Tools Reference
Current novyx-mcp builds register 102 tools. This page documents the stable public families and calls out support surfaces that are experimental or product-dependent. Local mode supports memory/context workflows with SQLite. Cloud mode is required for protected actions, approvals, shared audit evidence, and team governance.
Tier key: Free = available on all tiers including local mode | Starter+ = requires Starter tier or higher | Pro+ = requires Pro tier or higher | Enterprise = requires Enterprise tier
Core Memory
Store, retrieve, and manage memories.
| Tool | Description | Key Parameters | Tier |
|---|---|---|---|
remember | Store a memory observation in Novyx. Auto-tags with the current git repo name. | observation (str), tags (list), importance (int, 1-10), context (str), ttl_seconds (int) | Free |
recall | Search memories semantically using natural language. Scores combine cosine similarity, importance, and confidence. | query (str), limit (int), tags (list), min_score (float), explain (bool) | Free |
forget | Delete a memory by its UUID. | memory_id (str) | Free |
list_memories | List stored memories with optional tag filtering. | limit (int), tags (list) | Free |
supersede | Mark a memory as superseded by a newer one. The old memory remains for audit purposes. | old_memory_id (str), new_memory_id (str) | Free |
link_memories | Create a directed link between two memories (e.g. "related", "causes", "supports"). | source_id (str), target_id (str), relation (str) | Free |
unlink | Remove a link between two memories. | source_id (str), target_id (str) | Free |
get_links | Get all incoming and outgoing links for a memory. | memory_id (str), relation (str) | Free |
graph_edges | Query the memory graph edges with filters on direction and relation type. | memory_id (str), relation (str), direction (str: outgoing/incoming/both), limit (int) | Free |
Knowledge Graph
Build and query a subject-predicate-object knowledge graph.
| Tool | Description | Key Parameters | Tier |
|---|---|---|---|
add_triple | Add a knowledge graph triple (subject -> predicate -> object). Entities are auto-created. | subject (str), predicate (str), object_name (str) | Free |
query_triples | Query triples with optional filters. At least one filter should be provided. | subject (str), predicate (str), object_name (str) | Free |
delete_triple | Delete a knowledge graph triple by ID. | triple_id (str) | Free |
list_entities | List knowledge graph entities (the nodes in your graph). | limit (int), offset (int), entity_type (str) | Free |
get_entity | Get an entity and its associated triples. | entity_id (str) | Free |
delete_entity | Delete an entity and all its triples. | entity_id (str) | Free |
Context Spaces
Multi-agent collaboration through shared memory spaces.
| Tool | Description | Key Parameters | Tier |
|---|---|---|---|
create_space | Create a shared context space. The creator is the owner with full permissions. | name (str), description (str), allowed_agents (list), tags (list) | Free |
list_spaces | List all context spaces you own or have access to. | — | Free |
space_memories | List or search memories within a context space. | space_id (str), query (str), limit (int) | Free |
update_space | Update a context space (owner only). | space_id (str), name (str), description (str), allowed_agents (list), tags (list) | Free |
delete_space | Delete a context space (owner only). | space_id (str) | Free |
share_space | Share a space/tag with another user by email. Cloud only. | tag (str), email (str), permission (str: read/write) | Starter+ |
accept_shared_context | Accept a shared context invitation using a token. Cloud only. | token (str) | Starter+ |
shared_contexts | List all shared contexts you have access to or have shared. Cloud only. | — | Starter+ |
revoke_shared_context | Revoke a shared context invitation. Cloud only. | token (str) | Starter+ |
Rollback & Recovery
Memory rollback and checkpoint recovery helpers. These are supporting incident-review tools, not a guarantee that every external side effect or full agent state can be transactionally undone.
| Tool | Description | Key Parameters | Tier |
|---|---|---|---|
rollback | Rollback memory to a point in time. Supports ISO timestamps and relative expressions (e.g. "2 hours ago"). | target (str), dry_run (bool) | Starter+ |
rollback_preview | Preview what a rollback would undo without executing it. Always use before an actual rollback. | target (str) | Starter+ |
rollback_history | List past rollback operations with timestamps and undo counts. | limit (int) | Starter+ |
Audit & Evidence
Tamper-evident audit trail and integrity checks for memory and action evidence.
