Store a memory observation in Novyx.

Args: observation: The memory content to store. tags: Optional list of tags for categorization. importance: Importance score 1-10 (default 5). context: Optional context string. ttl_seconds: Optional time-to-live in seconds. Memory auto-expires after this duration.

Returns: JSON string with the stored memory UUID and details.

Create a reviewable draft without writing to canonical memory.

Args: observation: Proposed memory content. tags: Optional list of tags for categorization. importance: Importance score 1-10 (default 5). context: Optional context string. confidence: Confidence score 0-1 (default 1.0).

Returns: JSON string with the draft, review summary, and similar memories.

List current memory drafts.

Args: status: Optional filter by status (draft, merged, rejected).

Returns: JSON string with draft records.

Show a field-level diff for a memory draft.

Args: draft_id: Draft identifier returned by draft_memory. compare_to: Optional existing memory UUID to compare against.

Returns: JSON string with changed fields and a merge recommendation.

Merge a reviewed draft into canonical memory.

Args: draft_id: Draft identifier returned by draft_memory. supersede_memory_id: Optional older memory to mark as superseded.

Returns: JSON string describing the merged memory.

Get grouped review information for a branch/session of drafts.

Merge all open drafts in a branch/session.

Reject a draft without creating a memory.

Args: draft_id: Draft identifier returned by draft_memory. reason: Optional reason for rejection.

Returns: JSON string with the rejected draft state.

Reject all open drafts in a branch/session.

Search memories semantically using natural language.

Args: query: Natural language search query. limit: Maximum number of results to return (default 5). tags: Optional tag filter. min_score: Minimum similarity score 0-1 (default 0). explain: If true, include scoring breakdown for each result showing why it scored the way it did.

Returns: JSON string with matching memories and their scores.

Delete a memory by its UUID.

Args: memory_id: The UUID of the memory to delete.

Returns: JSON string indicating success or failure.

List stored memories with optional tag filtering.

Args: limit: Maximum number of memories to return (default 50). tags: Optional tag filter to narrow results.

Returns: JSON string with the list of memories.

Get memory statistics for the current account.

Returns: JSON string with total memories, average importance, tag distribution, etc.

Rollback memory to a point in time.

Supports ISO timestamps (e.g. '2026-01-15T10:00:00Z') and relative time expressions (e.g. '2 hours ago').

Args: target: ISO timestamp or relative time expression. dry_run: If true, preview changes without applying them.

Returns: JSON string with rollback results or preview.

Get the audit trail of memory operations.

Args: limit: Maximum number of audit entries to return (default 20). operation: Optional filter by operation type (CREATE, UPDATE, DELETE, ROLLBACK).

Returns: JSON string with audit trail entries.

Create a directed link between two memories.

Args: source_id: UUID of the source memory. target_id: UUID of the target memory. relation: Type of relationship (default 'related').

Returns: JSON string with the created link details.

Add a knowledge graph triple (subject -> predicate -> object).

Entities are auto-created by name if they don't exist.

Args: subject: The subject entity name. predicate: The relationship predicate. object_name: The object entity name.

Returns: JSON string with the created triple details.

Query knowledge graph triples with optional filters.

At least one filter should be provided. Returns all matching triples.

Args: subject: Filter by subject entity name. predicate: Filter by relationship predicate. object_name: Filter by object entity name.

Returns: JSON string with matching triples.

Create a shared context space for multi-agent collaboration.

Spaces let multiple agents share memories with fine-grained permissions. The creator is the owner and can grant read/write access to other agents or tenants.

Args: name: Name for the space. description: Optional description of the space's purpose. allowed_agents: Optional list of agent IDs that can access this space. tags: Optional tags for the space.

Returns: JSON string with the created space details including space_id.

List all context spaces you can access.

Returns spaces you own and spaces shared with you.

Returns: JSON string with list of spaces and their memory counts.

List or search memories within a context space.

Args: space_id: The space ID to query. query: Optional search query to filter memories semantically. limit: Maximum number of memories to return (default 50).

