Invariance MCP
Official@invariance/mcp
Ein MCP-Server (Model Context Protocol), der KI-Coding-Agenten mit der Invariance-Observability-Plattform verbindet. Er bietet Tools wie Claude Desktop, Cursor und Claude Code direkten Zugriff auf Ihre Runs, Nodes, Monitore, Signale, Findings, Reviews und mehr.
MCP ist ein offenes Protokoll, das es KI-Assistenten ermöglicht, externe Tools und Datenquellen zu nutzen. Dieser Server implementiert es für Invariance, sodass Ihr KI-Assistent Observability-Daten abfragen, Probleme untersuchen und das Verhalten von Agenten analysieren kann, ohne das Gespräch zu verlassen.
Installation
npm install -g @invariance/mcpOder führen Sie es direkt mit npx aus (empfohlen für MCP-Client-Konfigurationen):
npx @invariance/mcpRelated MCP server: Claude Code Starter Kit MCP
Einrichtung
Umgebungsvariablen
Variable | Erforderlich | Standard | Beschreibung |
| Ja | — | Ihr Invariance API-Schlüssel |
| Nein |
| API-Basis-URL |
| Nein |
| Transportmodus: |
| Nein |
| Port für SSE/HTTP-Transport |
| Nein |
| Request-Timeout in Millisekunden |
Sie erhalten Ihren API-Schlüssel unter app.invariance.ai/settings/api-keys.
Claude Desktop
Fügen Sie dies zu Ihrer Claude Desktop-Konfiguration hinzu (~/Library/Application Support/Claude/claude_desktop_config.json unter macOS):
{
"mcpServers": {
"invariance": {
"command": "npx",
"args": ["-y", "@invariance/mcp"],
"env": {
"INVARIANCE_API_KEY": "your-api-key"
}
}
}
}Claude Code
Fügen Sie dies zu Ihrer Claude Code-Konfiguration hinzu (.claude/settings.json oder ~/.claude/settings.json):
{
"mcpServers": {
"invariance": {
"command": "npx",
"args": ["-y", "@invariance/mcp"],
"env": {
"INVARIANCE_API_KEY": "your-api-key"
}
}
}
}Cursor
Fügen Sie dies zu Ihren Cursor MCP-Einstellungen hinzu (.cursor/mcp.json):
{
"mcpServers": {
"invariance": {
"command": "npx",
"args": ["-y", "@invariance/mcp"],
"env": {
"INVARIANCE_API_KEY": "your-api-key"
}
}
}
}Verfügbare Tools
Der Server stellt 37 Tools bereit, die die vollständige Invariance-API abdecken. Die Namen folgen dem Schema invariance_<resource>_<action>.
Runs (invariance_run_*)
start, get, list, finish, fail, verify, metrics
Nodes (invariance_node_*)
write, list
Monitore (invariance_monitor_*)
create, list, get, update, pause, resume, evaluate, executions, findings
Signale (invariance_signal_*)
emit, list, get, acknowledge, resolve
Findings (invariance_finding_*)
list, get, update
Reviews (invariance_review_*)
list, get, claim, unclaim, resolve
Agenten (invariance_agent_*)
me, set_key
Insights
invariance_narrative_get (LLM-synthetisierte Run-Zusammenfassung), invariance_ask (turn-basierte Q&A über Ihre KB + Runs), invariance_kb_pages_list, invariance_kb_page_get
Für komplexe Objektargumente (Monitor-Body, Signaldaten, Node-Input/Output, Run-Metadaten) akzeptieren die Tools JSON-kodierte Strings, die der Server vor der Weiterleitung an die API parst.
Legacy-Tool-Aliase
Die ursprünglichen 6 Tool-Namen aus früheren Versionen bleiben als Aliase erhalten, damit bestehende Client-Konfigurationen weiterhin funktionieren: invariance_create_run, invariance_get_run, invariance_list_runs, invariance_write_node, invariance_list_nodes, invariance_verify_run.
HTTP-Transport
Um den Server über Streamable HTTP anstelle von stdio auszuführen:
INVARIANCE_API_KEY=your-api-key INVARIANCE_MCP_TRANSPORT=http INVARIANCE_MCP_PORT=3000 npx @invariance/mcpDer Server stellt einen Streamable HTTP-Endpunkt unter http://127.0.0.1:3000/mcp und einen Health-Check unter http://127.0.0.1:3000/health bereit.
MCP-Clients, die HTTP-Transport unterstützen, können sich über die /mcp-Endpunkt-URL verbinden, anstatt einen Subprozess zu starten.
Fehlerbehebung
"INVARIANCE_API_KEY environment variable is required"
Stellen Sie sicher, dass Sie die Umgebungsvariable INVARIANCE_API_KEY in Ihrer MCP-Client-Konfiguration gesetzt haben. Siehe die Einrichtungsanleitungen oben.
