How is this different from connecting incident.io MCP directly?

Connected directly, an agent sees every tool the server exposes and holds the credential itself. Through TrustGate you choose which tools each agent gets, and every call is logged against the agent that made it.

What if the incident.io MCP API changes?

We track the incident.io MCP schema, so the tool catalogue stays current. Tools you have already allowed keep working, and anything new stays denied until you enable it.

Can I give different agents different incident.io tools?

Yes. Access is set per agent: one can get a handful of read tools, another the full set. You grant the whole MCP or single tools within it.

Where can I see what my agents did in incident.io?

Every call is logged against the agent that made it, with the tool it used. Activity can be sent on to your own monitoring.

Do I need to run the incident.io MCP server myself?

No. TrustGate connects to the incident.io MCP server and sits in front of it. Your agent only talks to your TrustGate URL, never to incident.io directly.

Connect incident.io MCP to your AI agents, securely.

Q: Is incident.io MCP secure?

The server itself exposes 91 tools, and some of them can write to your incident.io account. Behind TrustGate an agent only sees the tools you allow, requests and responses are inspected, and every call is logged. Anything outside the allowlist is denied before it reaches incident.io.

One secured endpoint for every incident.io call your AI agents make, scoped per user, inspected, and logged.

Get started for free

Get a demo

Try incident.io MCP now

What your agents can ask once incident.io MCP is connected.

Select the tools your AI agents can call

91 tools from the incident.io MCP schema.