Returns: JSON string with memories in the space.

Update a context space (owner only).

Args: space_id: The space ID to update. name: New name for the space. description: New description. allowed_agents: Updated list of allowed agent IDs. tags: Updated tags.

Returns: JSON string with the updated space details.

Delete a context space (owner only).

Args: space_id: The space ID to delete.

Returns: JSON string indicating success.

Report that cross-tenant space sharing is currently unavailable.

Novyx Cloud share tokens currently point to a hosted /join redemption route that is not available. Do not issue unredeemable tokens from MCP.

Args: tag: The tag or space tag to share. email: Email address of the recipient. permission: Access level — 'read' or 'write' (default 'read').

Returns: JSON string explaining that no cross-tenant share was created.

Get the full timeline of memory operations. The tape you scrub through.

Shows every create, update, delete, and rollback event with timestamps. Requires Pro tier or Novyx Cloud.

Args: since: Start of time range (ISO timestamp). until: End of time range (ISO timestamp). operations: Comma-separated filter: create, update, delete, rollback. limit: Maximum entries to return (default 100).

Returns: JSON string with timeline entries.

Reconstruct memory state at a specific point in time.

Returns all memories and their link graph as they existed at timestamp T. Requires Pro tier or Novyx Cloud.

Args: at: ISO timestamp to reconstruct state at. limit: Maximum memories to return (default 500).

Returns: JSON string with memory snapshot and edges.

Full biography of a single memory.

Shows creation, every update, every recall, every link, and deletion. Use this to understand why a memory exists and how it evolved. Requires Pro tier or Novyx Cloud.

Args: memory_id: UUID of the memory to inspect.

Returns: JSON string with the memory's full lifecycle events.

Diff memory state between two timestamps.

Shows what was added, removed, and modified in a time range. Requires Pro tier or Novyx Cloud.

Args: start: Start timestamp (ISO). end: End timestamp (ISO).

Returns: JSON string with added, removed, and modified memories.

Get Cortex autonomous intelligence status.

Shows whether Cortex is enabled, last run time, and consolidation/reinforcement stats. Requires Pro tier or Novyx Cloud.

Returns: JSON string with Cortex status and last run info.

Manually trigger a Cortex cycle.

Runs consolidation (merge duplicate memories) and reinforcement (boost frequently recalled memories, decay forgotten ones). Normally runs automatically every 6 hours. Requires Pro tier or Novyx Cloud.

Returns: JSON string with cycle results (consolidated, reinforced counts).

List auto-generated memory insights.

Cortex detects patterns across your memories and generates insights automatically. Requires Enterprise tier or Novyx Cloud.

Args: limit: Maximum insights to return (default 20).

Returns: JSON string with generated insights.

Get the current Cortex configuration.

Shows consolidation thresholds, reinforcement decay rates, and cycle schedule. Requires Pro tier or Novyx Cloud.

Returns: JSON string with Cortex configuration.

Mark a memory as superseded by a newer one.

The old memory remains in the system for audit purposes but is flagged as superseded. Use when information has been updated or corrected.

Args: old_memory_id: UUID of the memory being replaced. new_memory_id: UUID of the replacement memory.

Returns: JSON string confirming the supersede operation.

Remove a link between two memories.

Args: source_id: UUID of the source memory. target_id: UUID of the target memory.

Returns: JSON string confirming link removal.

Get all links for a memory.

Shows both incoming and outgoing connections in the memory graph.

Args: memory_id: UUID of the memory. relation: Optional relation type filter (e.g. "related", "causes", "supports").

Returns: JSON string with links list.

Query the memory graph edges with filters.

Browse the relationship graph between memories. Use to understand how memories are connected.

Args: memory_id: Optional UUID to filter edges involving a specific memory. relation: Optional relation type filter. direction: "outgoing", "incoming", or "both" (default). limit: Maximum edges to return (default 100).

Returns: JSON string with matching edges.

Delete a knowledge graph triple.

Removes a subject-predicate-object relationship from the knowledge graph.

Args: triple_id: ID of the triple to delete.