Server erscheint nicht in Ihrem MCP-Client
Überprüfen Sie, ob der Pfad zur Konfigurationsdatei für Ihren Client korrekt ist
Starten Sie den Client nach dem Bearbeiten der Konfiguration neu
Überprüfen Sie, ob
npx @invariance/mcpin Ihrem Terminal ohne Fehler ausgeführt wird
Authentifizierungsfehler
Überprüfen Sie, ob Ihr API-Schlüssel unter app.invariance.ai/settings/api-keys gültig ist.
Verbindungs-Timeouts
Wenn Sie eine benutzerdefinierte INVARIANCE_BASE_URL verwenden, überprüfen Sie, ob die URL erreichbar ist.
Mitwirken
Siehe CONTRIBUTING.md.
Lizenz
MIT
Available Tools
159 toolscortex_askA
Ask the READ-ONLY Cortex analyst (complex_query) an operational question and get a cited answer. Use this for questions like "Were refund SLAs met last week?", "Why did this run diverge?", "Which agents touched case_123?". The analyst is governed and EVIDENCE-CITED: every id in evidence_refs / affected_entities was observed through a read tool, and the runtime FAILS CLOSED against fabricated or cross-project ids (no answer is invented and no other tenant's data can leak). It only reads — it never mutates state. Returns the validated ComplexQueryResult {short_answer, reasoning_plan, evidence_refs, affected_entities, confidence, restricted_evidence_count, recommended_action, follow_up_questions}. mode="sync" (default) blocks for the answer; mode="async" enqueues then polls until the job is terminal. Note: the analyst only executes when the platform CORTEX_TOOL_RUNTIME_ENABLED flag is on.
Inputs: question (required), project_id (required), mode, target_ref, +1 more.
cortex_get_jobARead-only
Get a Cortex job's metadata and status (no artifacts). Returns the same safe-field projection as cortex_run_job: ids, status, target, actor, criteria, timestamps, error. Use cortex_get_result to fetch the structured result body.
Inputs: job_id (required).
cortex_get_resultARead-only
Get a Cortex job's structured result. Returns {job_id, status, result?}. The result is validated against the known per-kind schemas (workflow_eval / counterfactual_eval / outcome_attribution / complex_query / divergence_error_tracking). Raw artifacts (prompt input, raw model output) are NOT returned by this tool — they remain private on the platform.
Inputs: job_id (required).
cortex_job_runsARead-only
List the attempt history (audit-trail runs) for a Cortex job — one row per execution attempt, with status, model, metrics, timings, and any error. Read-only. Raw prompt/model artifacts are NOT included. Returns {runs: CortexJobRun[]}.
Inputs: job_id (required).
cortex_launchA
Launch a Cortex job through the GOVERNED launcher (POST /v1/cortex/jobs/launch) — the preferred path for the read-only complex_query analyst and divergence_error_tracking. The actor is resolved server-side from the API key and all evidence is ACL-filtered before prompt construction (fails closed against cross-project leak). mode="sync" runs now and embeds result/error; mode="async" enqueues — poll with cortex_get_result or cortex_job_runs. Returns {job_id, status, mode, deduplicated, result?, error?}. Idempotent when idempotency_key is supplied.
Inputs: mode (required), job_kind (required), project_id (required), target_ref (required), +3 more.
cortex_list_jobsARead-only
List Cortex jobs across accessible projects, newest first. Filter by status and/or kind. Read-only. Returns {data: CortexJob[], next_cursor}. Pass next_cursor back as cursor to page.
Inputs: kind, limit, cursor, status.
cortex_retry_jobA
Re-queue a failed or dead Cortex job for one more attempt (POST /v1/cortex/jobs/:id/retry). Returns {job_id, status}.
Inputs: job_id (required).
cortex_run_counterfactualA
Convenience wrapper around cortex_run_job for job_kind="counterfactual_eval": estimates what MIGHT have happened under a hypothetical change. Result is a HYPOTHESIS, not fact — it carries assumptions, evidence_refs, confidence, and uncertainty. question is required.
Inputs: question (required), project_id (required), target_ref (required), target_type (required), +4 more.
cortex_run_evalB
Convenience wrapper around cortex_run_job for job_kind="workflow_eval": checks whether a run/case/workflow met its criteria (e.g. SLA, policy compliance, action-item ownership). Returns the same job shape as cortex_run_job.
Inputs: project_id (required), target_ref (required), target_type (required), options, +4 more.
cortex_run_jobA
Enqueue a generic Cortex job (evals, counterfactuals, experiments, attributions). The actor is resolved server-side from the API key (a key bound to an agent_id runs as that agent; otherwise as the api_key actor). The platform ACL-filters input_refs and target access before prompt construction. Returns {job_id, status} plus, when a synchronous MVP runner completes the job inline, the validated result.
Inputs: job_kind (required), project_id (required), target_ref (required), target_type (required), +5 more.
invariance_agent_createA
Create a new agent inside one of the caller's projects. Requires a user-session JWT bearer (not an agent API key) — see invariance-cli inv auth signup / inv auth signin to obtain one.
Inputs: name (required), project_id (required), public_key.
invariance_agent_getARead-only
Fetch a single agent by ID. Requires a user-session JWT bearer.
