{"name":"verified-memory","displayName":"Verified Memory","description":"Stripe for AI memory — cryptographically verified, temporally valid, governed assertions with Proof Meter cost tracking and real-time delta sync. Zero-hop existence checks (0.01ms), sub-millisecond recall. Any agent, model, workflow, or human can write, read, and meter spend from a shared, trusted memory store. Works with Claude Code, Cursor, Windsurf, VS Code, and every MCP client. Shared verified memory across all your coding tools.","url":"https://mcp.verifiedstate.ai","version":"1.5.0","protocolVersion":"2024-11-05","endpoints":{"mcp":"https://mcp.verifiedstate.ai/mcp"},"tools":[{"name":"memory_ingest","description":"Ingest raw content into verified memory. Defaults to personal namespace. Use scope=\"team\" to store in team memory."},{"name":"memory_query","description":"Semantic search over verified memory. Returns ranked assertions with their active receipts."},{"name":"memory_verify","description":"Run the verification ladder on an assertion. Produces a signed receipt and updates assertion status."},{"name":"memory_health","description":"Get memory health stats for a namespace: assertion counts, verification ratio, stale/expired, open conflicts."},{"name":"meter_authorize","description":"Create a Proof Meter capability that grants an agent a spend budget. Returns a signed capability token."},{"name":"meter_spend","description":"Record a spend event against a Proof Meter capability. Returns a signed, hash-chained receipt."},{"name":"meter_budget","description":"Check remaining budget and spend summary for a Proof Meter capability."},{"name":"meter_settle","description":"Settle outstanding spend receipts into a Merkle-rooted settlement batch."},{"name":"meter_verify","description":"Cryptographically verify a Proof Meter spend receipt (hash, signature, chain, Merkle inclusion)."},{"name":"meter_receipts","description":"List and filter Proof Meter spend receipts for a namespace."},{"name":"session_save","description":"Save current session state to VerifiedState. Call this before ending any Claude Code session — captures what was built, decisions made, files changed, and next steps so the next session can pick up exactly here."},{"name":"session_load","description":"Load session context. If no project specified, loads the most recent session. Call at the start of every Claude Code session."},{"name":"session_list","description":"List all saved sessions/projects with their last activity time."},{"name":"session_delete","description":"Delete a saved session by project name. Retracts the session assertion (append-only — does not hard delete)."},{"name":"memory_subscribe","description":"Subscribe to namespace updates. After subscribing, subsequent tool calls will include any new assertions, conflicts, or changes since your last interaction."},{"name":"teams_list","description":"List all teams you belong to with their namespace IDs."},{"name":"session_end","description":"End the current session with an auto-generated summary. Optional — sessions auto-persist without this."},{"name":"memory_query_events","description":"Query the events log for a namespace. Returns state change events ordered by timestamp. Filter by entity, event type, source, or time range."},{"name":"working_state_get","description":"Retrieve the current working state for a session. Returns the live state block with goal, open loops, next action, and confidence."},{"name":"working_state_update","description":"Update (merge) the working state for a session. Top-level keys replace; arrays append. Creates a new block if none exists."},{"name":"alerts_get_unacknowledged","description":"Get active alerts (status=new or acknowledged, not resolved) for the current namespace. Use this at session start and whenever the user asks about system health, deployment status, or anything that may relate to active issues."},{"name":"alerts_acknowledge","description":"Mark an alert as acknowledged. Use when the user has been informed of the alert. Does not resolve the alert — only marks that the user is aware of it."},{"name":"alerts_resolve","description":"Mark an alert as resolved. Use only when the underlying issue has actually been fixed."}],"capabilities":{"tools":true,"streaming":false}}