Showing 91 tools

incident_listreadSearches and browses incidents with filters, sortable by workload.
incident_showreadFull incident details, optionally including the AI investigation findings, the post-mortem and the status update history.
incident_statsreadAggregate incident counts and workload across 10 or more dimensions: severity, status, type, mode, alert source, custom fields, roles and time periods, with breakdowns for working hours, late evening and overnight.
incident_update_listreadThe full status update history for an incident.
incident_updatewriteUpdates incident fields: status, severity, roles and custom fields.
incident_messagewritePosts a finding into an incident's channel, visible to everyone responding.
status_page_updatewriteDrafts a status page update for a human to publish. It never publishes, because publishing emails and texts everyone subscribed to the page.
incident_createhigh riskCreates a new incident, which starts the real response process.
incident_mergehigh riskMerges incidents that turn out to be the same problem.
incident_unmergehigh riskReverses a merge, restoring an incident as standalone.
alert_listreadSearches and browses alerts.
alert_showreadFull alert details with the incidents linked to it.
alert_source_listreadLists the configured alert sources.
alert_statsreadAlert counts with the workload from linked incidents, so alert noise can be traced through to responder cost.
alert_tag_listreadLists the alert tags available to apply.
alert_attachwriteAttaches an existing alert to an incident.
alert_detachwriteDetaches an alert from an incident.
alert_tag_applywriteAdds, removes or replaces tags on up to 50 alerts in one call.
alert_create_incidenthigh riskDeclares an incident from one or more alerts.
alert_resolvehigh riskResolves one or more firing alerts, silencing a live signal.
escalation_listreadSearches escalations, meaning pages that have been sent.
escalation_showreadFull escalation details with its transition history.
escalation_path_listreadLists the configured escalation paths.
escalation_path_showreadShows who would be paged at each level of an escalation path.
escalation_statsreadPaging counts by path, priority and time of day.
escalation_respondwriteAcknowledges or declines a page.
escalation_createhigh riskPages a user or an escalation path, reaching a real person's phone.
schedule_listreadLists the on-call schedules.
schedule_showreadSchedule details with current and upcoming shifts.
cover_request_listreadLists cover requests and how the candidates responded.
cover_request_createwriteAsks for volunteers to cover a shift.
cover_request_respondwriteResponds to a cover request the authenticated user is a candidate on.
cover_request_managewriteCancels a cover request, accepts an offer, or reminds candidates.
schedule_override_createhigh riskPuts someone on call for a period, or clears the schedule entirely for that period.
schedule_override_deletehigh riskRemoves an override, restoring the underlying rotation.
maintenance_window_listreadLists maintenance windows by status.
maintenance_window_alertsreadCounts the alerts a maintenance window held back.
notification_pause_listreadLists notification pauses for a user or schedule.
maintenance_window_updatewriteChanges a window's name, lead or times, or ends one now.
maintenance_window_deletewriteCancels a maintenance window that has not started.
notification_pause_updatewriteChanges a notification pause's times or reason.
notification_pause_deletewriteCancels an upcoming pause, or ends an active one now.
maintenance_window_createhigh riskSchedules a window where matching alerts do not page anyone.
notification_pause_createhigh riskPauses a person's notifications so escalations skip them entirely.
action_listreadLists the actions on incidents.
follow_up_listreadLists post-incident follow-ups.
follow_up_statsreadFollow-up counts grouped by owner, status and more.
action_createwriteCreates an action on an incident.
action_updatewriteUpdates an existing action.
follow_up_createwriteCreates a follow-up on an incident.
follow_up_updatewriteUpdates an existing follow-up.
action_deletehigh riskDeletes an incident action.
investigation_syncreadDownloads a full investigation as an archive: all findings, checks, conversation history and evidence, for programmatic LLM analysis.
investigation_feedbackwriteRecords whether an investigation was helpful.
investigation_starthigh riskStarts an AI investigation for an incident that nothing is investigating yet. It posts a message into the incident channel and invites the caller to it.
investigation_steerhigh riskFeeds information into a running investigation so it re-assesses its hypothesis. The steer shows up in the investigation's history.
ask_telemetryreadAn AI agent that queries logs, metrics, traces and dashboards across connected observability platforms, planning queries, handling pagination and time ranges and synthesizing results. Requires Investigations.
askhigh riskAn AI agent for on-call queries, schedule management and general questions. A single tool name covering an open-ended set of operations.
ask_incidenthigh riskAn AI agent for incident investigation and management actions. A single tool name covering an open-ended set of operations.
telemetry_inspectreadDiscovers the real labels and metric names on a connected datasource. Requires Investigations.
document_searchreadSearches the runbooks, guides, post-mortems and architecture notes synced into Investigations from sources like GitHub, Notion and Confluence, by meaning or by specific terms. Requires Investigations.
document_showreadReads a synced document in full, so an assistant can follow a runbook's steps rather than paraphrasing a summary. Requires Investigations.
duration_metrics_statsreadTime to acknowledge, time to resolve and other duration metrics with percentiles. Requires Response.
time_spent_statsreadResponder hours by person, period or custom field. Requires Response.
analysis_startreadStarts a structured operational analysis, downloading a workspace with playbooks (operational review, alert noise, on-call burden, team health, response effectiveness), the organisation's config and a branded HTML report template.
catalog_type_listreadLists catalog types such as Service and Team.
catalog_entry_listreadBrowses the entries in a catalog type.
catalog_entry_showreadFull catalog entry with all its attributes.
team_listreadLists the teams.
team_showreadTeam details with the escalation paths, alert sources and schedules it owns.
extension_connector_listreadLists connectors with their capabilities and connection health.
extension_plugin_listreadLists plugins with their skills and sync state.
extension_skill_feedback_listreadShows current feedback on a plugin's skills.
extension_skill_usage_listreadLists assessed skill loads with the run behind each one.
extension_verify_showreadShows a verification's status and result.
extension_plugin_syncwriteRe-syncs a plugin from its source repository.
extension_plugin_updatewriteEnables, disables, renames or changes which skills agents load.
extension_skill_feedback_updatewriteDismisses or reopens a skill feedback issue.
extension_verifywriteTests proposed extension content before landing it.
extension_upload_urlwriteGets a URL to upload content that is not in a readable repository.
extension_plugin_createhigh riskLoads a plugin from a source-control repository, bringing in the skills that agents will then follow as instructions.
resource_showreadReads organisation config, analysis playbooks or telemetry datasources. Calling it with resource organisation lists the severity levels, custom fields and roles needed for filtering and grouping.
resource_listreadLists the resources available to read.
custom_field_valuesreadLooks up the values a custom field accepts, and whether it can be set or is derived automatically.
external_service_listreadLists the third-party services whose status pages incident.io monitors, with the exact names to ask for.
external_service_statusreadChecks a monitored vendor's current status and recent changes, and says whether the vendor published it or incident.io inferred it from its own monitoring.
user_listreadFinds users by name or email.
help_queryreadAnswers questions about how incident.io itself works, from the product documentation.
nexus_health_showreadShows the health of the AI data sources that have been connected.
suggestion_listreadLists AI suggestions waiting on a human.
feedbackwriteSubmits feedback about the tools to incident.io.

What incident.io MCP exposes

incident.io MCP groups its 91 tools into 13 categories, each with a suggested starting policy.

All 91 tools as JSON, ready to paste.