Returns: JSON string confirming deletion.

List knowledge graph entities.

Entities are the nodes in your knowledge graph — subjects and objects of triples.

Args: limit: Maximum entities to return (default 100). offset: Pagination offset. entity_type: Optional type filter.

Returns: JSON string with entities list.

Get a knowledge graph entity and its associated triples.

Args: entity_id: ID of the entity.

Returns: JSON string with entity details and related triples.

Delete a knowledge graph entity and all its triples.

Removes the entity node and every triple where it appears as subject or object.

Args: entity_id: ID of the entity to delete.

Returns: JSON string confirming deletion.

Preview what a rollback would do without executing it.

Shows which operations would be undone if you rolled back to the target timestamp. Always use this before an actual rollback.

Args: target: ISO timestamp or relative expression (e.g. "2 hours ago").

Returns: JSON string with preview of operations to undo.

List past rollback operations.

Shows when rollbacks were performed, what they targeted, and how many operations were undone.

Args: limit: Maximum rollback events to return (default 50).

Returns: JSON string with rollback history.

Verify the integrity of the audit trail.

Checks the cryptographic hash chain (cloud) or entry consistency (local) to confirm no audit entries have been tampered with.

Returns: JSON string with verification result.

Create an execution trace to track a multi-step agent workflow.

Start a trace before a complex operation, add steps as you go, then complete it. Traces provide a full audit of agent reasoning.

Args: name: Name describing this trace (e.g. "research-and-summarize"). metadata: Optional JSON string with additional metadata.

Returns: JSON string with trace_id and status.

Add a step to an execution trace.

Record each significant action during a traced workflow. Include input and output data to make the trace useful for debugging.

Args: trace_id: ID of the active trace. step_name: Name of this step (e.g. "search-memories", "call-api"). input_data: Optional JSON string with step input. output_data: Optional JSON string with step output.

Returns: JSON string with step_id and status.

Mark an execution trace as complete.

Finalizes the trace. After completion, the trace and all its steps are immutable and available for audit.

Args: trace_id: ID of the trace to complete.

Returns: JSON string with completion status and step count.

Verify an execution trace's integrity.

Confirms all steps are present and the trace hasn't been tampered with.

Args: trace_id: ID of the trace to verify.

Returns: JSON string with verification result.

Run a memory health evaluation.

Scores your memory quality on a 0-1 scale based on staleness, conflicts, and superseded memories. Optionally acts as a CI gate with min_score.

Args: min_score: Optional minimum score threshold. If set, returns pass/fail gate result.

Returns: JSON string with score, breakdown, and optional gate result.

CI gate — pass or fail based on memory health score.

Use in CI/CD pipelines to block deploys when memory quality degrades.

Args: min_score: Minimum acceptable score (0.0 to 1.0).

Returns: JSON string with gate result (passed/failed) and score.

List past memory evaluation runs.

Track how memory quality has changed over time.

Args: limit: Maximum evaluations to return (default 50).

Returns: JSON string with evaluation history.

Detect memory drift over a time period.

Shows how many memories were created, deleted, and updated during the specified period. Useful for monitoring memory churn.

Args: days: Number of days to look back (default 7).

Returns: JSON string with drift metrics.

Get the full history of a single memory.

Shows every operation that affected this memory — creation, updates, links, supersedes — in chronological order.

Args: memory_id: UUID of the memory.

Returns: JSON string with chronological event list.

Time-travel recall — what would search have returned at a past timestamp?

Reconstructs the memory state at a historical point and runs a semantic search against it. Powerful for understanding how agent context evolved. Requires Pro tier or Novyx Cloud.

Args: query: Search query. at: ISO timestamp to search at. limit: Maximum results (default 5).

Returns: JSON string with historical search results.

Detect memory drift between two timestamps.

Compares memory state at two points in time and shows what changed. Requires Pro tier or Novyx Cloud.

Args: from_ts: Start ISO timestamp. to_ts: End ISO timestamp.

Returns: JSON string with drift analysis.

Get a snapshot of your current memory context.

