Skip to main content
Glama

Server Details

Run B2B outreach from your AI agent: 250+ tools for campaigns, leads, LinkedIn and email workflows.

Ownership verified
Status
Healthy
Uptime
100.0% over the last 24 hours
Last Tested
Transport
Streamable HTTP · MCP 2025-11-25
URL
ChatGPT
–
Claude
–

TDQS

A3.5/5.0

Scored across 105 tools

Disambiguation3/5

Most tools target distinct resources and actions, and descriptions often explicitly redirect to the right alternative. However, several edit_* tools also perform deletion (edit_competitor_watch, edit_signal_agent, edit_campaign), duplicating separate delete_* tools, and the sheer number of similar-sounding operations creates real selection burden.

Naming Consistency4/5

Nearly all tools follow a predictable campaignstack_verb_noun snake_case pattern (get_, edit_, delete_, send_, refresh_, etc.). Minor deviations like campaignstack_whoami and some very long compound names keep it from a perfect score.

Tool Count1/5

105 tools is far beyond the 3–15 well-scoped range and exceeds even the 50+ 'extreme mismatch' threshold. Even for a broad outreach platform, this volume makes the surface unwieldy and hard to navigate.

Completeness4/5

The surface covers nearly every resource lifecycle—campaigns, leads, lists, workflows, content, accounts, reviews, watchers, integrations, billing, and feedback. Minor gaps exist, such as no direct lead deletion or standalone email-send tool, but agents can work around them via workflows and edit operations.

Available Tools

105 tools
campaignstack_add_leads_to_external_destinationAdd leads to external destinationA
DestructiveIdempotent
Inspect

Add up to 25 leads to one provider list or sequence. This changes a third-party workspace and may make contacts available for outreach there. This reaches a real person. Before calling it, tell the user exactly who it goes to and read the wording back to them, and wait for an explicit yes.

Inputs: leads (required), provider (required), confirmed (required), destination (required), +1 more. Returns: ok, failed, result, results, +1 more

campaignstack_analyze_company_websiteAnalyze company websiteA
Read-onlyIdempotent
Inspect

Renamed V2 website-analysis read for a company's own public website. Cached analyses are reused; fresh AI extraction is credit-gated and charged to the bound workspace. Charges website_intel credits per call; campaignstack_estimate_cost gives the number before you commit.

Inputs: url (required), workspaceId. Returns: ok, items, failed, results, +1 more

campaignstack_answer_job_questionAnswer job questionA
Destructive
Inspect

Send the person's answers for an application stopped at needs_answer. Answers are saved to the watcher's application profile for future matching questions, then CampaignStack continues the application under review, credit, and account daily-limit gates. Use only answers the person gave you; read every answer back first. confirmed:true is required.

Inputs: answers (required), confirmed (required), applicationId (required), workspaceId. Returns: status, applicationId, creditAttempt

campaignstack_apply_to_jobApply to jobA
Destructive
Inspect

Start an application for exactly one matched posting. CampaignStack reads the posting, drafts a cover letter in the person's voice, then parks it for review or submits later under the watcher's automatic-mode and account-safety gates. Pass applicationId, or postingId only when exactly one live watcher matched that posting. confirmed:true is required after naming the posting and company and getting explicit approval.

Inputs: confirmed (required), postingId, workspaceId, applicationId. Returns: status, applicationId, creditAttempt

campaignstack_archive_job_watcherArchive job watcherA
Destructive
Inspect

Archive one job watcher permanently. It leaves live watcher lists and never searches or applies again, while previous applications remain readable. Use pause in campaignstack_edit_job_watcher for temporary stops. confirmed:true is required after naming the watcher/account back to the user.

Inputs: confirmed (required), watcherId (required), workspaceId. Returns: ok, item, results

campaignstack_cancel_workflow_node_leadsCancel workflow node leadsA
Destructive
Inspect

Confirmed drain operation for one workflow node. Waiting, event-waiting, and pending-review entries are cancelled; processing or already sent work cannot be recalled and is reported separately.

