eqho

Create a knowledge base. data accepts name (required), description and retrieval_config; description is not searchable content. Requires upstream Admin plus MCP config write. org_id overrides this call only and requires access; session organization is unchanged.

create_knowledge_base

Creates 1 campaign record. Dials nothing until status is ACTIVE and a phone number, agent and lead list are attached. Required in data: name. status defaults to CREATED when omitted (CREATED | ACTIVE | PAUSED | INACTIVE | PROCESSED). For outbound telephony this server requires rules.call_window (a campaign without one is never selected by the dialer). For inbound campaigns set inbound_lead_list or inbound calls fail. Read help(topic="campaign_schema") for task and condition shapes. Creating an ACTIVE campaign needs human approval. On the Claude connector (stateless HTTP) this action is declined with instructions to use the EQHO app; approval prompts work in MCP clients that support elicitation over a streaming transport. Args: data: Campaign configuration (name, status, agents, dispositions, lead_lists, phone_numbers, call_window/rules, tasks, custom_fields). org_id: Optional organization override for THIS call only. Superadmins may target any org; other users only orgs they can access (anything else raises a 403). The session's active org is not changed.

create_campaign

Creates 1 lead list (empty). Attach it to a campaign with campaign_id so uploaded leads become dialable by that campaign. Args: name: List name. campaign_id: Campaign to attach the list to. org_id: Optional organization override for THIS call only. Superadmins may target any org; other users only orgs they can access (anything else raises a 403). The session's active org is not changed.

create_lead_list

Creates 1 role record (agent behaviour instructions). Nothing goes live until an agent references the role. Use {{lead.first_name}}, {{agent.name}} style variables. Complex multi-step agents (qualification, booking, triage) work best with a structured workflow prompt; see help(topic="role_prompt"). Args: data: {name, description (the instructions)}. org_id: Optional organization override for THIS call only. Superadmins may target any org; other users only orgs they can access (anything else raises a 403). The session's active org is not changed.

create_role

Creates 1 agent record. No calls are placed. Required in data: name, voice_audio_id (list_voices), role_ids (list_roles), avatar ("MAN" or "WOMAN"; an agent without one 500s on a test call and cannot run voice engine v3), personality {adherence, empathy, humor} (0-1), and ai_settings.ai_model (the API rejects an agent without one: "AI model must be specified"). Optional: language (lowercase), other ai_settings fields, structured_greetings, action_ids, script_ids, knowledge_base_ids. For custom system_prompt.inbound/outbound, each supplied object needs sections and medium (e.g. "telephony" for phone calls, "sms" for texts). Omit system_prompt to use the default prompt. Explicit Realtime instructions must be nonblank; omission deliberately uses runtime defaults. Use realtime_blueprint for a persona draft. Legacy data_store_ids is normalized to knowledge_base_ids; conflicting aliases and invalid attachment IDs are refused. No voice/model/engine or prompt is silently replaced. Inspect metadata.agent_diagnostics. Args: data: Agent configuration. org_id: Optional organization override for THIS call only. Superadmins may target any org; other users only orgs they can access (anything else raises a 403). The session's active org is not changed.

create_agent

Delete an IVR flow (irreversible). Requires server human approval; clients without approval support fail closed. org_id overrides this call only and requires access; session organization is unchanged.

delete_ivr_flow

Delete a knowledge base and its content (irreversible). Requires upstream Manager+ plus MCP config write. Obtain user authorization first. org_id overrides this call only and requires access; session organization is unchanged.

delete_knowledge_base

Delete a knowledge source and its indexed content (irreversible). Requires upstream Admin plus MCP config write. org_id overrides this call only and requires access; session organization is unchanged.