Returns recent memories, stats, and audit activity — a quick overview of what your agent knows right now.

Returns: JSON string with context snapshot.

Get a full dashboard overview.

Combines stats, spaces, and recent activity into a single response. Use for periodic status checks or reporting.

Returns: JSON string with dashboard data.

Accept a shared context invitation.

When another user shares a context space with you, use this to accept and gain access to their memories. Requires Novyx Cloud.

Args: token: The sharing token from the invitation.

Returns: JSON string with acceptance result.

List all shared contexts you have access to.

Shows contexts shared with you and contexts you've shared with others. Requires Novyx Cloud.

Returns: JSON string with shared contexts list.

Revoke a shared context invitation.

Removes access for the recipient. Requires Novyx Cloud.

Args: token: The sharing token to revoke.

Returns: JSON string confirming revocation.

List pending Control approval requests.

Shows actions submitted by agents that require human approval before execution. Requires Novyx Control to be configured (NOVYX_CONTROL_URL + NOVYX_CONTROL_API_KEY).

Args: limit: Maximum approvals to return (default 20).

Returns: JSON string with pending approvals list.

Approve a pending agent action in Novyx Control.

Approves the action and triggers its execution against the target connector (GitHub, Slack, Linear, PagerDuty, or HTTP).

Args: approval_id: The approval ID (e.g. apr_act_xxx). approver_id: Your operator ID (e.g. usr_operator_default). reason: Optional reason for approval.

Returns: JSON string with the executed action details.

Check the current Control policy profile.

Shows which connectors require approval, risk tier rules, and auto-approve settings.

Args: connector: Optional connector to check (e.g. github, slack). environment: Environment to check against (default: production).

Returns: JSON string with policy profile and whether the connector requires approval.

List all active Control policies (built-in + custom).

Shows which policies are enforced on action submissions, including tenant-defined custom policies created via create_policy.

Args: enabled_only: If True (default), only show enabled policies.

Returns: JSON string with list of policies, their source (builtin/custom), and status.

Create a custom governance policy for your agent's actions.

Define rules with regex patterns and severity levels. Policies are evaluated alongside built-in policies on every action submission. Requires Starter plan.

Args: name: Policy name (alphanumeric, underscores, hyphens). E.g. "pii_protection". description: Human-readable description of what this policy enforces. rules: List of rule dicts. Each rule needs: - match: regex pattern to detect (e.g. "(ssn|social.security)") - severity: "critical", "high", "medium", or "low" - reason: (optional) violation message template, use {match} for matched text - context_requires: (optional) additional regex that must also match - confidence: (optional) 0.0-1.0, default 0.85 step_types: Which step types to evaluate (default: ["ACTION"]). whitelisted_domains: Domains to skip evaluation for.

Returns: JSON string confirming policy creation with version number.

Disable a custom governance policy.

Soft-deletes the policy so it no longer evaluates on action submissions. Built-in policies (FinancialSafety, DataExfiltration) cannot be deleted.

Args: policy_name: Name of the custom policy to disable.

Returns: JSON string confirming policy was disabled.

Check the health of your agent's memory.

Returns a health score (0-100), stale memory count, conflict count, and contradiction count. Use this to monitor memory quality over time.

Returns: JSON string with health score and breakdown.

List recent Control actions with their status.

Shows submitted, pending, approved, denied, executed, and failed actions.

Args: limit: Maximum actions to return (default 20).

Returns: JSON string with action list.

Submit an action to Novyx Control for governed execution.

The action is evaluated against your tenant's policy profile. If approval is required, it will be queued for human review; otherwise executed immediately.

Args: connector: Target connector (github, slack, linear, pagerduty, http). operation: Operation name (e.g. issues/create, messages/send). payload: JSON string with the full action envelope.

Returns: JSON string with action ID, status, and policy decision.

Get the status of a specific Control action.

Returns full action details including policy decision, approval status, execution result, and evidence (Novyx trace, certificate).

Args: action_id: The action ID (e.g. act_xxx).

Returns: JSON string with action details.

