams one managed memory layer across Claude, Gemini, MCP clients, and custom tooling. The product centers on tenant-safe operations, operator controls, and a visible four-store architecture instead of vague “AI memory” promises.

How to connect

How to connect eve-mcp to Claude

The server address to paste into Claude settings:

https://eve-mcp--eve.run.tools

  1. Open Claude (claude.ai or the desktop app).
  2. Go to Settings → Connectors.
  3. Click Add custom connector.
  4. Paste the server address copied below into Remote MCP server URL and click Add.
  5. In a chat, click + → Connectors and switch the new connector on.

Custom connectors are available on Free, Pro, Max, Team and Enterprise plans (Free is limited to one). On Team and Enterprise an organization Owner adds the connector first under Organization settings → Connectors.

Authorization

After you click Add, a sign-in window for the service opens. Sign in with your own account and approve access. Claude never sees your password.

How to connect eve-mcp to ChatGPT

The server address to paste into ChatGPT settings:

https://eve-mcp--eve.run.tools

  1. Open ChatGPT in a browser (chatgpt.com). A Plus, Pro, Business, Enterprise or Edu plan is required.
  2. Turn on developer mode once: Settings → Apps → Advanced settings → Developer mode.
  3. Open Settings → Connectors and click Create.
  4. Fill in the form: Name (anything), Description (one line about what the service does), MCP server URL (copy it below).
  5. Under Authentication choose OAuth if the service requires sign-in, otherwise None. Click Create.
  6. In a new chat open + → Apps/Connectors and enable the connector.

OpenAI has renamed this section before (Connectors → Apps/Plugins). If the label differs, search settings for "developer mode". On Business/Enterprise workspaces an admin must allow custom connectors first.

Authorization

On first use ChatGPT opens the service's sign-in window. Sign in and approve access.

How to connect eve-mcp to Cursor

The server address to paste into Cursor settings:

https://eve-mcp--eve.run.tools

Fastest: click Open in Cursor below and confirm the prompt.

Manually:

  1. In Cursor open Settings → Cursor Settings → MCP and click Add new global MCP server.
  2. Paste the JSON copied below into ~/.cursor/mcp.json (per project: .cursor/mcp.json in the repo root).
  3. Save the file. The server appears in the MCP list; authorize it there if asked.

Authorization

If the service needs sign-in, an authorize button appears next to the server in the MCP list.

Server tool list (14)

Raw names from tools/list. Only developers need these.

