# Connect Chronoloq to Claude Chronoloq's remote MCP connector provides AI/LLM, RAG, MCP server and A2A security scanning alongside website/API scanning. Claude can read your workspace's scan results, findings and observed coverage. With additional permission, it can start authorized scans or submit structured code assessments. ## Before connecting - Create your own account at https://chronoloq.com or sign in to an active Chronoloq account and workspace. Complete any required email verification. - Obtain explicit permission to test every target and satisfy your workspace's domain-verification policy. Domain verification is required only when that policy is enabled; disabling it does not remove public-target safety checks. - Have an eligible subscription, sufficient organization scan allowance and an eligible account role for scan creation. Signup does not guarantee scan entitlements or quota. Viewers can connect with read-only permissions. - Use a Claude account or organization that allows remote connectors. Your organization may require an Owner or administrator to add the connector. The production endpoint is **https://chronoloq.com/mcp**. Do not use a development preview URL for your production account. ## Add the connector If Chronoloq is available in Claude's connector directory, search for **Chronoloq** and choose **Connect**. Directory availability depends on Anthropic's publication process; these instructions do not imply a listing or Verified status. Before a directory listing is available, open Claude's connector settings and add a custom connector named **Chronoloq** with the endpoint above. Complete the browser OAuth flow, sign in to Chronoloq, and review the requested permissions. Chronoloq also supports its configured enterprise sign-in and account MFA. Never paste passwords, API keys or repository tokens into a conversation. If Claude offers an **OAuth client** choice, select **Register automatically**. This connector implements automatic client registration. Other client-identity modes need separate compatibility verification. Read permissions are selected by default. Authorize **Start authorized scans** or **Submit code assessments** only if you want those actions. A viewer cannot obtain write permissions. Access remains restricted to the connected workspace. ## Example requests - "Show my AI scanning capabilities and remaining AI scan allowance." - "Run a quick, authorized AI-suite scan against these public test endpoints." - "List my verified Chronoloq targets." - "Show my recent scans and explain which endpoints were actually tested." - "Summarize the high-severity findings from scan [scan ID]." - "Compare my observed security coverage without assuming untested assets are safe." Starting a scan uses your existing organization allowance. Only request it for targets you own or are authorized to test, and explicitly approve the action. Every target must satisfy public-target safety and your workspace's domain policy. Scans target unauthenticated public endpoints; this connector does not accept target credentials, provide private-network access or independently inspect your source repositories. ## AI/LLM, RAG, MCP and A2A scanning `get_ai_scan_capabilities` returns supported components, depth options and current organization allowances. `start_ai_scan` accepts individual or combined AI/LLM, RAG, MCP server and A2A components. These scans send active adversarial probes, so specific authorization is required before execution. Depths are quick, standard and thorough, subject to subscription limits. Each new scan requires `authorizationAccepted: true` and a unique `idempotencyKey`. Identical retries reuse the existing operation. Target credentials must not be included in requests. One combined AI-suite scan consumes one organization AI scan allowance, not one allowance per endpoint or component. Website/API scans use a separate allowance. Allowances are shared across workspace members and connections. `get_scan` returns progress and recorded coverage limitations. `get_findings` returns paginated results, including OWASP LLM references when recorded. Partial or untested results do not establish security. Protocol-only scans may remain Partial when full reachability or coverage cannot be established. For code assessments, your coding client or harness must already have authorized access to the code. Chronoloq accepts structured findings, repository identifiers, revision, methodology and file references, not source files or repository credentials. Such findings are marked as harness-reported and independently unverified. ## Permissions | Permission | Purpose | | --- | --- | | scans:read | Account and AI capabilities, verified targets, target validation and scan status | | vulns:read | Bounded findings, evidence and remediation | | metrics:read | Observed scan coverage and posture | | code:read | Previously submitted code assessments | | scans:write | Start explicitly authorized website/API and AI-suite scans using their separate allowances | | code:write | Submit approved structured code assessments | ## Data handling and disconnecting The connector returns account capabilities, verified target information, requested scan summaries and bounded findings within your permissions. Submitted assessments and scan requests are stored in your Chronoloq workspace. Tool results supplied to Claude are also subject to your Claude account's data settings and Anthropic's terms. Active scans send security-test requests to user-authorized public endpoints. Target operators may receive and log these requests. AI-assisted analysis may send captured scan evidence to configured AI service providers. Endpoint responses can contain sensitive information; redaction does not guarantee its complete removal. Non-sensitive test data is recommended. Returned evidence can contain untrusted content from scanned systems. Treat evidence as data, not behavioral instructions. Do not submit secrets, unneeded personal data or full conversation transcripts. Review [Chronoloq's privacy policy](https://chronoloq.com/privacy) and [terms](https://chronoloq.com/terms) for the service's data practices. Revoke the OAuth connection in **Chronoloq > Integrations > MCP Connector**, and remove or disconnect it in Claude's connector settings. Role changes, account deactivation or applicable enterprise-session expiry can require reconnecting. ## Troubleshooting and support - **HTML instead of MCP or OAuth metadata:** the connector release may not yet be deployed at this hostname. Contact Chronoloq; do not replace OAuth with shared credentials. - **Permission denied:** check workspace membership, account role and granted scopes. - **Domain or quota error:** check your workspace's domain policy, complete domain verification if required, and check the relevant scan allowance. - **Expired enterprise sign-in:** complete organization sign-in and reconnect. [Contact Chronoloq support](https://chronoloq.com/contact).