Get the full causal chain for a Control action.

Returns what the agent recalled, what policy fired, what memory state existed at that moment, and the audit trail — all in one call. Requires Novyx Cloud.

Args: action_id: The action UUID to explain.

Returns: JSON string with explanation (memories, policy, audit trail).

Save a recall baseline for regression testing.

Future eval runs will check if this query still returns the expected result. Free: 1 baseline, Starter: 5, Pro: unlimited.

Args: query: The recall query to baseline. expected_observation: Expected top result observation.

Returns: JSON string with the created baseline.

List all saved eval baselines.

Shows the recall queries and expected observations used for regression testing.

Returns: JSON string with baselines list.

Delete an eval baseline.

Args: baseline_id: The baseline UUID to delete.

Returns: JSON string confirming deletion.

Export the full audit log.

Downloads the complete audit trail in the specified format. Requires Pro tier or Novyx Cloud.

Args: format: Export format — "json", "csv", or "jsonl" (default: json).

Returns: JSON string with exported audit data.

Update Cortex configuration.

Tune how Cortex consolidates, reinforces, and decays memories. Requires Pro tier or Novyx Cloud.

Args: consolidation_threshold: Similarity threshold for merging duplicates (0.0-1.0). reinforcement_boost: How much to boost frequently recalled memories. decay_rate: How fast forgotten memories fade.

Returns: JSON string with updated Cortex configuration.

Get the anonymized threat intelligence feed.

Shows attack patterns detected across the Novyx network, anonymized to protect individual tenants. Requires Pro tier or Novyx Cloud.

Args: hours: Look back window in hours (default 24). min_severity: Minimum severity to include — low, medium, high, critical (default medium).

Returns: JSON string with threat signatures.

Get overall threat intelligence statistics.

Shows total signatures, active threats, mitigated count, and severity breakdown. Requires Pro tier or Novyx Cloud.

Returns: JSON string with threat stats.

Record a threat event for cross-tenant intelligence.

The event is fingerprinted, deduplicated, and added to the threat network. Requires Pro tier or Novyx Cloud.

Args: threat_event: JSON string describing the threat (pattern_type, details, severity).

Returns: JSON string with the created/updated threat signature.

Get trending threat signatures.

Shows the most active threats in the specified time window. Requires Pro tier or Novyx Cloud.

Args: hours: Look back window in hours (default 24). min_occurrences: Minimum occurrence count to include (default 2).

Returns: JSON string with trending threat signatures.

Find known threat signatures matching a threat event.

Compares the event fingerprint against the threat database. Requires Pro tier or Novyx Cloud.

Args: threat_event: JSON string describing the threat to match. min_similarity: Minimum similarity threshold 0.0-1.0 (default 0.8).

Returns: JSON string with matching signatures.

Get a specific threat signature by ID.

Returns full details including pattern hash, severity, occurrence count, and whether a defense has been deployed. Requires Pro tier or Novyx Cloud.

Args: signature_id: The threat signature ID.

Returns: JSON string with signature details.

Mark a threat signature as mitigated.

Indicates that the threat has been addressed. Requires Pro tier or Novyx Cloud.

Args: signature_id: The threat signature ID to mark as mitigated.

Returns: JSON string confirming mitigation.

List active auto-deployed defense rules.

Shows all defense rules currently protecting against detected threats. Requires Pro tier or Novyx Cloud.

Args: rule_type: Optional filter — block, rate_limit, quarantine, or alert_only.

Returns: JSON string with active defenses.

Deploy a defense rule against a threat signature.

Automatically blocks, rate-limits, quarantines, or alerts on matching threats. Requires Pro tier or Novyx Cloud.

Args: signature_id: The threat signature to defend against. rule_type: Defense type — block, rate_limit, quarantine, or alert_only. rule_config: Optional JSON string with rule configuration.

Returns: JSON string with the deployed defense details.

Remove a deployed defense rule.

Stops the defense from blocking/rate-limiting threats. Requires Pro tier or Novyx Cloud.

Args: defense_id: The defense rule ID to remove.