Inputs: nodeId (required), confirmed (required), workspaceId. Returns: ok, results, warnings, workflowId, +1 more

campaignstack_comment_on_postA
Destructive
Inspect

Ask CampaignStack to post a comment on a lead's most recent post from a connected LinkedIn account. Call this when the user has approved specific comment wording for a specific lead's post. platform is required (currently "linkedin"). Name the person with leadId, which is the supported form; profileUrl is accepted for someone already in the workspace. commentText must be the owner's own words or wording they approved: draft it with campaignstack_craft_message, or park it for approval with campaignstack_submit_comment_for_review instead of posting it here. The workspace's only connected LinkedIn account is used automatically, or pass accountId (campaignstack_get_accounts). A comment is public and stays under the owner's name. This reaches a real person. Before calling it, tell the user exactly who it goes to and read the wording back to them, and wait for an explicit yes. It goes out as soon as this call succeeds: there is no draft state and nothing to recall. One person per call. There is no bulk form, and calling this in a loop over a list is not the supported way to reach a list: add the leads with campaignstack_edit_lead_list and run a workflow over it with campaignstack_trigger_workflow, which paces the sends and applies the review step the workspace configured. Business hours do not apply. CampaignStack's daily limit for this action (and the weekly limit for invitations) still applies and counts sends already queued, so a run of calls stops at the limit; the owner sets those limits in the app. A paused account, or an action LinkedIn has challenged, refuses the send whatever the limit.

Inputs: platform (required), commentText (required), leadId, accountId, +3 more. Returns: jobId, status, success

campaignstack_configure_integrationConfigure integrationAInspect

Store or replace one provider integration using provider-nested credentials. This tool is non-OAuth because it accepts secrets; connected apps should direct the user to the workspace settings, Integrations section, in the CampaignStack app. Secrets are validated and never echoed back.

Inputs: provider (required), credentials (required), workspaceId. Returns: ok, failed, result, results, +1 more

campaignstack_connect_accountConnect accountAInspect

Return a browser handoff for connecting a new LinkedIn or Google account. Never asks for OAuth secrets, cookies, or passwords in chat.

Inputs: platform (required), workspaceId. Returns: ok, item, results, warning, +1 more

campaignstack_craft_messageCraft messageAInspect

Draft one outbound message without sending anything. Select channel linkedin/email and kind first_touch, connection_note, comment, comment_reply, followup, or meeting_followup. Returns sent:false; callers must read the draft back and use a separate send/action tool only after explicit approval. It never books a meeting for you: when a follow-up finds the lead accepted a time offered earlier, schedulingAcceptance names it (booked:false) for a person to book.

Inputs: kind (required), channel (required), leadId, postUrl, +14 more. Returns: ok, results, workspaceId

campaignstack_create_workspaceAInspect

Create a workspace and build it. Give a company website (acme.com), a LinkedIn profile URL, or both, and CampaignStack sets up what a new sign-up gets: the company context, the playbook, a first campaign with its ideal customer profile, a workflow, and a lead pool. Pass goal when the user told you what the workspace is for, in their words and in the first person: it becomes the campaign goal word for word, and without it one is written from the website and profile. Leave name out unless the user chose one; the company's name is used. A website alone is enough and needs no LinkedIn account: the workspace is whole and outreach waits until an account is linked with campaignstack_edit_workspace_accounts or connected under Accounts in the app. A profile URL is read through the user's own connected LinkedIn account, so it needs one. With neither URL you get a blank workspace and must name it. Returns the workspaceId at once and, when a build started, a generationId to follow with campaignstack_get_workspaces. Needs a user key or a signed-in connection: a workspace key is one workspace and cannot create another. One build runs at a time, and a free account owns one workspace.

Inputs: goal, name, profileUrl, websiteUrl. Returns: mode, next, workspaceId, generationId

campaignstack_decide_reviewDecide reviewA
Destructive
Inspect

Approve, reject, or edit-and-approve one pending review. To save a revised draft for a person to sign off instead of deciding it, use campaignstack_edit_draft. For kind workflow, an approved message goes out later, on the account's pacing (business hours, daily budget, spacing between sends), which can take hours; with priority:true it skips the queue and goes out within seconds when the account's send window is open, otherwise at the window's next opening. Use priority only when the user asked for this message to go out now; if campaignstack_get_reviews with this entryId shows a waitReason, tell them why it is waiting. For kind job_application, approval sends or schedules the application within limits; edit_and_approve accepts updatedContent { letter?, answers?: [{ question, answer }] }, and reject may include reason. confirmed:true is required.

Inputs: entryId (required), decision (required), confirmed (required), kind, +5 more. Returns: ok, results, warnings, workflowId, +1 more

campaignstack_delete_company_employee_watchA
Destructive
Inspect

Delete company employee watches by archiving them so they no longer refresh. Call this only after the user confirms the exact watches to remove. This permanently deletes a root record. Before calling it, name exactly what will be deleted, explain that it cannot be undone, and wait for an explicit yes. Batch edits are bounded. Return one result per requested item, and report partial failures instead of describing the whole batch as successful.