Inputs: id (required).
invariance_agent_listARead-only
List agents inside one of the caller's projects. Requires a user-session JWT bearer.
Inputs: project_id (required).
invariance_agent_meARead-only
Show the agent identity and API key associated with the current credentials. Useful for confirming which agent context the MCP server is operating as.
Inputs: no named parameters.
invariance_agent_set_keyA
Register or rotate the calling agent's Ed25519 public key. Once set, every node written by this agent must be signed with the matching private key (the server uses this key to verify signatures during invariance_run_verify).
Inputs: public_key (required).
invariance_askBRead-only
Ask a question against the agent's runs / knowledge base (turn-based session)
Inputs: message (required), model, max_turns, session_id.
invariance_capture_createA
Create a Capture — a standalone evidence record (session, conversation, trace). Captures don't need an execution upfront; link a capture to a run later with invariance_capture_link. Returns the capture session.
Inputs: source (required), tags, model, title, +5 more.
invariance_capture_getARead-only
Get a capture by id. Captures are standalone evidence; they don't need an execution upfront; link a capture to a run later with invariance_capture_link.
Inputs: id (required).
invariance_capture_linkA
Link a capture to an evidence-graph target (run/case/workflow_event/node). With only run_id this sets the capture's run_id (legacy, equivalent to invariance_capture_update). With target_id it creates a richer capture link; target_type defaults to "run". Captures are standalone evidence and don't need an execution upfront.
Inputs: id (required), run_id, link_type, target_id, +1 more.
invariance_capture_linksARead-only
List every evidence link on a capture (case/run/workflow_event/node), returning { links }. Captures are standalone evidence; link a capture to a target with invariance_capture_link.
Inputs: id (required).
invariance_capture_listARead-only
List captures (paginated). Captures are standalone evidence; they don't need an execution upfront. Filter by project_id, operator_id, session_type, source, run_id, or tags.
Inputs: tags, limit, cursor, run_id, +4 more.
invariance_capture_unlinkA
Detach a capture link. With link_id, deletes that specific evidence link. Without link_id, clears the capture's run_id (legacy). Captures are standalone evidence; they don't need an execution upfront.
Inputs: id (required), link_id.
invariance_capture_updateA
Update a capture: change status, reassign run_id, or merge metadata. Captures are standalone evidence; they don't need an execution upfront; link a capture to a run later with invariance_capture_link.
Inputs: id (required), tags, run_id, status, +1 more.
invariance_case_closeA
Close a case with an outcome — the common path. Equivalent to invariance_case_update with status="closed" + outcome + outcome_value_usd.
Inputs: id (required), outcome (required), closed_at, value_usd.
invariance_case_createA
Create a workflow-instance Case. A case owns many runs across time, agents, and humans (one loan, one audit, one claim). Returns the case; use its id when starting runs with invariance_run_start to link them.
Inputs: workflow_key (required), tags, owner, opened_at, +3 more.
invariance_case_event_createA
Attach a semantic workflow event to a case, e.g. "support.customer.escalated", "approval.granted", or "docs.received". Prefer this for meaningful workflow facts; keep raw execution trace data in runs/nodes.
Inputs: id (required), type (required), tags, payload, +5 more.
invariance_case_events_listARead-only
List semantic workflow events attached to one case. These are the queryable facts over the run/node evidence layer.
Inputs: id (required), limit, cursor.
invariance_case_evidenceARead-only
Show normalized evidence for a case: the case, linked runs, nodes, workflow events, actors, and outcome. Use this when you need the full workflow execution record instead of only the case row.
Inputs: id (required).
invariance_case_getARead-only
Get a case by id, including its linked runs (newest first, capped at 100). Use to inspect status, outcome, owner, custom_attrs, and the runs the case has accumulated.
Inputs: id (required).
invariance_case_listARead-only
List cases visible to the calling agent (paginated). Filter by tenant_id, end_user_id, workflow_key, status ("open" | "closed"), outcome, or tags.
Inputs: tags, limit, cursor, status, +4 more.
invariance_case_updateA
Update a case: change owner, merge custom_attrs (shallow), or transition status. Use invariance_case_close for the common "set outcome + close" path.
Inputs: id (required), tags, owner, status, +4 more.
invariance_create_runD
Alias of invariance_run_start
Inputs: name.
invariance_divergence_getBRead-only
Get a divergence by ID.
Inputs: id (required).
invariance_divergence_listBRead-only
List divergences (expected-vs-observed gaps) visible to the caller. Filter by run, kind, severity, status.
Inputs: kind, limit, cursor, run_id, +2 more.
invariance_divergence_updateB
Transition a divergence status: open | accepted | dismissed | converted_to_monitor.
Inputs: id (required), status (required).
invariance_dna_accept_edge_candidateA
Accept a proposed DNA edge candidate, making it eligible for promotion into a durable semantic link.
Inputs: id (required).
invariance_dna_list_edge_candidatesARead-only
List Company DNA edge candidates — discovered relationships between objects awaiting review. Filter by status (proposed/accepted/rejected/expired/promoted), object_id, or relation_kind. Output: {data: DnaEdgeCandidate[], next_cursor}.