delete_knowledge_base_source

Disconnect and end your current session. This will: - End your current session - Clean up session resources - Clear your connection context After disconnecting, you'll need to call connect() again to use other tools. Returns: Confirmation of disconnection Examples: # End your session disconnect()

disconnect

Discover what this EQHO MCP server can do. This is the recommended first call when connecting to the EQHO platform. It returns capabilities, **how to navigate EQHO** (connect → context → switch organization → data tools), and connection instructions. The discovery tool will: - Show you what features are available - Include a **navigation** section: org switching, get_context, switch_organization - Explain how to connect and get started - Provide quick start workflows for common tasks Args: query: Optional natural language query to filter capabilities. Examples: "How do I make calls?", "campaign management", "reporting and analytics" Returns: System capabilities, connection instructions, and getting started guide Examples: # Discover everything await discover_capabilities() # Specific capability await discover_capabilities(query="How do I access call recordings?")

discover_capabilities

Prepare a local Realtime two-prompt draft; creates nothing and never dials. Common Workflows: guide -> draft -> review -> create_role/create_agent. Examples: realtime_blueprint(data={name: ..., company: ..., role: ..., goal: ..., persona: ..., emotional_switch: ..., backend_model: ..., business_rules: ...}). Available Actions: local draft only. Required brief: name, company, role, goal, persona, emotional_switch, backend_model and business_rules. Optional voice_audio_id is an actual org voice row ID, not a guessed voice name. Review unresolved questions and suggested defaults. startup_history is analyzed only, never saved. Estimates do not prove provider acceptance or runtime readiness.

realtime_blueprint

List industry briefs or draft a Realtime build; creates nothing and never dials. Common Workflows: list -> brief -> gather approved facts/persona -> draft -> review -> separately authorized create handlers -> saved configuration audit. Examples: agent_builder(operation='brief', industry='towing'). For draft, data uses realtime_blueprint's required name/company/role/goal/persona/ emotional_switch/backend_model/business_rules fields and optional fields. No hardcoded voice, disposition or resource IDs; no automatic price/ETA, campaign activation, partial-stack retry or live-readiness claim. Industry suggestions are not silently inserted into the supplied business rules.

agent_builder

Get one agent's full config: voice, personality, greetings, roles, actions, scripts, feature flags. Realtime budget/persona guidance is in metadata.agent_diagnostics, including compact tiers. Estimates are advisory; rendered prompts, greeting mapping and live-call execution remain unverified. Args: detail: summary (identity and status), core (working configuration), full (default; complete existing response including history). agent_id: Agent ID (from list_agents). org_id: Optional organization override for THIS call only. Superadmins may target any org; other users only orgs they can access (anything else raises a 403). The session's active org is not changed.

get_agent

Get one call. summary: status, duration, disposition, numbers. core: the full call record. full: adds the transcript turns and the recording URL (long calls = large response; page the transcript with transcript_offset/transcript_limit). Args: call_id: Call ID (from list_calls). detail: summary | core | full. transcript_offset: First transcript turn to include (full only). transcript_limit: Max transcript turns to include (full only). org_id: Optional organization override for THIS call only. Superadmins may target any org; other users only orgs they can access (anything else raises a 403). The session's active org is not changed.

get_call

Get the QA scorecard for a call: every grader's pass/fail, score and reasoning, plus the overall pass rate (empty when nothing graded yet). Args: call_id: Call to grade-check. org_id: Optional organization override for THIS call only. Superadmins may target any org; other users only orgs they can access (anything else raises a 403). The session's active org is not changed.

get_call_scorecard

Get one campaign: agents, phone numbers, dispositions, call window, tasks. Args: campaign_id: Campaign ID (from list_campaigns). detail: summary (id, name, status, agents, phone numbers), core (adds dispositions, call_window, rules, lead lists), full (everything incl. tasks and SIP config; large). mention_keys: Exact saved campaign.custom_fields keys to select; mention definitions are separate from these campaign values. Common Workflows: read the intended campaign and agent template, then select the actual saved mention keys before editing. Available Actions: read only; use update_campaign for mention edits. Examples: get_campaign(campaign_id="<ID>", mention_keys=["agent_greeting"]). fields: Explicit top-level fields to return instead of a tier (id/name always kept), e.g. ["tasks", "pre_call_tasks"]. org_id: Optional organization override for THIS call only. Superadmins may target any org; other users only orgs they can access (anything else raises a 403). The session's active org is not changed.