Inputs: watchIds (required), confirmed (required), workspaceId. Returns: results, warnings

campaignstack_delete_competitor_watchA
Destructive
Inspect

Delete competitor watches and disable their scheduled automations. Call this only after the user confirms the exact watches to remove. This permanently deletes a root record. Before calling it, name exactly what will be deleted, explain that it cannot be undone, and wait for an explicit yes. Batch edits are bounded. Return one result per requested item, and report partial failures instead of describing the whole batch as successful.

Inputs: watchIds (required), confirmed (required), workspaceId. Returns: results, warnings

campaignstack_delete_contentDelete contentA
Destructive
Inspect

Permanently delete draft or failed content. Requires confirmed:true. This permanently deletes a root record. Before calling it, name exactly what will be deleted, explain that it cannot be undone, and wait for an explicit yes.

Inputs: confirmed (required), contentIds (required), workspaceId. Returns: ok, results, workspaceId

campaignstack_delete_lead_listDelete lead listA
Destructive
Inspect

Delete up to 50 lead lists from the bound workspace. The list row is removed and entries drain asynchronously. This permanently deletes a root record. Before calling it, name exactly what will be deleted, explain that it cannot be undone, and wait for an explicit yes.

Inputs: confirmed (required), leadListIds (required), workspaceId. Returns: ok, items, failed, results, +1 more

campaignstack_delete_signal_agentA
Destructive
Inspect

Delete signal agents by running the supported archive/remove contract and clearing linked watch pointers. Call this only after the user confirms the exact agents to remove. This permanently deletes a root record. Before calling it, name exactly what will be deleted, explain that it cannot be undone, and wait for an explicit yes. Batch edits are bounded. Return one result per requested item, and report partial failures instead of describing the whole batch as successful.

Inputs: agentIds (required), confirmed (required), workspaceId. Returns: results, warnings

campaignstack_delete_signal_watchA
Destructive
Inspect

Delete person signal watches by archiving them so they no longer refresh. Call this only after the user confirms the exact watches to remove. This permanently deletes a root record. Before calling it, name exactly what will be deleted, explain that it cannot be undone, and wait for an explicit yes. Batch edits are bounded. Return one result per requested item, and report partial failures instead of describing the whole batch as successful.

Inputs: watchIds (required), confirmed (required), workspaceId. Returns: results, warnings

campaignstack_delete_workflowDelete workflowA
Destructive
Inspect

Permanently delete one workflow and its associated records. Requires confirmed:true. Use edit_workflow set_enabled:false to pause without deleting history.

Inputs: confirmed (required), workflowId (required), workspaceId. Returns: ok, results, warnings, workflowId, +1 more

campaignstack_delete_workspaceA
DestructiveIdempotent
Inspect

Delete a workspace you own and cancel its subscription. Stops its workflows and scheduled work. Requires a user key or signed-in connection and confirmed:true. Confirm the exact workspace with the user before calling.

Inputs: confirmed (required), workspaceId (required). Returns: deleted, workspaceId

campaignstack_delete_workspace_memoryA
Destructive
Inspect

Permanently delete one workspace memory candidate or active fact. Requires confirmed:true. This permanently deletes a root record. Before calling it, name exactly what will be deleted, explain that it cannot be undone, and wait for an explicit yes.

Inputs: memoryId (required), confirmed (required), workspaceId. Returns: id, deleted, success

campaignstack_diagnose_workflow_sourcingDiagnose workflow sourcingA
Read-onlyIdempotent
Inspect

Explain why one lead did or did not enter one workflow, including score, dedup, exclusions, provider filters, and verdict. Read-only diagnostic only.