Inputs: limit, cursor, status, object_id, +2 more.
invariance_dna_list_edgesARead-only
List durable Company DNA edges — relationships between objects, e.g. the operational graph derived from a run (refund -[REQUIRED]-> policy). Filter by run_id, kind, or entity_id. Output: {data: DnaEdge[], next_cursor}.
Inputs: kind, limit, cursor, run_id, +1 more.
invariance_dna_list_object_mentionsARead-only
List Company DNA object mentions — extracted references (in events or chunks) that may resolve to a DNA object. Filter by event_id, chunk_id, object_id, or mention_type. Output: {data: DnaObjectMention[], next_cursor}.
Inputs: limit, cursor, chunk_id, event_id, +3 more.
invariance_dna_list_objectsBRead-only
List Company DNA objects — the canonical operational entities in a project's DNA graph (e.g. tickets, services, policies). Filter by kind or a free-text query q. Output: {data: DnaObject[], next_cursor}.
Inputs: q, kind, limit, cursor, +1 more.
invariance_dna_promote_edge_candidateA
Promote an accepted DNA edge candidate into a durable semantic link. The candidate must be accepted, carry a semantic_similarity signal, and have at least two evidence chunks (otherwise the API returns 422). Idempotent: re-promoting returns the existing link with already_promoted=true. Set dry_run=true to preview the would-be link without writing.
Inputs: id (required), dry_run.
invariance_dna_reject_edge_candidateA
Reject a DNA edge candidate so it is never promoted.
Inputs: id (required).
invariance_doctorARead-only
Run a health check on this MCP server: verifies API key auth, API reachability, and reports server name/version. Mirrors inv doctor --json from the CLI. Returns {checks: [{name, status, message}], summary: {pass, fail, warn}}.
Inputs: no named parameters.
invariance_eval_case_createC
Add a case (input + expected) to an eval suite.
Inputs: body (required), suite_id (required).
invariance_eval_case_create_from_runA
Snapshot an existing production run as a new eval case in a suite (captures the run's input + output as expected).
Inputs: body (required), suite_id (required).
invariance_eval_case_listBRead-only
List cases for an eval suite (paginated).
Inputs: suite_id (required), limit, cursor.
invariance_eval_dataset_append_exampleA
Append a single example row (input + expected output) to an existing dataset.
Inputs: id (required), body (required).
invariance_eval_dataset_createB
Create a reusable eval dataset (a named collection of input/expected example rows used to drive experiments).
Inputs: body (required).
invariance_eval_dataset_examples_listBRead-only
List example rows for a dataset (paginated).
Inputs: id (required), limit, cursor.
invariance_eval_dataset_getARead-only
Get a single eval dataset by ID.
Inputs: id (required).
invariance_eval_dataset_listARead-only
List eval datasets visible to the calling agent (paginated).
Inputs: limit, cursor.
invariance_eval_dataset_seed_suiteA
One-call eval setup for agents: create a dataset, append rows, create a linked suite, create one case per row, and optionally start the eval run. This is the preferred MCP path for turning JSON examples into runnable evals.
Inputs: body (required).
invariance_eval_experiment_compareARead-only
Compare two scored eval runs case-by-case (CompareResponse: per-case ScoreDelta entries + aggregate deltas per scorer). Use to surface regressions vs. a baseline.
Inputs: id (required), baseline (required).
invariance_eval_experiment_runA
Execute an experiment against an existing eval run: applies a list of scorer specs to every case result and (optionally) records a baseline run for later compare. Populates eval_results.scores. Built-in scorer names: exact_match, contains, numeric_tolerance (config.tolerance: number), json_match, levenshtein.
Inputs: id (required), body (required).
invariance_eval_run_getBRead-only
Get an eval run by ID (status, aggregate counts, timestamps, scorer specs).
Inputs: id (required).
invariance_eval_run_resultsARead-only
List per-case results for an eval run (paginated). Each result has output, expected, scores (per-scorer 0..1), and pass/fail.
Inputs: id (required), limit, cursor.
invariance_eval_scorer_createA
Register a scorer (named scoring rule that maps an output+expected pair to a 0..1 score). Built-in scorer kinds: exact_match, contains, numeric_tolerance, json_match, levenshtein.
Inputs: body (required).
invariance_eval_scorer_listARead-only
List scorers visible to the calling agent (paginated).
Inputs: limit, cursor.
invariance_eval_scorers_list_builtinARead-only
List the built-in scorer kinds available on the platform (name + config schema). Use this to discover what you can pass in scorer_specs to invariance_eval_experiment_run.
Inputs: no named parameters.
invariance_eval_suite_createA
Create an eval suite (the legacy grouping for cases + runs). New work should generally prefer datasets; suites remain for back-compat and curated case sets.
Inputs: body (required).
invariance_eval_suite_getCRead-only
Get an eval suite by ID.
Inputs: id (required).
invariance_eval_suite_listBRead-only
List eval suites visible to the calling agent (paginated).
