LeadJourney

Delete a send-back rule. LeadJourney stops sending that event to the app; records already written there stay. The first call without `confirm: true` deletes nothing and explains what would stop — put that to the user and only call again with `confirm: true` once they agree.

delete_app_outbound

Delete one app trigger. Events of that type stop becoming conversions; conversions already recorded stay. The first call without `confirm: true` deletes nothing and explains what would stop — put that to the user and only call again with `confirm: true` once they agree.

delete_app_trigger

Delete a custom traffic channel. Built-in channels cannot be deleted — switch them off with `update_custom_traffic_channel` instead. Historic traffic already attributed to the channel stays in reports; the channel simply stops being offered. Links still carrying its key will no longer be classified, so retire the links first. Pass `confirm: true` to carry it out.

delete_custom_traffic_channel

Delete a dashboard and every report card on it. There is no undo, and the cards are not recoverable — confirm with the user first, and pass `confirm: true` to carry it out.

delete_dashboard

Delete a tag, its rules, and the tag from every lead carrying it. Reports and saved filters that segment on the tag lose that segment. Pass `confirm: true` to carry it out.

delete_tag

Delete a tracking domain. Every tracking link and short link already published on that host stops resolving the moment it goes — including ones printed on material nobody can recall. Pass `confirm: true` to carry it out.

delete_tracking_domain

Before `request_ad_creation`: the fields a new campaign, ad set and ad take on this platform (required ones, allowed values, text limits, defaults), and the account's own building blocks to refer to by id — Meta Pages, pixels and images; Google conversion actions. Read-only.

describe_ad_creation

Describe how an AI agent workflow is put together: the report sources it can pull, the date ranges, the report formats, the destinations it can send to, and the schedule options. Call this before `create_agent` — the keys are not guessable, and an unrecognised one is dropped without an error. An agent is three steps in a fixed order: GET DATA (which report, over which dates) → GENERATE REPORT (a prompt, and the format) → DELIVER (email, Slack or webhook). The schedule decides when it runs.

describe_agent_config

Everything needed to send LeadJourney data back to an app — e.g. push every new lead into HubSpot as a contact with its UTM source, or every won conversion into a Salesforce opportunity. This is the opposite direction of triggers. Returns the `catalog` (events LeadJourney can send with their `source_fields`, the app's objects, and the write modes each object supports), this workspace's `conversion_types`, the `existing_rules`, and app `notes` (for HubSpot: writing may need a reconnect). Pass `object` to also get `writable_fields` — the app fields that object can be written to. Call it before create_app_outbound, and again with `object` once you know what to write.

describe_app_outbound

Everything needed to build a trigger for one app without guessing. A trigger turns an event in the app (a booked meeting, a won deal, a paid invoice) into a LeadJourney conversion. Returns: - `trigger_types` — the only accepted values for `trigger_type`, with what each one fires on. - `conversion_types` — this workspace's conversion types; `conversion_type` must be one of them, exactly. - `recommended_triggers` — which trigger type fits a lead, a booked call, a sale. - `default_fields`, `target_fields`, `preferred_paths`, `known_paths` — which LeadJourney fields to map and the app paths that usually hold them. - `click_id` — how the click id reaches this app and which path carries it. Mapping it is what ties the conversion to the ad click. - `extra_fields` — app-specific options passed in `options`. Call this before create_app_trigger, every time.

describe_app_triggers

What can be read live from one ad platform, and which of its accounts this workspace has connected. Call this before `get_live_ad_report` or `get_live_ad_entities`: metric, breakdown and field names are the platform's own and differ per platform, and an invented one fails. Live reads ask the platform at this moment, with every metric and setting it exposes — unlike `get_campaign_report`, which is LeadJourney's synced data joined with CRM revenue. Use the live tools for platform-side detail (search terms, impression share, frequency, quality ranking, targeting, bidding settings) and the LeadJourney reports for real leads and revenue; the best answers combine both.

describe_live_ad_data

List every metric, dimension, filter field and attribution setting the report builder accepts, for the current workspace. Call this BEFORE `run_report` or `add_dashboard_report` — the ids are not guessable, and an unknown one produces an empty report rather than an error.

describe_report_builder

List every field, operator and value a tag rule can be built from in this workspace: rule fields, comparison operators, tag types, the integrations leads arrive from, the workspace's traffic channels and its conversion types. Call this before `create_tag`. A rule referencing a field or value that does not exist saves without complaint and then never matches a single lead.