Inputs: workflowId (required), leadId, profileUrl, workspaceId. Returns: error, result

campaignstack_edit_accountEdit accountA
Destructive
Inspect

Edit one connected LinkedIn account. Use operation=set_timezone with an IANA timezone to move automation hours and budget days; null restores the country default. Use operation=set_owner_workspace to change the workspace that pays for maintenance/watchers. Use operation=set_watcher to turn the account's connection watcher, message watcher, own-posts watcher or invitation hygiene on or off (state is in campaignstack_get_accounts include.maintenance): name the account and kind and get a yes first, because enabling a watcher starts drafting outbound messages for review. The first enable creates the watcher and its review-first workflow, later calls pause or resume it; one watcher per kind per account across all workspaces (DUPLICATE_*_EXTERNAL means another workspace runs it; it has to be turned off there first) and it needs a live session (AGENT_ACCOUNT_UNAVAILABLE). Use operation=delete only after naming the account and getting explicit confirmation: delete removes the account from every workspace and clears its CampaignStack session/proxy/budgets/credentials, unlike campaignstack_edit_workspace_accounts unlink which only removes workspace access. With platform google, operation=set_scheduling_hours sets the hours meetings are offered in, which also bound when campaign email sends; it needs a person key and only changes a Google account that person connected.

Inputs: platform (required), operation (required), accountId, workspaceId. Returns: ok, item, results, warning, +1 more

campaignstack_edit_campaignEdit campaignA
Destructive
Inspect

Create, update, archive, unarchive, or edit singular campaign components: ICP, persona, roadmap phase, topic (add, reweight or remove by topicId or exact taxonomy name; an unknown one fails with the valid names), a LinkedIn search watch on a topic (search_watch: start, stop or resume it, and choose which of the workspace's LinkedIn accounts run it; it takes any taxonomy topic, even one the campaign does not carry; a workspace runs one watch per topic, so a second watch returns enabled:false with a disabledReason), signal weight, and automation state. An icp operation already schedules a rescore of already-scored leads automatically; set rescore:true (optionally with rescoreLeadIds) on it to also score every unscored campaign lead against the new criteria, since that is not automatic. This can spend credits, paid provider quota, or model-backed work. Tell the user the cost or budget meaning first when the tool reports one. Also permanently deletes a whole campaign and cascades its data (delete), or a single ICP, persona, or phase (delete_icp/delete_persona/delete_phase); each requires confirmed:true and cannot be undone, so name what will be deleted and wait for an explicit yes first. Use archive to stop work without deleting history, and restores a previous campaign guidance or offer addendum version with the top-level restoreVersionId field, using a versionId from campaignstack_get_campaigns' guidanceHistory include. Each operation returns success or failure independently. Score impacts and topic weights are bounded; archive requires confirmed:true and does not send messages; unarchive reverses it and needs no confirmation.

Inputs: operations (required), workspaceId, restoreVersionId. Returns: ok, results, warnings, workspaceId

campaignstack_edit_companiesEdit companiesAInspect

Patch per-workspace company overlays (tags/notes) for up to 50 shared or private companies. This does not mutate shared firmographics.

Inputs: companies (required), workspaceId. Returns: ok, items, failed, results, +1 more

campaignstack_edit_company_employee_watchAInspect

Create company employee watches or change their active/paused state. Call this when the user asks CampaignStack to watch a company for new hires or pause/resume an existing company watch. It does not delete watches; use campaignstack_delete_company_employee_watch for removal. Batch edits are bounded. Return one result per requested item, and report partial failures instead of describing the whole batch as successful.

Inputs: operations (required), workspaceId. Returns: results, warnings

campaignstack_edit_competitor_watchA
Destructive
Inspect

Create competitor watches, pause/resume them, assign LinkedIn accounts used by CampaignStack to observe those competitors, or delete them. Omit sendAccountIds (or pass []) to collect engagers without contacting anyone. A delete operation runs the same removal as campaignstack_delete_competitor_watch and requires confirmed:true on that operation. Batch edits are bounded. Return one result per requested item, and report partial failures instead of describing the whole batch as successful.

Inputs: operations (required), workspaceId. Returns: results, warnings

campaignstack_edit_contentEdit contentAInspect

Create, update, duplicate (copyFromId), or request legal desired-state transitions for content. generator.type=lead_magnet drafts a post from a lead-magnet candidate; generator.type=search_console_demand drafts a post from Search Console demand. Generators create draft posts only. Create accepts media; update accepts patch.media and patch.mediaMode (replace by default, append to add, [] to remove). Images use public URL or raw base64 with mimeType and optional alt: JPEG, PNG, GIF or WebP, 5 MB cap, blog upload URL safety checks. Existing mediaIds remain supported; postType is inferred when uploading media. Scheduling/approval/reject/retry transitions are validated and return per-operation results; nothing is published by this tool.