Inputs: limit, cursor.
invariance_eval_suite_runA
Kick off an eval run: executes every case in the suite against a target (agent / recipe / inline override) and stores per-case results.
Inputs: suite_id (required), body.
invariance_finding_getARead-only
Get a finding by ID.
Inputs: id (required).
invariance_finding_listBRead-only
List findings (durable, structured issues raised by monitors or agents) visible to the caller, paginated.
Inputs: limit, cursor.
invariance_finding_updateA
Transition a finding to a new status: "open" (active), "review_requested" (escalated to a human/agent reviewer), "resolved" (fixed), or "dismissed" (intentionally ignored / false positive).
Inputs: id (required), status (required).
invariance_get_runC
Alias of invariance_run_get
Inputs: id (required).
invariance_guardrail_createB
Create a guardrail (from a recipe or finding). Required: title. Optional: recipe_id, finding_id, rule, mode, status.
Inputs: title (required), mode, rule, status, +3 more.
invariance_guardrail_getARead-only
Get a guardrail by ID.
Inputs: id (required).
invariance_guardrail_listARead-only
List per-agent guardrails. Filter by status or recipe_id.
Inputs: limit, cursor, status, recipe_id.
invariance_guardrail_promoteA
Promote a guardrail to a new lifecycle status: suggested → accepted → shadow → active_monitor → rejected.
Inputs: id (required), to (required).
invariance_guardrail_updateC
Patch a guardrail (mode, status, monitor_id).
Inputs: id (required), mode, status, monitor_id.
invariance_kb_page_createA
Create a knowledge-base page
Inputs: body (required), path (required), title (required), kind, +1 more.
invariance_kb_page_deleteCDestructive
Delete a knowledge-base page
Inputs: id (required).
invariance_kb_page_getBRead-only
Get a knowledge-base page by ID
Inputs: id (required).
invariance_kb_pages_listCRead-only
List knowledge-base pages
Inputs: kind, limit, cursor, search.
invariance_kb_page_updateC
Update fields on a knowledge-base page
Inputs: id (required), body, kind, title, +1 more.
invariance_kb_session_append_messageC
Append a message to a KB ask session
Inputs: id (required), content (required), role.
invariance_kb_session_createB
Create a multi-turn ask session for the agent KB
Inputs: model, title.
invariance_kb_session_deleteBDestructive
Delete a KB ask session
Inputs: id (required).
invariance_kb_session_list_messagesARead-only
List all messages in a KB ask session in order
Inputs: id (required).
invariance_list_nodesD
Alias of invariance_node_list
Inputs: run_id (required).
invariance_list_runsC
Alias of invariance_run_list
Inputs: no named parameters.
invariance_memory_readARead-only
Record a memory read by an agent against a subject (customer/account/policy/...) and return the current MemoryRecord (if any). Use this whenever an agent consults a remembered belief — it produces an auditable MemoryAccess event tying that belief to a node in the run, which the divergence detectors use to flag stale or unsupported memory.
Inputs: key (required), used_for (required), subject_id (required), subject_type (required), +2 more.
invariance_memory_writeA
Record a memory write by an agent: set or update a belief (claim) about a subject. Returns the new MemoryAccess + MemoryRecord. Defaults: source="agent_write", confidence=1.0. Provide provenance (EvidenceRef[] as JSON) when the claim is derived from authoritative records (CRM/ticket/policy doc) so downstream divergence checks can verify it.
Inputs: key (required), value (required), used_for (required), subject_id (required), +7 more.
invariance_metrics_agentsARead-only
Per-agent usage rollup over a time window: run counts, node counts, cost. Useful for agent-by-agent comparison.
Inputs: window_hours.
invariance_metrics_overviewARead-only
Cross-run rollup over a time window: total runs, nodes, errors, cost, latency, etc. Use this to ground "what is happening across my agents" questions.
Inputs: window_hours.
invariance_monitor_createA
Create a monitor that evaluates an event-shaped predicate against nodes/runs and optionally emits signals, findings, or reviews when matched.
Inputs: body (required).
invariance_monitor_evaluateA
Manually evaluate a monitor right now against an explicit input scope (returns the resulting execution plus any signals/findings/reviews produced).
Inputs: id (required), input.
invariance_monitor_executionsCRead-only
List past evaluation executions for a monitor (each has status, trigger, matched_node_ids, timing).
Inputs: id (required), limit, cursor.
invariance_monitor_findingsARead-only
List findings produced by a monitor across all of its executions.
Inputs: id (required), limit, cursor.
invariance_monitor_getBRead-only
Get a monitor by ID
Inputs: id (required).
invariance_monitor_listARead-only
List monitors visible to the calling agent (paginated).
Inputs: limit, cursor.
invariance_monitor_pauseA
Disable a monitor so it stops firing (preserves the spec; use invariance_monitor_resume to re-enable).
Inputs: id (required).
invariance_monitor_preview_evaluatorARead-only
Dry-run a monitor evaluator against history to see which nodes it would match and why. Writes nothing. Returns {sampled, matched, matches:[{run_id,node_id,matched,reason,observed_value?}]}.
