titan_run_template is the escape hatch. It runs an allowlisted Titan template directly, giving you access to template-specific payload fields and proxy location control that the dedicated tools do not surface.
For most work, prefer titan_search, titan_fetch, or titan_crawl—they run the same templates with parameters shaped for agents.
Required scope: mcp:templates:run
Parameters
string
required
Slug of an allowlisted template. Anything outside the allowlist returns
template_not_found. List valid slugs with titan_list_templates.string[]
Target URLs, up to 100. Required for templates that take static URLs, which is all eight current templates.
object
Template-specific options. Valid keys are defined by the template’s
input_schema.integer
Convenience shortcut merged into
payload as max_results.string[]
Geographic locations to route the run through. Useful when a target serves different content by region.
boolean
default:"true"
Wait for results inside the call.
integer
default:"30"
Seconds to wait, capped at 30.
string
Deduplicate repeated calls for 24 hours. See Idempotency and retries.
Example
Run the generic page extraction template with options the dedicated fetch tool does not expose:Response
Records come back as raw template output rather than a tool-specific shape, because the shape is whatever the template’soutput_schema defines.
template object echoes which template ran, so an agent handling several templates can branch on it without tracking state.
Allowlist enforcement
Only the eight templates in the MCP allowlist are runnable. User-owned, admin, inactive, and experimental templates return:Credits
Credit estimates follow the template’s capability:
Final billing is always based on records actually delivered. See Credits and usage.
When to use this instead of a dedicated tool
Next steps
- titan_list_templates — discover slugs and their schemas
- titan_get_run — poll runs that outlive the wait window