Inputs: operations (required), workspaceId. Returns: ok, results, workspaceId

campaignstack_edit_conversationsEdit conversationsA
Idempotent
Inspect

Batch small inbox state changes: mark conversations read, or cancel/edit a pending campaignstack_send_message sendAt row (scheduledMessages, id from campaignstack_get_conversations include.scheduledMessages). Sending remains campaignstack_send_message.

Inputs: workspaceId, conversations, scheduledMessages. Returns: ok, item, results, warning, +1 more

campaignstack_edit_draftEdit draftAInspect

Save new words on one draft without changing its approval state. Call this when the user asks to revise a draft for someone to sign off later, or to reword a message that is already approved and scheduled. Workflow drafts awaiting review can teach durable writing preferences from the edit immediately. To stop a scheduled draft, use campaignstack_recover_review with mode cancel_scheduled, or reopen for a conversation draft. A draft waiting for review stays waiting (a person still approves it, and the approval is recorded as an edit of the model's draft); a scheduled message keeps its slot; a job application keeps its status and is not approved. It never approves, rejects or sends (campaignstack_decide_review does that).

Inputs: entryId (required), confirmed (required), updatedContent (required), kind, +1 more. Returns: ok, results, warnings, workflowId, +1 more

campaignstack_edit_exclusionsAInspect

Edit exclusion policy for a workspace, signal agent, or workflow. Call this when the user asks to add/remove blocked profiles, change filter rules, or enable/disable exclusion rules. To block a profile, use add_manual_profiles without exclusionId: it adds to the workspace block list (created on first use) and stops running workflows from drafting to or contacting that person. Child targets are additive only: they can add restrictions but cannot release workspace restrictions. Batch edits are bounded. Return one result per requested item, and report partial failures instead of describing the whole batch as successful.

Inputs: target (required), operations (required), workspaceId. Returns: results, warnings

campaignstack_edit_identityEdit identity voiceAInspect

Edit or restore one identity voice profile language. Provide identityId, linkedinAccountId, or googleAccountId, or none of them when the workspace has one LinkedIn account; full core/overlays replace the stored object. Restores are versioned.

Inputs: language (required), core, active, overlays, +6 more. Returns: ok, results, warnings, workspaceId

campaignstack_edit_job_watcherEdit job watcherAInspect

Create, update, pause, or resume job watchers in one bounded request. Create binds a watcher to one connected LinkedIn account; update replaces searches only when queries is supplied and merges applicationProfile so old answers remain; pause stops new searches/applications; resume starts again if the account is usable. If outboundMode is automatic or applyEnabled is true, confirm that the user wants applications to proceed under the account's limits before calling.

Inputs: operations (required), workspaceId. Returns: ok, item, results

campaignstack_edit_lead_listEdit lead listAInspect

Create set/query lead lists, update list metadata/query, or add/remove up to 50 members. Query-list membership is predicate-owned; explicit add/remove requires an editable set list.

Inputs: op (required), name, query, members, +3 more. Returns: ok, items, failed, results, +1 more

campaignstack_edit_leadsEdit leadsAInspect

Apply bounded lead edits. Metadata patches write workspace overlays for global or private leads; contact patches are global-row contact updates only after workspace membership proof; profile link/unlink is provenance-bounded to private leads. Max 50 leads per call.

Inputs: leads (required), workspaceId. Returns: ok, items, failed, results, +1 more

campaignstack_edit_notificationsEdit notificationsAInspect

Mark notifications read by ids, or mark all notifications read for the resolved workspace. User-key and OAuth callers are restricted to their own notification rows plus legacy shared rows.

Inputs: all, ids, workspaceId. Returns: ok, failed, result, results, +1 more

campaignstack_edit_signal_agentA
Destructive
Inspect

Create signal agents, update their account assignment, script or approval mode (approvalMode: whether what the agent drafts waits for a person), pause/resume them, replace their additive exclusions, or restore a previous agentScript version (restore_guidance, using a versionId from campaignstack_get_signal_agents include.guidanceHistory). On a create, omitted accountIds and actAccountIds default to the workspace's only LinkedIn account; with several, ask the person which accounts read the target (accountIds) and which send and act (actAccountIds). actAccountIds [] means collect without contacting anyone. Setting status to 'archived' runs the same permanent removal as campaignstack_delete_signal_agent and requires confirmed:true on that operation. Batch edits are bounded. Return one result per requested item, and report partial failures instead of describing the whole batch as successful.