Inputs: body (required).
invariance_monitor_preview_targetARead-only
Dry-run a monitor target against history to see which runs/nodes it would inspect. Writes nothing. Returns {run_ids, node_ids, counts:{runs,nodes}, truncated}.
Inputs: body (required).
invariance_monitor_resumeA
Re-enable a paused monitor so it begins firing again.
Inputs: id (required).
invariance_monitor_updateA
Patch an existing monitor (partial update; only included fields change).
Inputs: id (required), patch (required).
invariance_narrative_getARead-only
Fetch (or regenerate) the LLM-synthesized narrative for a run
Inputs: run_id (required), refresh.
invariance_node_listBRead-only
List nodes for a run in append order (paginated).
Inputs: run_id (required), limit, cursor.
invariance_node_writeA
Append a single node (one unit of work) to an open Invariance run. Use this to record tool calls, LLM calls, logs, context attachments, or handoffs as they happen.
Inputs: run_id (required), action_type (required), type, error, +4 more.
invariance_operator_createA
Create a new operator in one of the caller's projects. Operators are the actors whose work shows up in the company brain — create operator_type='agent' for an autonomous worker (Claude Code, a scripted agent, a coding bot) and operator_type='human' for a teammate whose screen recordings, microphone capture, meetings, and Granola notes you want to ingest. Requires a user-session JWT bearer.
Inputs: name (required), project_id (required), operator_type (required).
invariance_operator_getARead-only
Fetch a single operator by ID. Requires a user-session JWT bearer.
Inputs: id (required).
invariance_operator_listARead-only
List operators inside one of the caller's projects. Filter by operator_type to find all human teammates or all autonomous agents. Requires a user-session JWT bearer.
Inputs: project_id (required), operator_type.
invariance_operator_meARead-only
Show the operator identity associated with the current credentials. An "operator" is the unified actor model — every Claude Code session, autonomous agent, AND human teammate is an operator. Use this to confirm which operator context the MCP server is acting as (e.g. before recording session events, attaching runs, or writing notes to the company brain).
Inputs: no named parameters.
invariance_receipt_batchB
Record many external receipts in one call. Requires an AGENT API key (operator tokens get 403).
Inputs: receipts (required).
invariance_receipt_createA
Record one external receipt (proof an external action happened). Requires an AGENT API key (operator tokens get 403).
Inputs: receipt (required).
invariance_receipt_getBRead-only
Get an external receipt by ID.
Inputs: id (required).
invariance_receipt_listBRead-only
List external receipts. Filter by run, node, source, kind, external_id, business object.
Inputs: kind, limit, cursor, run_id, +5 more.
invariance_recipe_getARead-only
Get a recipe by ID or slug.
Inputs: id (required).
invariance_recipe_listARead-only
List built-in operational-check recipes (registry of controls). Promote one into a guardrail via invariance_guardrail_create.
Inputs: limit, cursor.
invariance_recipe_updateB
Patch a recipe (enabled, default_mode).
Inputs: id (required), enabled, default_mode.
invariance_review_claimA
Claim a pending review for the calling agent — sets status to "claimed" so other agents do not pick it up. Pair with invariance_review_resolve when done, or invariance_review_unclaim to release.
Inputs: id (required), notes.
invariance_review_getARead-only
Get a review by ID.
Inputs: id (required).
invariance_review_listARead-only
List reviews (work items requesting agent/human adjudication of a finding or run) in the queue, paginated.
Inputs: limit, cursor.
invariance_review_resolveA
Close a review by recording a decision: "passed" (looks good, no action), "failed" (issue confirmed, should not ship), or "needs_fix" (issue confirmed, fix-and-retry).
Inputs: id (required), decision (required), notes.
invariance_review_unclaimA
Release a previously-claimed review back to "pending" so another agent can pick it up. Does not record a decision — use invariance_review_resolve for that.
Inputs: id (required), notes.
invariance_run_failA
Close a run with failure — sets status to "failed" and stores the optional error string in metadata.error. Use this when the agent aborted due to an exception or unrecoverable error. For successful completion, use invariance_run_finish.
Inputs: id (required), error.
invariance_run_finishA
Close a run successfully — sets status to "completed". Use this when the agent finished its work without errors. For failures, use invariance_run_fail instead.
Inputs: id (required).
invariance_run_forkA
Fork a run from a specific node — creates a new run that branches off the parent at from_node_id. Useful for "what-if" replays during agent debugging.
Inputs: id (required), from_node_id (required), name, metadata.
invariance_run_getARead-only
Get details of an Invariance run (status, metadata, aggregate counts, timestamps).
Inputs: id (required).
invariance_run_inspectARead-only
Composite triage view for a run — fetches run, metrics, narrative, recent nodes, and open findings in parallel and returns {run, metrics, narrative, recent_nodes, open_findings}. Mirrors inv run inspect. Best first call when an agent is asked to debug a run.
Inputs: id (required), limit.
invariance_run_listBRead-only
List runs visible to the calling agent in reverse-chronological order (paginated).