describe_tag_rules

Look at a website and work out which CMS it runs on and which form builder its forms use — the two things that decide how the tracking script and form capture get installed. Use it before walking someone through setup, so the instructions match what they actually have rather than a generic version.

detect_website_stack

Disconnect an app from this workspace: its webhook is removed, its triggers go with it, and no more events arrive until the user reconnects in Settings → Apps (connecting again cannot be done from here). The first call without `confirm: true` disconnects nothing and explains what would be lost — put that to the user and only call again with `confirm: true` once they agree.

disconnect_app

Retrieve ad creative performance data for the user's current workspace. Use this tool to analyze which ads, images, videos, and creative assets are performing best. Returns metrics like impressions, clicks, conversions, spend, revenue, ROAS, and cost per conversion for each creative. Supports filtering by date range, traffic channel (Facebook, Google, etc.), creative type, and attribution method (first_click or last_click). Call `switch_workspace` first if the user wants results for a different workspace.

get_ad_creative_report

The ad platform's own optimisation advice for a connected account: Google's recommendations with their estimated impact plus the account and campaign optimization scores; Meta's opportunity recommendations. Read-only — nothing is applied. Treat these as the platform's suggestions, not LeadJourney's: a platform optimises for the conversions it can see, so check a "raise budget" or "broaden match" suggestion against real leads and revenue in `get_campaign_report` before passing it on. `describe_live_ad_data` says whether a platform has recommendations at all.

get_ad_platform_recommendations

One ad platform's whole setup as a checklist: connect → pick ad account → pixel (Meta, X) → choose which events are sent → map each event to the platform's event name. Each step says whether it is done, whether the user or Atlas does it, and a `where` link to that exact screen. Also returns the connected ad accounts (ids, names, pixel — never a token), every workspace event with its `send_to_ads` switch and its current name on this platform, and the platform's `standard_events`. Use it before update_ad_event_mapping, and to walk a user through what is still missing.

get_ad_platform_setup

Read one AI agent workflow back in full: every step and its config, the destinations, the schedule, and the last runs with their status and any error. Start here when an agent is not producing what was expected — the run log says which step stopped.

get_agent

Read a CRM's real fields and values for trigger conditions ("only when <field> is <value>"): without `field`, the fields of `object` that can be a condition; with `field`, the values it can take — pipeline stage ids, lead statuses, picklist entries. Use the value (not the label) in create_app_trigger options, e.g. {"attribute_name": "stage_id", "attribute_value": "12"} or {"property_name": "dealstage", "property_value": "contractsent"}. Never guess a stage or status id.

get_app_field_options

The log of what LeadJourney sent to an app, newest first: each push's event, object, status (success / failed / skipped), HTTP status and error. Use it when the user asks whether leads are reaching their CRM, or why one did not.

get_app_outbound_deliveries

Show one app's connection in this workspace: whether it is connected, which account it is connected to, a live health check against the provider (token revoked, webhook missing, unreachable), and every trigger with its id, trigger type, conversion type, field mappings and active flag. Use the trigger ids from here with update_app_trigger and delete_app_trigger. When `health.healthy` is false, tell the user the `reason` and that reconnecting in Settings → Apps fixes a revoked connection.

get_app_status

Retrieve campaign performance data for the user's current workspace. Use this tool to analyze marketing campaign metrics including clicks, conversions, revenue, spend, ROAS, cost per lead, and cost per conversion. Supports filtering by date range, traffic channel (Facebook, Google, TikTok, etc.), conversion type, and attribution method (first_click or last_click). Call `switch_workspace` first if the user wants results for a different workspace.

get_campaign_report

Get everything a CRM needs to send conversions back to this workspace: the postback URL, how to authenticate, and which parameters the payload takes. The response includes the workspace's postback SECRET. Treat it as a credential — it belongs in the CRM's webhook settings, not in a chat log, a ticket or an email.

get_crm_webhook_info

Get the authenticated user's identity (email) and currently selected workspace. Call this first whenever the user asks about "my" data so you know the active context.

get_current_workspace

Fetch a single lead's full journey (events, tags, emails, phones) by `lead_id`. Same payload as the lead-overview modal.