Inputs: operations (required), workspaceId. Returns: results, warnings

campaignstack_edit_signal_watchAInspect

Create person signal watches or change their active/paused state. Call this when the user asks CampaignStack to watch a lead for profile changes or pause/resume an existing person watch. It does not delete watches; use campaignstack_delete_signal_watch for removal. Batch edits are bounded. Return one result per requested item, and report partial failures instead of describing the whole batch as successful.

Inputs: operations (required), workspaceId. Returns: results, warnings

campaignstack_edit_userEdit userAInspect

Edit the calling user's notification preferences only. Requires a user API key; workspace keys cannot change personal preferences.

Inputs: notificationPreferences (required). Returns: ok, results, warnings, workspaceId

campaignstack_edit_workflowEdit workflowAInspect

Create workflows (from a graph, or from a whole workflow file exported by the workflow editor), edit metadata, set whether the workflow sends without review (approvalMode), enable or pause execution, set LinkedIn account pools, set email sender account, set workflow-level outreach intents, and restore a previous node instructions version (restore_guidance, using a versionId from campaignstack_get_workflows include.guidanceHistory). Graph changes belong in campaignstack_edit_workflow_graph. Each operation returns independently.

Inputs: operations (required), workspaceId. Returns: ok, results, warnings, workflowId, +1 more

campaignstack_edit_workflow_graphEdit workflow graphAInspect

Atomically patch or replace one workflow graph. replace supplies the full non-anchor graph. patch starts from the current non-anchor graph, upserts/removes nodes, and replaces edges when supplied. A patched node's config is merged into the stored one: only the keys you send change, and a key sent as null is cleared. Invalid graphs are rejected before persistence; active-lead safety remains enforced by the workflow backend. Optional split-node optimization runs after the graph write.

Inputs: mode (required), workflowId (required), graph, patch, +2 more. Returns: ok, results, warnings, workflowId, +1 more

campaignstack_edit_workspaceEdit workspaceAInspect

Write workspace settings, branding, compliance, restore one playbook version, or accept/reject the pending playbook proposal. Only supplied components change. Branding affects client-facing shared artifacts only; compliance toggles the fixed lawful-basis notice; playbook restores are versioned. Read the pending proposal with campaignstack_get_workspaces' playbookProposal include before deciding it.

Inputs: branding, settings, compliance, workspaceId, +2 more. Returns: ok, results, warnings, workspaceId

campaignstack_edit_workspace_accountsEdit workspace accountsA
Idempotent
Inspect

Link or unlink already-connected LinkedIn/Google accounts to a workspace. Nothing is re-authenticated or sent.

Inputs: operations (required), workspaceId. Returns: ok, item, results, warning, +1 more

campaignstack_edit_workspace_integrationsEdit workspace integrationsA
Destructive
Inspect

Link an integration the person owns into this workspace, unlink it from this workspace, or disconnect it by deleting this workspace's stored credentials. Disconnect and unlink do not undo provider-side data already sent.

Inputs: operations (required), workspaceId. Returns: ok, failed, result, results, +1 more

campaignstack_edit_workspace_membersEdit workspace membersA
Destructive
Inspect

Apply bounded member and invitation operations with per-operation results. Removes, invitations, and revocations require confirmed:true because they affect another person's access or send/cancel an email. Admin/owner grants are intentionally unavailable here.

Inputs: operations (required), workspaceId. Returns: ok, results, warnings, workspaceId

campaignstack_edit_workspace_memoryA
Idempotent
Inspect

Apply a pending workspace memory candidate so it becomes active in future prompts. A no-op when the memory is already active.

Inputs: memoryId (required), workspaceId. Returns: id, status, changed, success

campaignstack_enrich_companiesEnrich companiesBInspect

Run provider firmographic enrichment for up to 25 companies. Company firmographics are shared public-domain data; provider/API-key errors are reported per company.

Inputs: companyIds (required), provider, workspaceId, clayRoutineId. Returns: ok, items, failed, results, +1 more

campaignstack_enrich_leadsEnrich leadsAInspect

Run contact enrichment for up to 25 leads with bounded include flags. Email is on by default; phone must be explicitly requested and can cost more. Per-lead provider errors are returned independently. Charges vary by mode (email reveal: contact_email_reveal credits per result found; phone reveal: contact_phone_reveal credits per result found). campaignstack_estimate_cost accepts variant or variantCounts before you commit. Email is on by default; phone is charged separately only when include.phone is true.