Inputs: limit, cursor.
invariance_run_llm_callsARead-only
List LLM calls for a run in append order (paginated). Each entry includes model, tokens, cost, latency, and the underlying node_id.
Inputs: run_id (required), limit, cursor.
invariance_run_metricsBRead-only
Aggregate metrics for a run: total_input_tokens, total_output_tokens, total_cache_read/write, total_cost_usd, llm_call_count, tool_call_count, error_count, total_latency_ms.
Inputs: id (required).
invariance_run_node_type_metricsARead-only
Aggregate metrics for a single typed-node kind within a run (counts, latency stats, custom-field roll-ups).
Inputs: type (required), run_id (required).
invariance_run_node_typesARead-only
List the typed-node kinds present in a run (one row per registered type with a count). Pair with invariance_run_node_type_metrics for per-type aggregates.
Inputs: run_id (required).
invariance_run_operational_graphBRead-only
Get the operational graph for a run — entities, edges, findings, a completeness score (business_object_linked, policy_context_found, owner_found, approval_context_found, downstream_state_change_found), and a missing_evidence list naming the unsupported dimensions.
Inputs: run_id (required).
invariance_run_startA
Start a new Invariance run (the container for a sequence of nodes). The returned run is in status "open" — you must close it later with invariance_run_finish (success) or invariance_run_fail (error).
Inputs: name, case_id, metadata, tenant_id, +1 more.
invariance_run_verifyARead-only
Verify the cryptographic proof chain for a run — recomputes node hashes and Ed25519 signatures end-to-end. Returns {valid, node_count, head_hash, first_invalid_node_id, reason}.
Inputs: id (required).
invariance_saved_view_createC
Create a saved query view over executions/events/runs/nodes/captures.
Inputs: name (required), spec (required), source (required), viz, +1 more.
invariance_saved_view_deleteBDestructive
Delete a saved view by ID.
Inputs: id (required).
invariance_saved_view_getARead-only
Get a saved view by ID.
Inputs: id (required).
invariance_saved_view_listARead-only
List saved query views (name, source, spec, viz, visibility).
Inputs: no named parameters.
invariance_saved_view_runA
Run a query and return the result. Pass EITHER saved_view_id OR source+spec (exactly one).
Inputs: spec, source, saved_view_id.
invariance_saved_view_updateB
Patch a saved view (partial; only included fields change).
Inputs: id (required), viz, name, spec, +2 more.
invariance_session_append_noteA
Append a freeform text note to an existing session as a custom event with payload {text}. USE THIS WHEN: jotting a thought during a Claude Code task ("trying approach X next"), capturing a meeting takeaway, annotating a screen recording, or recording a partial transcript chunk from microphone capture. The note becomes part of the company brain timeline for that session.
Inputs: text (required), session_id (required).
invariance_session_attach_runA
Attach an existing run to an existing agent-session (PATCH). USE THIS when a Claude Code task that started a session later starts producing a run — call this to link the run's graph back to the session timeline so the brain can correlate transcript/notes with operational nodes.
Inputs: run_id (required), session_id (required).
invariance_session_createA
Open a new agent-session — the canonical container for an operator's bounded chunk of work in the company brain. CALL THIS at the START of: a new Claude Code task (source='api'), a screen recording for a teammate (source='screen_recording'), a mic capture session (source='microphone'), a meeting (source='meeting'), ingestion of a Granola note (source='granola_note'), or a manual note-taking session (source='manual_note'). Events (transcript chunks, tool calls, screenshots, notes) are appended to this session via invariance_session_append_note or the events sub-route. Link a session to a run via agent_id+run_id (or call invariance_session_attach_run later).
Inputs: source (required), external_session_id (required), title, run_id, +3 more.
invariance_session_getARead-only
Fetch a single agent-session by ID, including its source, timestamps, attached run/agent, and metadata.
Inputs: id (required).
invariance_session_listARead-only
List agent-sessions, optionally filtered by source, agent, run, or status. Use this to find all Claude Code work for an agent (source="api"), all meetings ingested today (source="meeting"), or all screen recordings for a human teammate.
Inputs: run_id, source, status, agent_id.
invariance_session_record_summary_to_kbA
Persist a summary of an agent-session as a knowledge-base page (so it becomes searchable, durable company brain content beyond the raw session timeline). USE THIS at the END of a Claude Code task, after a meeting wraps, or once a screen-recording has been reviewed — to capture the takeaways. The KB page is created under path 'sessions/<session_id>' by default.
Inputs: body (required), title (required), session_id (required), path, +1 more.
invariance_signal_acknowledgeA
Acknowledge a signal — moves status from "open" to "acknowledged" (someone has seen it). Use invariance_signal_resolve to mark fully resolved.
Inputs: id (required).
invariance_signal_emitA
Emit a manual signal (alert/notification) — typically attached to a run/node and used to flag noteworthy events for review or downstream automation. severity defaults to "info" if omitted. run_id/node_id auto-fill from INVARIANCE_RUN_ID/INVARIANCE_NODE_ID env vars when present.