get_lead_journey

Retrieve leads and their conversion data from the user's current workspace. Use this tool to analyze lead performance, track conversions, and understand customer journey. Supports filtering by date range, traffic channel, conversion type, campaign, ad set, asset, keyword, ad, and attribution method (first_click or last_click). Call `switch_workspace` first if the user wants results for a different workspace.

get_leads

The account, campaigns, ad sets, ads or keywords exactly as the ad platform holds them right now — status, budgets, bidding strategy, targeting, networks, schedule, tracking templates, approval status. Read-only. Use this to audit settings ("which campaigns still run on the Display Network", "which ad sets have no frequency cap", "is auto-tagging on") before recommending a change. `fields` takes the platform's own field names; leave it empty for the defaults `describe_live_ad_data` lists. Without `ids` every live (not removed) entity at the level is returned, up to `limit`.

get_live_ad_entities

Performance straight from the ad platform's reporting API, right now: any metric and breakdown the platform offers, at account, campaign, ad set, ad or keyword level. Read-only — nothing on the ad account changes. Call `describe_live_ad_data` first for the platform's metric and breakdown names. Pass `date` in breakdowns for one row per day; leave it out for totals over the range. Conversions here are what the platform itself counts; for leads and revenue from the CRM use `get_campaign_report` or `run_report`, and compare the two.

get_live_ad_report

Returns the authenticated user's profile (id, name, email).

get_my_profile

Show where the workspace has got to in onboarding — the checklist, the current step, the detected website stack — together with the live state of the pieces that actually have to work: tracking domains and their verification, whether any conversion events are defined, and whether leads have started arriving. Start here when someone says setup is not working. The checklist can read as finished while a domain is still unverified, and that is the usual cause.

get_onboarding_status

Retrieve traffic channel performance data for the user's current workspace. Use this tool to compare performance across different advertising platforms (Facebook Ads, Google Ads, TikTok Ads, etc.). Returns aggregated metrics like total clicks, conversions, revenue, spend, ROAS, and cost per acquisition for each traffic source. Supports filtering by date range and attribution method (first_click or last_click). Ideal for understanding which channels drive the best results and where to allocate budget. Call `switch_workspace` first if the user wants results for a different workspace.

get_traffic_channel_report

Which fields a Google Ads resource has, and which metrics and segments can be selected with it — so a `google_ads_search` query is right the first time. Pass `resource` (e.g. "campaign", "search_term_view", "keyword_view") to get its attribute fields plus the metrics and segments it accepts; pass `field` (e.g. "metrics.search_impression_share") to get one field's type, enum values and where it can be used.

google_ads_field_metadata

The ad changes asked for in this workspace — pending ones waiting for confirmation, and the log of applied, failed, cancelled and expired ones with before and after values — plus what the current user may ask for (per platform and action) and the workspace's money limits. Use it to answer "did my change go through", to avoid asking for a change that is already pending, and to check permissions before proposing one.

list_ad_changes

Every ad platform LeadJourney sends conversions to — Meta, Google, Microsoft (Bing), LinkedIn, Reddit, X, ChatGPT, Taboola, Outbrain, TikTok — with whether it is connected, whether an ad account is picked, whether the connection expired, and the next step. Start here when the user talks about ads, pixels, the Conversions API or offline conversions.

list_ad_platforms

List the AI agent workflows in the current workspace: what each one reports on, where it sends the result, its schedule, and when it last ran. Use it to find the `agent_id` for `get_agent`, `update_agent`, `run_agent` or `delete_agent`.

list_agents

List every app LeadJourney integrates with (Cal.com, Calendly, HubSpot, Salesforce, Typeform, Stripe, …) and, for each one, whether it is connected in this workspace and how many conversion triggers it has. The `app` value of each row is the key every other app tool takes. Start here when the user mentions an app, or asks what is connected.

list_apps

List the dashboards in the current workspace and the report cards on each. Use it to find the `dashboard_id` for `add_dashboard_report` or `delete_dashboard`, and the `report_id` for `remove_dashboard_report`.

list_dashboards

List the workspace's tags and the rules that apply each one automatically. A tag with no rules is applied by hand only.

list_tags

List the workspace's tracking domains and where each one has got to in verification. A domain only works once its status is `success`; anything else means the CNAME has not been seen yet, and links on it will not resolve.

list_tracking_domains