Inputs: leads (required), include, provider, workspaceId, +1 more. Returns: ok, items, failed, results, +1 more

campaignstack_estimate_costA
Read-onlyIdempotent
Inspect

What a costed final MCP tool would spend, before you call it. Returns creditsPerUnit, the total for the count you name, the workspace's remaining free balance, and whether the spend is allowed. It only calculates, never reserves or spends. A paid workspace overflows to overage in its subscription currency (USD or EUR). The response includes the current overageRate and currency from billing. It reports a null balance rather than a refusal. Prices come from the same table the charge is made from, so single-action tools cannot drift from billing. For tools with multiple charge modes, pass variant for one mode or variantCounts for the full exposure; otherwise the tool refuses to guess.

Inputs: tool (required), count, variant, workspaceId, +1 more. Returns: per, note, tool, count, +11 more

campaignstack_explain_capabilitiesA
Read-onlyIdempotent
Inspect

Explain how the pieces of this workspace fit together, so you can pick the right tool without opening several to compare them. Call this when you are choosing between similar tools, when a request does not obviously map to one tool, or when an error tells you to consult it. topic: overview (default), watchers (how to choose between the eight ways to monitor something), workflows (how a workflow gets its leads, and which tool creates which kind), platforms (what can be acted on and what is tracking-only), outreach_and_safety (who sends, what the limits are, and what to confirm with the user first). It is a static reference and returns no live counts: each topic names the list tool that does (campaignstack_get_signal_agents, campaignstack_get_workflows, and so on). Costs nothing and reads no workspace data.

Inputs: topic. Returns: topic

campaignstack_explain_draftA
Read-only
Inspect

Why a draft says what it says: each sentence attributed to the piece of configuration that produced it (the offer context, a playbook section, the campaign guidance, an agent's script, the outreach intent, the voice profile, a node's instructions, an agent memory, the contact details, the calendar availability, the conversation so far, the product's built-in rules, the lead's own facts), plus the tool and field that changes each one. Works on in-app and MCP-crafted drafts alike. Every citation is re-checked against the prompt section it names, so an unverifiable one comes back as unattributed rather than as a plausible guess. Call this when a draft reads wrong and you need to know WHICH setting to edit; campaignstack_preview_craft_prompt shows the same prompt for a draft that has not been written yet. Pass include ["attempts"] to see whether a draft held for review was rewritten on each attempt and why each one was rejected. Charges mcp_craft credits per call; campaignstack_estimate_cost gives the number before you commit. One model call over a stored draft and its prompt; writes nothing.

Inputs: entryId, include, workspaceId, craftAuditLogId. Returns: draft, fixes, caller, verdict, +4 more

campaignstack_export_leads_to_crmExport leads to CRMA
DestructiveIdempotent
Inspect

Export up to 25 leads to a connected CRM provider. Today this supports HubSpot contact upsert, optional company association, and optional timeline note. This reaches a real person. Before calling it, tell the user exactly who it goes to and read the wording back to them, and wait for an explicit yes.

Inputs: leads (required), confirmed (required), note, provider, +3 more. Returns: ok, failed, result, results, +1 more

campaignstack_export_linkedin_connectionsExport LinkedIn connectionsAInspect

Start a runner job that exports an account's first-degree LinkedIn connections into a lead list. This does not send connection requests.

Inputs: listName, accountId, workspaceId, keepUpToDate, +1 more. Returns: ok, item, results, warning, +1 more

campaignstack_follow_profileA
Destructive
Inspect

Ask CampaignStack to follow a profile from a connected LinkedIn account, so their posts reach the owner's feed. Call this when the user asks to follow a specific person. platform is required (currently "linkedin"). Name the person with leadId, which is the supported form; profileUrl is accepted for someone already in the workspace. The workspace's only connected LinkedIn account is used automatically, or pass accountId (campaignstack_get_accounts). A follow is visible to that person and notifies them; it does not send a connection request, use campaignstack_send_connection_request for that. This reaches a real person. Before calling it, tell the user exactly who it goes to and read the wording back to them, and wait for an explicit yes. One person per call. There is no bulk form, and calling this in a loop over a list is not the supported way to reach a list: add the leads with campaignstack_edit_lead_list and run a workflow over it with campaignstack_trigger_workflow, which paces the sends and applies the review step the workspace configured. Business hours do not apply. CampaignStack's daily limit for this action (and the weekly limit for invitations) still applies and counts sends already queued, so a run of calls stops at the limit; the owner sets those limits in the app. A paused account, or an action LinkedIn has challenged, refuses the send whatever the limit.