memory_storeStore a piece of knowledge into memory. Args: text: The exact text content to remember (required for semantic/episodic/auto, optional for preference store) source: The source origin of the memory (required for semantic/episodic/auto, optional for preference store) source_agent: Agent/tool calling this API (audit trail). Default "unknown". store: Target memory store — "auto" (default), "semantic", "episodic", "preference", or "learned_rules" context: Context scope (personal, naya, es) — default "personal" metadata: Optional dictionary with extra context entity_refs: Optional list of related entities category: Category for preference store (e.g. 'architecture') key: Key for preference store value: Value for preference store confidence: Confidence score for preference store (0.0–1.0, default 1.0)
memory_searchSearch memory using a natural language query. Args: query: The natural language question or keywords source_agent: Agent/tool calling this API (audit trail). limit: Max number of results (default 10) min_similarity: Minimum confidence threshold (0.0-1.0) store: Memory store to search — "semantic" (default), "episodic", "learned_rules", "preference", or "all" merge_policy: Cross-store merge policy when store="all": interleave, weighted_rrf, or calibrated_fusion. context: Context scope (personal, naya, es). Defaults to "all", meaning omitted context is unscoped cross-context search. Use explicit "personal" to keep preference reads scoped. visibility: Visibility scope ("PERSONAL" or "SHARED"). Defaults to PERSONAL. category: Filter by memory category (e.g. "architecture", "security"). Use "all" or omit to search all categories.
memory_get_preferencesGet stored preferences for a context. Use at session start to load behavioral preferences. Args: source_agent: Agent/tool calling this API (audit trail). Default "unknown". context: Context scope (personal, naya, es) category: Optional category filter (e.g., 'architecture', 'communication_style') scope: Preference scope (user or tenant)
memory_pre_compactionSave a list of critical facts as episodic entries before context compaction. Use when you have specific facts to preserve verbatim. Blocking — waits for write confirmation. Args: session_id: Current session UUID critical_facts: List of critical facts that must survive compaction source_agent: Agent/tool calling this API (audit trail). Default "unknown". context: Context scope (personal, naya, es) tool_name: Source tool name
memory_session_startLog the beginning of a conversation or work session. Args: summary: Brief description of the session goal source_agent: Agent/tool calling this API (audit trail). Default "unknown". session_id: Optional UUID. Auto-generated if omitted. context: Context scope (personal, naya, es) — default "personal" details: Optional dictionary of context
memory_session_endLog the end of a session. Automatically triggers the AI learning and reflexion pipeline. Args: summary: Brief description of the outcome (success/failure) source_agent: Agent/tool calling this API (audit trail). Default "unknown". session_id: Optional UUID matching the session_start context: Context scope (personal, naya, es) — default "personal" details: Optional dictionary of context status: Session outcome — "success", "failure", or "unknown" (optional)
memory_extractExtract memorable facts, preferences, and events from transcript text. Batch-processes transcript using AI classification. Optionally auto-stores extracted items with dedup checking. Args: transcript: Raw transcript text to extract memories from source: Source identifier (e.g., claude_code, gemini_cli) source_agent: Agent/tool calling this API (audit trail). Default "unknown". session_id: Optional session UUID for provenance tracking auto_store: If true, automatically store extracted items (with dedup) context: Context scope (personal, naya, es) -- default "personal" min_importance: Only extract items at or above this importance (1-10, default 4) use_extraction: If true, use atomic fact extraction (produces SPO fields) instead of classifier
memory_ingestStart a batch ingestion job from a conversation export file. Triggers asynchronous processing of an export file (ChatGPT, Claude, Gemini, etc.) through the ingestion pipeline. Args: file_path: Absolute path to the export file source_agent: Agent calling this tool (audit trail) source_type_hint: Override format auto-detection (chatgpt|claude_code|claude_desktop|gemini) source_priority: Priority for stored claims (1=bulk, 2=inferred, 4=tool, 5=explicit) batch_size: Number of turns per extraction batch (default 100) dry_run: If True, count facts without writing to memory predicate_allowlist: If set, only store claims with predicates in this list Returns: JSON with job_id and status=queued
memory_ingest_statusPoll the status of a running ingestion job. Args: job_id: The job ID returned from memory_ingest source_agent: Agent calling this tool (audit trail) Returns: JSON with job_id, status, total, processed, failed, errors
memory_feedbackRecord consumer feedback on a retrieved memory chunk. Args: chunk_id: UUID of the semantic chunk being rated outcome: Feedback outcome — "helpful", "not_helpful", or "correction" source_agent: Agent/tool calling this API (audit trail). Default "unknown". context: Optional query or context text for this retrieval correction_text: Corrected text when outcome="correction"
memory_forgetSoft-delete (retract) a semantic memory chunk by its chunk_id. Sets retracted_at and valid_to so the chunk is excluded from future searches but preserved for audit purposes. Args: chunk_id: UUID of the semantic chunk to forget source_agent: Agent/tool calling this API (audit trail). Default "unknown".
memory_updateUpdate an existing memory. If text changes, facts are re-extracted. Only provided fields are updated; None fields keep existing values. Args: chunk_id: UUID of the semantic chunk to update text: New text content (triggers re-embedding and claim re-extraction if changed) metadata: New metadata dictionary (replaces existing metadata entirely) entity_refs: New entity references list (replaces existing refs entirely) source_agent: Agent/tool calling this API (audit trail). Default "unknown".
memory_pre_compactDistill raw conversation messages into memories using AI extraction before context compaction. Use when you have raw conversation text to process. Extracts decisions, preferences, and learned patterns via Gemini Flash. Applies per-session (max 3) and per-agent-hour (max 30) rate limits. Skips if the same session+content hash was already processed. Args: session_id: Current session UUID messages: List of {role, content} dicts (user/assistant/system) source_agent: Agent/tool calling this API (audit trail). Default "unknown". context: Context scope (personal, naya, es) — default "personal"
memory_ingest_urlIngest a web URL into memory. Fetches the web page, extracts clean content, chunks it, and stores each chunk with trust_level: "external" metadata. Full SSRF protection applied. Args: url: HTTPS URL to ingest (HTTP/other schemes blocked for security) source_agent: Agent calling this tool (audit trail) context: Context scope (personal, naya, es) — default "personal" importance: Override importance score (1-10) for all extracted chunks metadata: Optional extra metadata to attach to each chunk entity_refs: Optional list of related entities Returns: JSON with status, chunk_ids, title, word_count, url