get_campaign

Get help and guidance based on your current context. This tool provides intelligent assistance by understanding where you are in your workflow and what you're trying to accomplish. It adapts its responses based on: - Whether you're connected - Which organization you're in - What tools you've recently used - Common next steps for your current state Args: topic: Specific topic or question (optional) Examples: "campaign management", "call recordings", "how do I switch organizations?" context_aware: Provide context-specific help (default: True) Returns: Relevant guidance, examples, and next steps Examples: # General help based on current state help() # Specific topic help(topic="How do I analyze campaign performance?") # Get all available topics help(context_aware=False)

help

Read one editable IVR draft and its node tree. org_id overrides this call only and requires access; session organization is unchanged.

get_ivr_flow

Read inbound routing configuration for a claimed number. This does not prove live answering. org_id overrides this call only and requires access; session organization is unchanged.

get_ivr_number_routing

Read a knowledge base, its sources and ingestion status. Use source IDs, not rag_file_id. org_id overrides this call only and requires access; session organization is unchanged.

get_knowledge_base

Read indexed chunks of a knowledge source, optionally filtered by query. org_id overrides this call only and requires access; session organization is unchanged.

get_knowledge_base_chunks

Read a knowledge-base revision without restoring or changing it. org_id overrides this call only and requires access; session organization is unchanged.

get_knowledge_base_revision

Get one lead in full by ID: names, timezone, custom_fields (CRM ids), extracted_data. Phone/email lookup returns the grouped search summary; use its returned ID with get_lead(lead_id=...) for the full record. Calls carry lead_id; a call ID is a different identifier. With include_timeline=True (lead_id only) the lead's timeline is added as data.timeline = {items, next_cursor}: creation, edits, call-now requests, conversations and blocklisting, newest first, up to 50 entries (GET /v1/leads/{lead_id}/timeline). If the timeline cannot be read, data.timeline is null and a warning says so; it is never an empty list. phone/email lookup uses GET /v1/leads/lookup through the grouped search adapter. ID lookup uses the call's lead_id, not its call ID. A 404 remains a failed read; check the ID and organization rather than assuming why the lead is missing or searching other organizations automatically. Args: lead_id: Lead ID (takes precedence). phone: Look up by phone number (E.164). email: Look up by email. org_id: Optional organization override for THIS call only. Superadmins may target any org; other users only orgs they can access (anything else raises a 403). The session's active org is not changed. include_timeline: Also fetch the lead's timeline (first 50 entries). Requires lead_id; refused with phone/email lookup.

get_lead

Org-wide or per-campaign performance: dashboard widgets, KPIs, disposition breakdown, duration distribution, call volume. KPIs and per-campaign dashboards default to total_conversations and contact_rate. Override with filters.metrics; query_metrics() lists the available metric IDs. Per-campaign dashboards group by campaign and require an explicit filters.time_granularity (hour/day/week/month/year). Date bounds are inclusive start, exclusive end. duration and call_count send start_date / end_date as start_time / end_time and campaign_id as campaign_ids. disposition (GET /v1/reports/dispositions) has no campaign filter: with a campaign_id it returns UnsupportedFilter; use list_dispositions(campaign_id=...). The org dashboard (GET /v1/reports/dashboard) takes only user_id, so date filters are not applied to it (a warning says so). Args: campaign_id: Scope to one campaign (per-campaign performance). start_date: ISO datetime lower bound. Bounds are sent as given (this server does not convert time zones), while the web app counts days in the organization's reporting time zone, so daily totals can differ. To match a local day, pass an offset, e.g. 2026-10-01T00:00:00-06:00. end_date: ISO datetime upper bound (exclusive); same offset rule. view: dashboard (default) | kpis | disposition | duration | call_count. org_id: Optional organization override for THIS call only. Superadmins may target any org; other users only orgs they can access (anything else raises a 403). The session's active org is not changed. filters: Extra query params. The v1 report views send only the params their endpoint documents and warn about the rest.