Inputs: title (required), data, type, run_id, +3 more.
invariance_signal_getBRead-only
Get a signal by ID.
Inputs: id (required).
invariance_signal_listARead-only
List signals visible to the calling agent (paginated, newest first).
Inputs: limit, cursor.
invariance_signal_resolveA
Resolve a signal — moves status to "resolved" (the underlying issue has been addressed).
Inputs: id (required).
invariance_verify_runD
Alias of invariance_run_verify
Inputs: id (required).
invariance_workflow_createB
Create a workflow definition with typed fields, expected steps, allowed outcomes, and custom metrics.
Inputs: key (required), display_name (required), description, custom_metrics, +3 more.
invariance_workflow_deleteADestructive
Delete a workflow definition. Existing cases/runs/events are retained.
Inputs: key (required).
invariance_workflow_event_createA
Record a semantic workflow event on a case (e.g. "refund.issued", "approval.approved", "human.handoff"). Links case/run/node evidence + external refs and is bridged into DNA. Pass idempotency_key to make external retries safe.
Inputs: type (required), case_id (required), payload, actor_id, +5 more.
invariance_workflow_event_listBRead-only
List semantic workflow events across cases. Filter by case, workflow, tenant, actor, type, or time window.
Inputs: to, from, type, limit, +7 more.
invariance_workflow_getARead-only
Get one workflow definition by workflow key.
Inputs: key (required).
invariance_workflow_listARead-only
List workflow definitions: typed fields, expected steps, allowed outcomes, and custom metrics.
Inputs: no named parameters.
invariance_workflow_observability_executionsARead-only
List per-execution health for a workflow (status, stale flag, health, reasons, evidence mix, cost/tokens).
Inputs: workflow_key (required).
invariance_workflow_observability_getBRead-only
Get the observability rollup for one workflow by key.
Inputs: workflow_key (required).
invariance_workflow_observability_listARead-only
List workflow observability rollups (per workflow_key: execution/open/closed counts, evidence mix, cost & token totals).
Inputs: no named parameters.
invariance_workflow_updateA
Patch a workflow definition. Existing cases/runs/events keep their workflow_key.
Inputs: key (required), description, display_name, custom_metrics, +3 more.
invariance_write_nodeD
Alias of invariance_node_write
Inputs: run_id (required), action_type (required), input, output.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 159 tools
Several tools are explicit aliases (list_runs/run_list, create_run/run_start, write_node/node_write) and others intentionally overlap (capture_link vs capture_update, cortex_run_job vs cortex_run_eval/counterfactual, case_event_create vs workflow_event_create). These duplicates make it genuinely hard to know which tool to select despite mostly clear resource boundaries.
The dominant invariance_<resource>_<action> pattern is readable and mostly snake_case, but there are many deviations: aliases invert the order (list_runs vs run_list), noun-phrase names appear (invariance_monitor_findings, invariance_run_node_types), and the cortex_* family follows its own conventions. It is consistent enough to guess, but not predictable across a 159-tool surface.
159 tools is far beyond a well-scoped MCP server; the calibration guidance treats 50+ as an extreme mismatch. The count is inflated by aliases, overlapping wrappers, and many near-duplicate subdomain endpoints, making the toolset unwieldy regardless of individual tool quality.
Coverage is broad across runs, monitors, cases, evals, sessions, DNA, Cortex, and more, so core workflows exist. However, several subdomains have lifecycle dead ends: eval datasets/scorers/suites have create/read/append but no update/delete, agents/operators/captures lack update or delete paths, and monitors cannot be deleted. These are workable gaps but noticeable in a 159-tool API.
Maintenance
Related MCP Connectors
Errors, logs and AI-call costs for apps on Node.js, Next.js, Cloudflare Workers or the browser. Ask your AI what’s broken, why, and how to fix it, right inside Cursor or Claude.
AI agent observability for production traces, natural-language insights, and improvement loops.
Query your AI agent traces, root-cause verdicts and issues, and run investigations, from your agent.
AI code governance and code review for Claude Code, Copilot and Cursor. Checks rules and security.
Related MCP Servers
- AlicenseAqualityDmaintenanceConnects AI assistants to Warpmetrics telemetry data to monitor AI agent performance, execution runs, and LLM costs. It allows users to query success rates, latency, and spend metrics directly through natural language interfaces.2063 npmMIT
- AlicenseCqualityDmaintenanceEnables AI-powered automated testing, security scanning, code review, and maintenance tasks directly within Claude Code or desktop.1252 PyPI4MIT
- FlicenseNot gradedqualityDmaintenanceEnables AI-driven incident response by connecting Claude to monitoring tools like Prometheus, Grafana, Loki, PagerDuty, and Slack for automated investigation and runbook generation.9-

Confident AIofficial
AlicenseNot gradedqualityBmaintenanceConnects AI-powered tools to the Confident AI platform for evaluating, observing, and iterating on AI quality, enabling cloud evaluations, dataset management, prompt versioning, production tracing, and human annotations directly from editors like Cursor and Claude Code.2MIT