LeadJourney

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

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

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

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

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

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

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

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 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

List the traffic channels configured in the current workspace: the built-in ones (Meta Ads, Google Ads, organic search, …) and any custom channels this workspace added. `is_custom` decides what can be done to a channel — only custom ones can be edited or deleted. `traffic_channel_key` is the id every report filter and UTM link uses.

list_traffic_channels

List every channel the UTM Builder can produce a tracking link for, and which of the five link KINDS each one is. Call this before `build_utm_link` — the kind decides which arguments that tool needs. - `paid`: a static template the ad platform fills in at click time. Nothing to configure; it is pasted into the platform's tracking field. - `organic`: a link you tag by hand before posting. - `email`: a newsletter or automation link, optionally carrying the recipient's address so the lead is identified. - `fixed`: a built-in channel with a fixed `lj_channel` value, tagged by hand (Google My Business). - `custom`: one of this workspace's own custom traffic channels.

list_utm_channels

List every workspace the authenticated user can access, with the current workspace flagged. Use this before any reporting work so the user can confirm which workspace is active or pick a different one with `switch_workspace`.

list_workspaces

Remove one report card from a dashboard. The dashboard itself and every other card stay. Get both ids from `list_dashboards`.

remove_dashboard_report

Run an AI agent workflow immediately, ignoring its schedule. It runs in the background and DELIVERS for real — emails go out, Slack messages post, webhooks fire — so use it to prove a new agent works before activating it, not to preview one. Read the outcome back with `get_agent`, which shows the run's status and any error.

run_agent

Run a custom report against the current workspace and return the rows. Any combination of metrics, one dimension (`group_by`), an optional second dimension, filters and an attribution setting. Call `describe_report_builder` first for the valid metric / dimension / filter ids — an id this tool does not know produces empty rows, not an error. To keep a report, pass the same arguments to `add_dashboard_report`.

run_report

Switch the authenticated user's current workspace. Pass the `workspace_id` from `list_workspaces`. After switching, every reporting tool (campaigns, traffic channels, ad creatives, leads) will operate on the new workspace.

switch_workspace

Change an AI agent workflow: its name, its data step, its prompt, its destinations, its schedule, or its status (activate, pause, archive). `destinations` REPLACES every existing destination — send back the ones you are keeping, or they are gone. Call `get_agent` first and work from what is there. Anything you leave out entirely is left alone.

update_agent

Rename a custom traffic channel, change its description or type, or switch it on and off. Built-in channels can only be switched on and off — their label and type are fixed. Renaming re-derives the channel key. Links already published with the old key keep reporting under it, so rename only when nothing is live yet, or accept the split.

update_custom_traffic_channel

Change a tag's name, colour, type or rules. Passing `rules` REPLACES the whole set — every rule you mean to keep has to be sent back, or it is gone. Call `list_tags` first and send the existing rules along with the new one. Leave `rules` out entirely to change only the name or colour.

update_tag

Change a tracking domain's host, or retry verification on one that failed. Changing the host starts verification again from scratch, and every short link already published on the old host stops resolving — so treat it as creating a new domain, not editing a label. `retry: true` re-runs verification on a failed domain, which is what to do once the CNAME has actually been added.

update_tracking_domain

Fetch a page and check whether this workspace's LeadJourney tracking script is actually on it. The single most useful check when leads are not appearing. It reads the page's server-rendered HTML, so a script injected later by a tag manager or a client-side app can be missed even though it works in a browser. A negative result is a reason to look, not proof it is absent.

verify_tracking_script