get_dashboard

Get one role's full instructions (can be 10K+ words). Args: detail: summary (identity and status), core (working configuration), full (default; complete existing response including history). role_id: Role ID (from list_roles or get_agent). org_id: Optional organization override for THIS call only. Superadmins may target any org; other users only orgs they can access (anything else raises a 403). The session's active org is not changed.

get_role

SMS registration (A2P/10DLC) fields for the organization, as the API stores them. Read-only. Returns the Twilio (a2p_status, a2p_registration, eqho_brand_share_pending) and Telnyx (sms_compliance_status, ten_dlc_brand_id, ten_dlc_campaign_id, messaging_profile_id) fields of the organization record, and the responses of GET /v1/organizations/a2p/has-verified-a2p-campaign, /v1/organizations/a2p/brand-info and /v1/organizations/telnyx/10dlc/status verbatim. A read that fails is reported as "unavailable" with a warning. ``enum_reference`` lists the values the API defines for twilio.a2p_status. Args: org_id: Optional organization override for THIS call only. Superadmins may target any org; other users only orgs they can access (anything else raises a 403). The session's active org is not changed.

a2p_status

Get your current session context and connection state. Use this to understand: - Which organization you're currently working with - Your permissions and capabilities - Connection health and status - Session expiration time - Available organizations you can switch to This is helpful when you're unsure about your current context or need to verify which organization's data you're accessing. Args: include_health: Include connection health metrics (default: True) Returns: Current session state, active organization, and connection info Examples: # Check current context get_context() # Quick check without health metrics get_context(include_health=False)

get_context

List the AI agents in the active org (name, voice, status), paginated. Args: limit: Max agents to return (upstream default ~16, up to 500). skip: Agents to skip (pagination). name: Case-insensitive substring filter on the agent name. org_id: Optional organization override for THIS call only. Superadmins may target any org; other users only orgs they can access (anything else raises a 403). The session's active org is not changed.

list_agents

List calls with server-side filters (summarized rows). Use hours=24 for "what happened today"; use q for free-text search across transcripts. Args: detail: Omit for the existing summarized response. summary returns compact rows, core adds working fields, full keeps complete rows. Explicit tiers preserve upstream pagination metadata. campaign_id: Filter to one campaign. agent_id: Filter to one agent. lead_id: All calls for one lead. call_type: inbound or outbound. start_date: ISO datetime lower bound on call start. end_date: ISO datetime upper bound on call start. hours: Convenience: calls that started in the last N hours (sets start_date when start_date is omitted). q: Free-text search across transcript, greeting and summary. limit: Rows per page (max 500). page: Page number, 1-based (sent as API offset). disposition_level_3: Exact stored disposition names or, with campaign_id, configured IDs or unique normalized names. Resolved values are the configured names stored on calls. Resolution also requires campaigns read scope. Raw filters preserve historical stored strings without reconciliation. sort_by: API sort field (default start). failed_only: Busy, failed, no-answer, or canceled calls. Conflicting status filters are rejected; compatible narrower filters are kept. order_by: asc or desc on start time (default desc). org_id: Optional organization override for THIS call only. Superadmins may target any org; other users only orgs they can access (anything else raises a 403). The session's active org is not changed. filters: Escape hatch: extra query params forwarded as-is. custom_fields: Optional {key: value} filter on conversation custom fields, the same keys webhook events publish as conversation_custom_fields. Exact, case-sensitive string match (strings only; a number stored as 12345 is not matched by '12345'); all pairs must match; dotted keys reach nested fields ({"order.id": "A17"}); at most 10 pairs, keys up to 128 characters, values up to 256.

list_calls

Tasks configured on a campaign (task_type: webhook, http_request, email, text/SMS, call_forward, try_again, llm_call, update_lead), with their ids, grouped by their stored ``type`` (pre_call, post_call, and unset for tasks without one; unset runs as post_call). try_again is post_call only: its retry counter has no meaning before the first attempt. Read-only; call this before update_campaign_tasks. Credential fields inside task settings (auth token/password, secrets, Authorization-style headers) are returned as "<redacted>". Sending the echo back to update_campaign_tasks keeps the stored credentials: "<redacted>" values are restored from the current task with the same id. Args: campaign_id: Campaign whose tasks to list. org_id: Optional organization override for THIS call only. Superadmins may target any org; other users only orgs they can access (anything else raises a 403). The session's active org is not changed.