Category	Covers	Access	Suggested policy
Extensions	Connectors and their health, plugins loaded from source-control repositories with their skills and sync state, skill feedback and usage, upload URLs, and verifying proposed extension content	Read / Write	extension_plugin_create loads a plugin from a repository, which is instructions the agents then follow. Treat it as a supply-chain control, not a convenience, and verify content before it lands
Alerts	Searching and showing alerts, alert sources, counts with workload, tags, attaching and detaching alerts from incidents, resolving firing alerts and declaring an incident from them	Read / Write	alert_resolve silences a live signal and alert_create_incident starts a real incident; alert_tag_apply touches up to 50 alerts in one call
Incidents	Creating, searching, showing and updating incidents, aggregate stats across 10 or more dimensions, status update history, posting findings into the incident channel, merging and unmerging, and drafting status page updates	Read / Write	Follow stats then list then show rather than paginating. status_page_update only produces a draft: publishing notifies every subscriber and stays with a person
Config & Meta	Organisation config and the resources available to read, custom field values, monitored third-party service status, users, AI data source health, pending AI suggestions, product documentation answers and tool feedback	Read / Write	Read resource_show with resource organisation before filtering or grouping, and custom_field_values before setting a field, since it also says whether a field is settable or derived
Maintenance & Pauses	Scheduling windows where matching alerts do not page, counting what a window held back, and pausing a person's notifications so escalations skip them	Read / Write	Both creation tools suppress paging, which is the one failure mode where nothing visibly breaks until an incident is missed. Require approval and bound the duration
Actions & Follow-ups	Creating, listing, updating and deleting incident actions, and creating, listing and updating post-incident follow-ups with counts grouped by owner and status	Read / Write	action_delete is the only destructive tool here; follow-ups have no delete
On-call & Cover	Listing schedules and their current and upcoming shifts, overriding who is on call for a period or clearing it, and asking for, responding to and managing shift cover requests	Read / Write	schedule_override_create changes who gets paged, including clearing the schedule entirely for a period. Approve it and check the window
Escalations	Paging a user or an escalation path, searching and showing escalations with transition history, listing paths and who they would page at each level, acknowledging or declining a page, and paging counts by path, priority and time of day	Read / Write	escalation_create reaches a real person's phone. escalation_path_show is the safe way to see who would be paged before doing it
Catalog & Teams	Catalog types such as Service and Team, catalog entries with their attributes, and teams with the escalation paths, alert sources and schedules they own	Read	Read-only. team_show is the quickest way to scope an analysis to one team's surface
Investigations	Starting an AI investigation on an incident, steering a running one's hypothesis, downloading a full investigation as an archive, and recording whether it was helpful	Read / Write	Both start and steer are visible to responders: starting posts into the incident channel, and steering appears in the investigation history. investigation_sync pulls the complete evidence set out as an archive
AI Agents	Three agent tools: one for on-call queries and schedule management, one for incident investigation and management actions, and one for querying logs, metrics, traces and dashboards	Read / Write	ask and ask_incident are agents that take actions, so a single tool name stands in for an open-ended set of operations. The tool name tells you nothing about what the call does
Telemetry & Documents	Discovering the real labels and metric names on a connected datasource, and searching and reading the runbooks, guides, post-mortems and architecture notes synced into Investigations	Read	Read-only, but synced documents come from GitHub, Notion and Confluence, so their content is third-party text; use document_show when the agent has to follow a runbook step by step
Analytics	Time to acknowledge and resolve with percentiles, responder hours by person, period or custom field, and starting a structured analysis that downloads playbooks, org config and a report template	Read	time_spent_stats attributes hours to named people, which is employee performance data in most jurisdictions

Connect incident.io MCP to your AI agent

Four steps, under ten minutes.

01
Sign in to NeuralTrust
Open the NeuralTrust platform, sign in and go to Getting started, then Connect your tools.
02
Choose your agent
When asked which agent you use, pick Claude Code, Codex, Cursor, Windsurf, Copilot or your own code.
03
Connect incident.io
Find incident.io in the tool list and connect it.
04
Connect your agent
Follow the steps on screen to add the config to your agent, then send the test prompt.

Permissions down to the single tool

You do not allow incident.io MCP or block it. You allow the exact tools each agent can call, one by one.

incident.io MCP	91 TOOLS AVAILABLE
Code review agent	4 tools, read only
Issue triage bot	9 tools
Platform team assistant	all 91 tools

Tool by toolConnecting the server is not the unit of access. Each tool is granted on its own.
Default denyA tool you have not enabled is never exposed to the agent.
Approval on writesTools that change or delete things can require a human to confirm.

Is incident.io MCP secure?

One endpoint instead of raw credentials. Every call is checked against policy and logged.

AI agentsConnect through your MCP client configuration.	NeuralTrustApply controls based on your configuration.	incident.io MCP91 documented tools. Some require the Response, On-call or Investigations product: they still appear in the tool list but return an error explaining what the account is missing. Team, Pro and Enterprise plans only.
Coding assistants	Connection & credentialsper user	Extensions11 tools
Internal copilots	Tool access policiesdeny by default	Alerts10 tools
Workflow automations	Request & response inspectioninline	Incidents10 tools
Compatible MCP clients	Activity monitoringalerts and logs	All other tools60 tools