> ## Documentation Index
> Fetch the complete documentation index at: https://docs.manypi.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Tool reference
> Every tool the ManyPI MCP server exposes.
Sixteen tools, grouped the way you use them. Each lists its arguments and its
safety annotation — the hint MCP clients use to decide what to run
automatically and what to confirm with you first.
## Agent
### `run_agent`
Write · reaches the web
Run the ManyPI agent on a natural-language goal: scrape sites, research,
analyse data, turn a site into an API. Asynchronous — returns a `run_id` and
`conversation_id`.
| Argument | Type | |
| -------- | ------ | ------------------------------------------------------------ |
| `goal` | string | **Required.** What you want ManyPI to do, in plain language. |
For finding leads, prefer `generate_leads`.
### `get_agent_run`
Read-only
Check the status and result of agent and lead-gen runs.
| Argument | Type | |
| ----------------- | ---- | -------------------------------------------------------------------- |
| `conversation_id` | uuid | From `run_agent` / `generate_leads`. Omit for your most recent runs. |
| `limit` | 1–50 | How many runs to return. |
Statuses: `queued` / `planning` / `running` (working), `paused` (waiting on you
— the question is in `result_summary`), `completed` / `failed` / `cancelled`
(done — see `result_summary`).
### `reply_to_agent`
Write · reaches the web
Answer an agent that is waiting for input, or send a follow-up to a finished
run. Resumes the **same** run if it was paused; otherwise starts a follow-up run
in the same conversation.
| Argument | Type | |
| --------- | ------ | --------------------------------------------------------------- |
| `run_id` | uuid | **Required.** The run to answer. |
| `message` | string | **Required.** Your answer or follow-up, up to 8,000 characters. |
## Lead generation
### `generate_leads`
Write · reaches the web
Find and enrich B2B leads. Describe your ideal customer; poll `get_agent_run`
for progress and read results with `search_leads`.
| Argument | Type | |
| ------------- | ------ | ----------------------------------------------------------------------------------- |
| `description` | string | **Required.** e.g. *"e-commerce agencies in Germany, 5–50 employees, with emails"*. |
### `search_leads`
Read-only
Search the workspace's saved leads.
| Argument | Type | |
| -------- | ------ | -------------------------------------------------- |
| `query` | string | Matches company, name, email or domain. |
| `status` | enum | `new` · `qualified` · `contacted` · `disqualified` |
| `limit` | 1–500 | Defaults to 100. |
## Outreach content
### `list_inboxes`
Read-only
List connected sending inboxes with per-inbox health: today's sends versus the
warmup-aware cap, 14-day delivered/failed/opened, and a 0–100 health score. Use
an inbox `id` as `smtp_config_id` elsewhere. Takes no arguments.
### `list_sequences`
Read-only
List saved email sequences with their inline steps. Takes no arguments.
### `create_sequence`
Write
Create the email steps a campaign sends.
| Argument | Type | |
| ------------- | ------------ | ---------------------------- |
| `name` | string | **Required.** Internal name. |
| `description` | string | Optional. |
| `steps` | array (1–10) | **Required.** Ordered steps. |
Each step: `subject`, `body_html`, `wait_days` (0–90), `thread_mode`
(`new` / `reply`). Step 0 goes out on enrollment and is always `new`; a `reply`
step threads under the first email.
Merge variables: `{{first_name}}`, `{{company}}`, `{{title}}`, `{{domain}}`,
`{{location}}`. Bodies are simple HTML — ``, `
`, ``, ``.
### `draft_email`
Read-only
AI-draft a cold email or a whole follow-up sequence. Returns
`{subject, html}` or `{steps: […]}`. **Persists nothing** — feed the result
into `create_sequence` or `send_email`.
| Argument | Type | |
| -------------- | --------------------- | --------------------------------------------------------- |
| `instructions` | string | **Required.** What you are selling / your ask. |
| `mode` | `email` \| `sequence` | Defaults to `email`. |
| `steps` | 2–5 | Number of steps when `mode` is `sequence`. Defaults to 3. |
| `lead_id` | uuid | An example lead to tailor the tone toward. |
## Campaigns
### `list_campaigns`
Read-only
List campaigns with per-campaign stats — enrolled, active, completed, sent,
opened, replied. Takes no arguments.
### `get_campaign`
Read-only
One campaign's full analytics: headline stats, a per-step sent/opened
breakdown, and every enrolled lead with its progress.
| Argument | Type | |
| ------------- | ---- | ------------- |
| `campaign_id` | uuid | **Required.** |
### `create_campaign`
Write
Bind a sequence to a sending inbox.
| Argument | Type | |
| ---------------- | ------ | -------------------------------------------------------- |
| `name` | string | **Required.** |
| `sequence_id` | uuid | **Required.** From `list_sequences` / `create_sequence`. |
| `smtp_config_id` | uuid | **Required.** From `list_inboxes`. |
| `reply_to` | email | Override the reply-to address. |
| `daily_limit` | 1–2000 | Never exceeds the inbox's own cap. |
### `update_campaign`
Destructive · idempotent
Pause, resume, archive, or change a campaign's name, reply-to or daily limit.
Pausing stops further sends without losing progress.
| Argument | Type | |
| --------------------------------- | ---------------------------------- | ------------- |
| `campaign_id` | uuid | **Required.** |
| `status` | `active` \| `paused` \| `archived` | |
| `name`, `reply_to`, `daily_limit` | | |
Annotated destructive because it overwrites existing fields and archiving is
this API's closest thing to a delete — but applying the same patch twice lands
on the same state, so it is also idempotent.
## Sending
Both tools below queue **real email to real people** and require a paid ManyPI
plan. Nothing sent can be recalled.
### `enroll_leads`
Destructive · reaches the web
Enroll leads into a campaign, queuing the sequence for real sending and
respecting the inbox's warmup and daily caps. Leads with no email, or already
enrolled, are skipped.
| Argument | Type | |
| ------------- | ---------------- | ---------------------------------- |
| `campaign_id` | uuid | **Required.** |
| `lead_ids` | uuid\[] (1–1000) | **Required.** From `search_leads`. |
### `send_email`
Destructive · reaches the web
Send a single one-off email to a lead now, outside any sequence. Renders merge
variables against the lead, respects the inbox's cap, and marks the lead
contacted.
| Argument | Type | |
| ---------------- | ------ | ------------------------------------------------------ |
| `smtp_config_id` | uuid | **Required.** From `list_inboxes`. |
| `subject` | string | **Required.** Merge variables allowed. |
| `html` | string | **Required.** Simple HTML. |
| `lead_id` | uuid | Used for merge variables and as the default recipient. |
| `to` | email | Explicit recipient; defaults to the lead's address. |
### `outreach_stats`
Read-only
Workspace-wide activity: 14 days of sends (sent/failed/opened per day), replies
per day, active enrollments, and today's sent versus the combined cap. Takes no
arguments.
## Scraping
### `list_scrapers`
Read-only
List the workspace's scrapers — id, name, target URL, type. Takes no arguments.
### `run_scraper`
Write · reaches the web
Run a saved scraper. Asynchronous — returns a `runId`; read the extracted data
with `get_scraper_runs`.
| Argument | Type | |
| ------------ | ------ | ------------------------------------------- |
| `scraper_id` | string | **Required.** From `list_scrapers`. |
| `url` | url | Override the scraper's target for this run. |
### `get_scraper_runs`
Read-only
Recent scraper runs with status and extracted data.
| Argument | Type | |
| -------- | ----- | - |
| `limit` | 1–100 | |