| Tool | Description | Key Parameters | Tier |
|---|---|---|---|
audit | Get the audit trail of memory operations (CREATE, UPDATE, DELETE, ROLLBACK). | limit (int), operation (str) | Free |
audit_verify | Verify audit trail integrity. Checks the cryptographic hash chain (cloud) or entry consistency (local). | — | Starter+ |
Execution Traces
Track and verify multi-step agent workflows.
| Tool | Description | Key Parameters | Tier |
|---|---|---|---|
trace_create | Create an execution trace to track a multi-step workflow. | name (str), metadata (JSON str) | Starter+ |
trace_step | Add a step to an active execution trace. Include input/output for debugging. | trace_id (str), step_name (str), input_data (JSON str), output_data (JSON str) | Starter+ |
trace_complete | Mark an execution trace as complete. The trace and steps become immutable. | trace_id (str) | Starter+ |
trace_verify | Verify an execution trace's integrity. Confirms all steps are present and untampered. | trace_id (str) | Starter+ |
Replay
Time-travel debugging for memory state.
| Tool | Description | Key Parameters | Tier |
|---|---|---|---|
replay_timeline | Get the full timeline of memory operations. The tape you scrub through. | since (ISO str), until (ISO str), operations (comma-separated str), limit (int) | Pro+ |
replay_snapshot | Experimental snapshot helper. Reconstructs which memories existed at a timestamp but may back-fill present field values. Disabled by default in production. | at (ISO str), limit (int) | Pro+ |
replay_lifecycle | Full biography of a single memory: creation, updates, recalls, links, and deletion. | memory_id (str) | Pro+ |
replay_diff | Diff memory state between two timestamps. Shows added, removed, and modified memories. | start (ISO str), end (ISO str) | Pro+ |
replay_memory | Get the full chronological history of a single memory. | memory_id (str) | Pro+ |
replay_recall | Time-travel recall: what would search have returned at a past timestamp? | query (str), at (ISO str), limit (int) | Pro+ |
replay_memory_drift | Detect memory drift between two timestamps. Compare state and show changes. | from_ts (ISO str), to_ts (ISO str) | Pro+ |
Cortex
Experimental
Cortex is a supporting memory-maintenance surface and may be disabled by default. Do not build production action-control flows that depend on Cortex being available.
Memory intelligence -- consolidation, reinforcement, and insights.
| Tool | Description | Key Parameters | Tier |
|---|---|---|---|
cortex_status | Get Cortex status: enabled state, last run time, consolidation/reinforcement stats. | — | Pro+ |
cortex_run | Experimental memory-maintenance cycle. Disabled by default unless experimental features are enabled. | — | Pro+ |
cortex_config | Get the current Cortex configuration: thresholds, decay rates, cycle schedule. | — | Pro+ |
cortex_insights | List experimental synthetic memory insights when Cortex is enabled. | limit (int) | Enterprise |
Eval
Experimental
Eval is a supporting quality signal for memory/context. It is not the primary production-action gate.
Memory quality scoring and CI gates.
| Tool | Description | Key Parameters | Tier |
|---|---|---|---|
eval_run | Run a memory health evaluation. Scores quality on a 0-1 scale based on staleness, conflicts, and superseded memories. | min_score (float) | Starter+ |
eval_gate | Experimental memory-health threshold check. Disabled by default unless experimental features are enabled. | min_score (float) | Pro+ |
eval_history | List past evaluation runs. Track quality over time. | limit (int) | Starter+ |
eval_drift | Detect memory drift over a time period. Shows creates, deletes, and updates. | days (int) | Starter+ |
Actions & Control
Governed agent actions, human-in-the-loop approval workflows, and policy-as-code authoring.
| Tool | Description | Key Parameters | Tier |
|---|---|---|---|
list_pending | List pending Control approval requests awaiting human review. | limit (int) | Starter+ |
approve_action | Approve a pending agent action. Triggers execution against the target connector (GitHub, Slack, Linear, PagerDuty, HTTP). | approval_id (str), approver_id (str), reason (str) | Starter+ |
check_policy | Check the current Control policy profile. Shows which connectors require approval. | connector (str), environment (str) | Starter+ |
action_history | List recent Control actions with their status (submitted, pending, approved, denied, executed, failed). | limit (int) | Starter+ |
create_policy | Create or update a custom Control policy. Pass agent_id to scope it to a single agent (Pro+). | name (str), description (str), rules (list), step_types (list), whitelisted_domains (list) | Starter+ |
list_policies | List active Control policies (built-in + tenant custom). | enabled_only (bool, default True) | Free |
delete_policy | Disable a custom policy. Built-in policies cannot be deleted. | policy_name (str) | Starter+ |
Memory Drafts
Propose, review, and merge memory changes before they become canonical.
