PhantomBuster

PhantomBuster

This connector drives PhantomBuster through its API v2 (https://api.phantombuster.com/api/v2): the agents (Phantoms set up in your workspace) that extract and enrich leads from LinkedIn, Sales Navigator, Google Maps, Instagram and other sites, and automate outreach. It lists and reads agents, launches and stops them, follows their runs (containers) and reads the results, edits an agent's settings and schedule, works with the leads lists and stored leads of the workspace storage, and reports plan usage.

Setup (2 minutes)

1. In PhantomBuster open Workspace settings (https://phantombuster.com/workspace-settings, also in the navbar menu) and add an API key. It is shown only once: copy it.
2. Paste it into PHANTOMBUSTER_API_KEY. It is sent in the X-Phantombuster-Key header and gives full access to that one workspace.
3. Run phantombuster_get_org to check the key.

PhantomBuster also runs its own MCP server (https://mcp.phantombuster.com, OAuth sign-in) with a similar set of tools; this connector uses the API key instead.

Workflow

• Workspace and usage: phantombuster_get_org (id, plan, time zone, s3Folder), phantombuster_get_org_resources (execution time, email and AI credits used and left), phantombuster_list_running_containers, the CSV exports phantombuster_export_agent_usage and phantombuster_export_container_usage, dashboard folders with phantombuster_list_agent_groups.
• Find an agent: phantombuster_list_agents (filter with agent_ids, input_types, output_types), then phantombuster_get_agent for its settings; with_manifest: "true" shows the input fields its script accepts. Deleted ones: phantombuster_list_deleted_agents. Scripts: phantombuster_list_scripts, phantombuster_get_script. Always work with agent ids, never names: two agents can share a name.
• Run an agent: phantombuster_launch_agent (returns the containerId), phantombuster_launch_agent_soon for a delayed run, phantombuster_abort_agent to stop it. Then poll phantombuster_get_agent_output every minute or so (status, progress, new log lines with from_output_pos and prev_container_id) until status is finished; do not poll in a tight loop.
• Results: phantombuster_list_containers (runs of an agent), phantombuster_get_container, phantombuster_get_container_output (log) and phantombuster_get_container_result (the data of a run). Result files are also at https://phantombuster.s3.amazonaws.com/<orgS3Folder>/<s3Folder>/result.csv (and result.json), with both folders from phantombuster_get_agent.
• Configure: phantombuster_save_agent changes the name, saved argument (input), schedule (launch_type, repeated_launch_preset), time limit and notifications, or creates an agent from a script. Saving never launches a run.
• Leads storage (beta API): phantombuster_list_lead_lists, phantombuster_get_lead_list, phantombuster_list_leads_in_list; create or change a list with phantombuster_save_lead_list, add or update one lead with phantombuster_save_lead. Installed switched off: phantombuster_delete_leads, phantombuster_delete_lead_list, phantombuster_delete_agent and phantombuster_unschedule_all_agents (stops every schedule).

Launching is real: confirm first. A launch acts on the connected social accounts (visits profiles, sends connection requests or messages) and uses execution time of the plan. Show the user the agent, its input and the volume, and launch only after a clear yes. Too many actions per day can get a LinkedIn account restricted; keep the daily limits the Phantom suggests.

Launch argument: argument replaces the saved input settings for that run (and is saved too with save_argument); bonus_argument merges a few fields over the saved ones for one run. The field names are those of the script manifest, e.g. spreadsheetUrl, numberOfProfiles, sessionCookie (never ask the user to paste a cookie into the chat; connected accounts are set up in PhantomBuster).

Leads list filters: a list is a saved filter. Top level {"and": [ ... ]} of clauses {"filter": {"<field>": {"entity": "lead", "operator": "contains", "valueToCompare": "CTO"}}}, nested and / or, or {"__global_search__": "text"}. Lead fields are snake_case (linkedin_job_title, location, professional_emails), company fields camelCase with "entity": "company" (industriesV2, employeeCount). Operators: equals, notEquals, contains, notContains, in, notIn, greaterThan, lowerThan, greaterOrEquals, lowerOrEquals, beforeDateTime, afterDateTime (ISO 8601), beforeNowPlusOffset / afterNowPlusOffset (milliseconds), isSet, isNotSet, include, some, every. Check a new list with phantombuster_list_leads_in_list and include_total_count.

Reading the answers: ids are strings of digits. Dates are Unix timestamps in milliseconds. Run status: starting, running, finished, launch error, never launched; exitCode 0 and endType finished mean success. Usage values in phantombuster_get_org_resources are seconds of execution time and counts of credits.

Paging: phantombuster_list_containers returns limit runs; for older ones pass the endedAt of the last run as before_ended_at. phantombuster_list_leads_in_list pages with page_size and offset. The other lists return everything at once.

Errors: 401 "API key not found" for a wrong or revoked key, 400 for a missing or wrong parameter (or a launch argument that is too big), 404 for an unknown id. A launch fails when the plan's execution time is used up or too many runs are in parallel (maxParallelism).