Returns: JSON string confirming removal.

Measure the effectiveness of a deployed defense.

Returns a 0-1 score based on blocks vs false positives. Requires Pro tier or Novyx Cloud.

Args: defense_id: The defense rule ID to evaluate.

Returns: JSON string with effectiveness score.

Record that a defense blocked a threat.

Used to track defense performance. Mark false positives to auto-tune rules. Requires Pro tier or Novyx Cloud.

Args: defense_id: The defense rule that blocked the threat. is_false_positive: Whether this block was a false positive (default False).

Returns: JSON string confirming the record.

Get overall defense statistics.

Shows total defenses, active count, average effectiveness, and block counts. Requires Pro tier or Novyx Cloud.

Returns: JSON string with defense statistics.

Get a recommended defense strategy for a threat signature.

Returns the suggested rule type, confidence level, and reasoning. Requires Pro tier or Novyx Cloud.

Args: signature_id: The threat signature to get a recommendation for.

Returns: JSON string with recommendation (rule_type, confidence, reasoning).

Check if a threat correlates with attacks on other tenants.

Cross-references the threat fingerprint against the network database. Requires Pro tier or Novyx Cloud.

Args: threat_event: JSON string describing the threat to correlate.

Returns: JSON string with correlation result and matching signatures.

Detect an ongoing attack campaign.

Looks for coordinated attack patterns across multiple tenants. Returns campaign details if detected. Requires Pro tier or Novyx Cloud.

Args: hours: Look back window in hours (default 24).

Returns: JSON string with campaign details or campaign_detected: false.

Check if multiple threat events are part of a coordinated attack.

Analyzes timing and pattern similarity across events. Requires Pro tier or Novyx Cloud.

Args: threat_events: JSON string with array of threat events to analyze. time_window_hours: Optional time window for correlation (default: auto).

Returns: JSON string with is_coordinated boolean.

Find threat signatures related to a given signature.

Uses pattern similarity to find related threats. Requires Pro tier or Novyx Cloud.

Args: signature_id: The signature ID to find related threats for. similarity_threshold: Minimum similarity 0.0-1.0 (default 0.7).

Returns: JSON string with related signatures.

Get real-time memory stream connection status.

Shows active connections, max allowed, and event bus metrics. Requires Novyx Cloud.

Returns: JSON string with stream status.

Introspect the MCP tool surface.

Returns a machine-readable registry of every tool exposed by this server, each with a status (functional / cloud_only / cloud_only_hard_fail / disabled / stub), category (memory, graph, runtime, control, ...), and a one-line description. Use this to answer "what can this MCP actually do, and which parts require Novyx Cloud?" without having to call every tool.

Args: status: Optional filter — only return tools with this status. category: Optional filter — only return tools in this category.

Returns: JSON string: {counts, tools}. counts is over the full registry (not the filtered view); tools is filtered if status/category given.

Create a first-class agent entity in the Novyx Runtime.

Novyx is provider-agnostic — you must specify which LLM backend the agent uses. Use "litellm" to reach models not directly supported (Gemini, Mistral, Cohere, local Ollama, etc.).

Args: name: Human-readable agent name. provider: LLM provider (required): "openai", "anthropic", or "litellm". model: LLM model name (required, e.g. "gpt-4o", "claude-sonnet-4-6"). agent_id: Custom agent ID (auto-generated if omitted). description: Agent description. instructions: System prompt / instructions. capabilities: List of enabled capability pack names. memory_scope: Memory scope for the agent (e.g. "private", "shared"). policy_profile: JSON string of policy profile configuration.

List all agents for the current tenant.

Args: status: Filter by status (active, paused, archived). limit: Max results.

Get an agent by ID.

Delete an agent.

Update an existing agent's configuration.

Args: agent_id: Agent to update. name: New agent name. model: New LLM model name. provider: New LLM provider. instructions: New system prompt / instructions. capabilities: New list of enabled capability pack names. memory_scope: New memory scope. policy_profile: JSON string of policy profile configuration. config: JSON string of additional configuration.