| Tool | Description | Key Parameters | Tier |
|---|---|---|---|
draft_memory | Create a reviewable draft without writing to canonical memory. Returns similar memories and a review summary. | observation (str), tags (list), importance (int), context (str), confidence (float), branch_id (str) | Free |
memory_drafts | List current memory drafts with optional status and branch filtering. | status (str: draft/merged/rejected), branch_id (str) | Free |
draft_diff | Show a field-level diff for a memory draft, with a merge recommendation. | draft_id (str), compare_to (str) | Free |
merge_draft | Merge a reviewed draft into canonical memory. Optionally supersede an older memory. | draft_id (str), supersede_memory_id (str) | Free |
reject_draft | Reject a draft without creating a memory. | draft_id (str), reason (str) | Free |
memory_branch | Get grouped review information for a branch/session of drafts. | branch_id (str) | Free |
merge_branch | Merge all open drafts in a branch/session at once. | branch_id (str) | Free |
reject_branch | Reject all open drafts in a branch/session. | branch_id (str), reason (str) | Free |
System
Dashboard, statistics, and health monitoring.
| Tool | Description | Key Parameters | Tier |
|---|---|---|---|
dashboard | Full dashboard overview combining stats, spaces, and recent activity. | — | Free |
memory_stats | Get memory statistics: total count, average importance, tag distribution. | — | Free |
memory_health | Check memory health on a 0-100 scale. Reports stale count, conflicts, and contradictions. | — | Free |
context_now | Snapshot of current memory context: recent memories, stats, and audit activity. | — | Free |
Runtime v2
First-class agent lifecycle, missions, capabilities, checkpoints, and supervisor interventions. Full documentation →
| Tool | Description | Key Parameters | Tier |
|---|---|---|---|
create_agent | Register a persistent agent in the Novyx Runtime. Required since v2.5.0: provider and model are no longer optional. Pass provider="openai", "anthropic", or "litellm". See the novyx-agent 2.0 upgrade guide. | name (str), provider (str, required), model (str, required), capabilities (list) | Free |
list_agents | List all agents for the current tenant. | status (str), limit (int) | Free |
get_agent | Get an agent by ID. | agent_id (str) | Free |
delete_agent | Delete an agent. | agent_id (str) | Free |
create_mission | Create a mission (bounded job) for an agent. | agent_id (str), goal (str), constraints (list), success_criteria (list) | Free |
list_missions | List missions for the current tenant. | agent_id (str), status (str), limit (int) | Free |
get_mission | Get a mission by ID. | mission_id (str) | Free |
pause_mission | Pause a running mission. | mission_id (str) | Free |
resume_mission | Resume a paused mission. | mission_id (str) | Free |
cancel_mission | Cancel a mission. | mission_id (str) | Free |
create_capability | Register a capability pack (tool bundle with governance). | name (str), tools (list), risk_levels (object) | Free |
list_capabilities | List registered capability packs. | — | Free |
create_checkpoint | Create a checkpoint for a mission (rollback point). | mission_id (str), label (str) | Free |
list_checkpoints | List checkpoints for a mission. | mission_id (str) | Free |
rollback_to_checkpoint | Rollback a mission to a previous checkpoint. | mission_id (str), checkpoint_id (str), reason (str) | Free |
create_intervention | Record a supervisor intervention. | intervention_type (str), mission_id (str), rationale (str) | Free |
list_interventions | List supervisor interventions. | mission_id (str), agent_id (str), intervention_type (str) | Free |
Tool Count by Category
| Category | Tools |
|---|---|
| Core Memory | 9 |
| Knowledge Graph | 6 |
| Context Spaces | 9 |
| Rollback & Recovery | 3 |
| Audit & Compliance | 2 |
| Execution Traces | 4 |
| Replay | 7 |
| Cortex | 4 |
| Eval | 4 |
| Actions & Control | 7 |
| Memory Drafts | 8 |
| Runtime v2 | 17 |
| System | 4 |
| Subtotal documented above | 84 |
| Additional governance/runtime/support tools | 18 |
| Current registry total | 102 |
Exact tool counts can change as experimental tools move in or out of the MCP registry. Treat the action-gate workflow as the product surface; use counts only as release-specific reference data.