list_campaign_tasks

List the campaigns in the active org (summarized rows). Args: status: Filter by status: ACTIVE, PAUSED, CREATED, INACTIVE, PROCESSED. Use status="ACTIVE" for "what is running right now". limit: Rows to return. skip: Rows to skip (pagination). org_id: Optional organization override for THIS call only. Superadmins may target any org; other users only orgs they can access (anything else raises a 403). The session's active org is not changed. filters: Escape hatch: extra query params forwarded as-is.

list_campaigns

Disposition values as the API reports them. Read-only. - stored: rows of GET /v2/reports/dispositions ([{disposition_level_3, count}]) in API order, verbatim, including rows whose label is null or empty. Without campaign_id, a failed v2 read falls back to GET /v1/reports/dispositions (stored_source says which endpoint answered; the v1 report has no campaign filter, so with campaign_id there is no fallback and stored is "unavailable"). - configured: the dispositions configured on the campaign (name, parent_disposition, contacted, conversion, enabled); only with campaign_id. - calls_without_disposition: meta.total of GET /v1/calls?dispositioned=false for the same window. Call this before filtering by disposition: the filter is an exact string match and a guessed value returns zero rows, not an error. Args: campaign_id: Restrict to one campaign (sent as campaign_ids). start_date: ISO datetime lower bound, sent as created_gte (default: 90 days ago). end_date: ISO datetime upper bound, sent as created_lt. org_id: Optional organization override for THIS call only. Superadmins may target any org; other users only orgs they can access (anything else raises a 403). The session's active org is not changed.

list_dispositions

List editable IVR drafts in the active organization. org_id overrides this call only and requires access; session organization is unchanged.

list_ivr_flows

Read claimed phone numbers for IVR routing, with offset pagination. org_id overrides this call only and requires access; session organization is unchanged.

list_ivr_numbers

Read registered softphone seat inventory for IVR routing. org_id overrides this call only and requires access; session organization is unchanged.

list_ivr_seats

List available IVR draft templates. org_id overrides this call only and requires access; session organization is unchanged.

list_ivr_templates

Read knowledge-base revision history. offset is 1-based and limit is 1–100. org_id overrides this call only and requires access; session organization is unchanged.

list_knowledge_base_revisions

List knowledge bases in the active organization. org_id overrides this call only and requires access; session organization is unchanged.

list_knowledge_bases

Lead lists in the org, or one list in full when list_id is given. Read-only. Args: list_id: Return this one list. campaign_id: Lists attached to this campaign. q: Name search. unassigned: Only lists not attached to any campaign. limit: Rows per page. page: Page number, 1-based (sent as the API's offset). org_id: Optional organization override for THIS call only. Superadmins may target any org; other users only orgs they can access (anything else raises a 403). The session's active org is not changed.

list_lead_lists

List or search leads (GET /v1/leads). With list_id: the leads on that lead list. Without: search the org by q / campaign_id / time. For a phone or email lookup use get_lead. Names are genuinely null for leads created by inbound calls; use get_lead for the full record (custom_fields, CRM ids). Args: detail: Omit for the existing summarized response. summary returns compact rows, core adds working fields, full keeps complete rows. Explicit tiers preserve upstream pagination metadata. list_id: Lead list to page through. campaign_id: Leads on one campaign. q: Free-text search. start_time: ISO-8601 lower bound on creation time. end_time: ISO-8601 upper bound on creation time. limit: Rows per page. page: Page number, 1-based (sent as the API's ``offset``). org_id: Optional organization override for THIS call only. Superadmins may target any org; other users only orgs they can access (anything else raises a 403). The session's active org is not changed. filters: Escape hatch: extra query params forwarded as-is.

list_leads