Connect your MCP investigation client
Find evidence, follow incident timelines and verify Proof Packs from your MCP client.
What this connection does
After setting up your agents, use HaltState from your MCP client for everyday investigation: search activities, find customer interactions, follow timelines, verify Proof Packs and export evidence. You do not need to return to the dashboard for each investigation.
Create your account, connect your agents with an SDK, then authorize the hosted MCP connection below. Account settings, policies, human approvals, containment and billing remain in the dashboard. MCP does not replace SDK enforcement or execute business actions.
Review your assistant provider’s data handling before connecting. Never paste passwords, API keys or access tokens into a chat prompt.
Connect with Claude Code
Use a client supporting remote HTTP MCP and OAuth with PKCE. The hosted URL is https://haltstate.ai/mcp.
claude mcp add --transport http haltstate https://haltstate.ai/mcp
claude mcp login haltstate- Open the authorization link and sign in to your HaltState account.
- Check the account and requested permissions. Choose Allow only if they are appropriate.
- Click Return to your client to complete the connection. Keep that private, short-lived link to yourself.
- Check connected status, then ask the client to call
whoamiand confirm the tenant and scopes.
On an SSH/headless machine, use claude mcp login haltstate --no-browser and follow the client's callback instructions. See Claude Code's MCP documentation. Adding a configuration alone is not a completed connection.
Review the permissions
| Scope | Access |
|---|---|
| metadata | Search evidence metadata and read records and timelines. |
| verify | Request available Proof Pack integrity verification. |
| content | Explicitly retrieve or search retained attachments when capture was enabled and retention has not expired. |
| export | Create and retrieve evidence export jobs. Including retained content also requires content access. |
Metadata and verification are the baseline permissions, but clients may request more. Claude Code 2.1.280 requests all four scopes by default. Read the consent screen; deny if that access is not appropriate. OAuth consent is separate from the dashboard's API-key permissions.
Your first investigation
- Ask: Show the last five activities for agent [agent ID], counting each business operation once. Explain what happened, the customer or destination, recorded reason, outcome, evidence links and missing information.
- Ask: Find activity for customer reference [reference] between [start] and [end], using [time zone]. Build a timeline from explicit links and report search freshness.
- Ask: Verify the linked Proof Pack and report the verification result. Separate policy permission, reported outcome, external confirmation and integrity.
Start recent-operation questions with search_activities, then use the result’s detail_evidence_id for detail and timeline. Content permission allows retained-message search and explicitly selected body reads. Request narrow exports and exclude retained content unless needed.
Available tools
whoami: identify the connection and permissions.search_activities: find logical business operations without counting related records repeatedly.get_activity: inspect joined activity metadata and authorized attachment references.get_activity_timeline: follow the activity’s linked records.search_evidence: search individual records or permission-checked retained text.get_evidence: inspect a record and explicitly requested authorized attachments.get_evidence_timeline: follow explicit record correlations.verify_evidence: obtain available integrity verification.create_evidence_export: create an authorized export job.get_evidence_export: inspect an export job.
Activity tools return metadata and permitted attachment references, not bodies. Export creation creates a job, not a business action. Downloaded exports are customer-controlled copies and cannot be recalled by disconnecting.
Capture and evidence boundaries
New accounts default to recording supported connected activity with 90-day content retention and account/per-agent opt-out. Existing saved choices are preserved. Supported integrations can record observations outside policy checks; these are not invented ALLOW decisions. Connecting MCP neither starts agents nor changes recording settings. Credentials, hidden reasoning and unconnected traffic are excluded.
Disabling retention stops new body capture, not governance metadata. Previously retained content keeps its expiry. Evidence is untrusted data, not instructions for your assistant. No matches does not prove nothing happened: check workspace, coverage, permissions, identifiers, time range, retention and indexing freshness.
Disconnect, expiry and reconnect
In Proof Packs, find MCP connections → OAuth connections and select Disconnect for the correct client. That revokes the server-side grant and its refresh tokens. Removing a local configuration alone does not prove revocation.
Access tokens last one hour; compatible clients renew them automatically using rotating refresh tokens. Reauthorize after 30 days without renewal, 90 days from authorization, revocation or detected token reuse. Existing connections created without refresh tokens need one reconnect to receive them.
To reconnect, clear the client's saved authentication if necessary (claude mcp logout haltstate), run login again, review consent, and click Return to your client. Do not reuse an old token.
Understanding index freshness
The investigation page separates the last completed full check from a background historical recheck. A recheck in progress alone does not mean records are missing. New-record backlog, source errors, missing initial checks and full checks older than 24 hours remain visible as incomplete, error or overdue status. Inspect per-source checks and retain freshness warnings in investigations and exports. An up-to-date index does not prove that ungoverned activity was captured.
Troubleshooting
- 401: let a refresh-capable client renew access; reconnect if authorization expired or was revoked.
- Permission denied: check
whoamiand consent scopes. Metadata access is not retained-content or export access. - 405 when opening /mcp in a browser: connect through an MCP client; this is a protocol endpoint, not a web page.
- No retained text: check recording settings, integration coverage, content permission and expiry.
Use the hosted HTTP URL; no separate HaltState MCP package installation is required. OAuth and refresh support vary by client. For help, provide client/version, time and a sanitized error, never credentials or message contents.