Inputs: platform (required), leadId, accountId, profileUrl, +2 more. Returns: jobId, status, success

campaignstack_get_accountsGet accountsA
Read-onlyIdempotent
Inspect

Read one or many connected accounts. LinkedIn rows include safety by default because it gates outreach; maintenance is opt-in and bounded. Google rows omit LinkedIn-only safety, budgets and ramp fields and carry schedulingHours (meeting hours, default 9 to 18). Results are one page. When pagination reports more, say so rather than presenting the page as the whole set, and page through before counting anything.

Inputs: ids, limit, cursor, include, +2 more. Returns: items, slots, warnings, available, +6 more

campaignstack_get_billing_urlGet billing URLA
Read-onlyIdempotent
Inspect

Use this when the user wants to upgrade or downgrade their workspace subscription. Returns the workspace billing settings URL, where the signed-in human reviews and confirms the change. This read-only tool does not change billing or make a charge. Instead use campaignstack_get_workspaces for subscription state and balances.

Inputs: workspaceId. Returns: url, workspaceId, requiresHumanConfirmation

campaignstack_get_campaignsGet campaignsA
Read-onlyIdempotent
Inspect

Read campaigns and optional campaign planning details in one stable collection envelope. Use this when the user needs campaign identity, ICP, personas, roadmap, topics, weights, metrics, automations, job history, guidance history, ICP match scores, or the topic vocabulary (include.topicVocabulary, the ids a campaign topic can be set to). Pass include.personas.ids or include.roadmap.ids to target one specific persona or phase directly instead of scanning the capped list, and include.icpScores.icpId to read match scores for one ICP. Omit ids to list matching records, pass one id for one record, or pass several ids for several records. The response still uses the same collection envelope. Use include only for extra details the user asked for. Large, computed, historical, or external includes must use their documented caps and must report when data is truncated or paginated. This read-only tool does not change data. Instead use campaignstack_get_workspaces for workspace-level setup, and campaignstack_get_identities for sender accounts or voice. The result returns items, pagination, missingIds when requested ids are unavailable, and warnings for capped includes.

Inputs: ids, limit, cursor, filters, +2 more. Returns: items, warnings, missingIds, pagination, +1 more

campaignstack_get_companiesGet companiesA
Read-onlyIdempotent
Inspect

Read companies by ids or page workspace companies in one bounded V2 envelope. Optional includes cover signals, funding, discovery rows, employees, metric history, reviews, news, and a bounded jobs page with filters. Use nearest specialist tools for deeper company intelligence or refreshes; this read does not fetch new data. Results are one page. When pagination reports more, say so rather than presenting the page as the whole set, and page through before counting anything.

Inputs: ids, limit, query, cursor, +2 more. Returns: items, warnings, missingIds, pagination

campaignstack_get_company_employee_watchesA
Read-onlyIdempotent
Inspect

Get company employee watches with optional ids, company filter, state filter, and pagination. Returns the stable V2 collection shape and never leaks watches outside the resolved workspace.

Inputs: ids, limit, cursor, filters, +2 more. Returns: items, warnings, missingIds, pagination

campaignstack_get_competitor_watchesA
Read-onlyIdempotent
Inspect

Get competitor watches with optional ids, filters, pagination, and a bounded latestPosts include. Use this before changing a watch so the named target and current state can be read back.

Inputs: ids, limit, cursor, filters, +2 more. Returns: items, warnings, missingIds, pagination

campaignstack_get_contentGet contentA
Read-onlyIdempotent
Inspect

Read one or many content records in a stable items[] envelope. Includes can add bounded calendar projection, suggestions, media/full records, and analytics; include.limit caps each include. Results are one page. When pagination reports more, say so rather than presenting the page as the whole set, and page through before counting anything.

Inputs: ids, limit, cursor, filters, +2 more. Returns: items, warnings, missingIds, pagination

campaignstack_get_conversationsGet conversationsA
Read-onlyIdempotent
Inspect

Read inbox conversation summaries, unread count, or full threads when include.messages is true. Always returns a stable items/pagination envelope. A conversation with hasSuspiciousInbound true holds a message that offers something of value and asks the reader to act on it (the shape of a fake recruiter or investor): say so to the person before you book, open a link or reply on their behalf. include.scheduledMessages lists this thread's pending campaignstack_send_message sendAt rows. A workflow draft that has not gone out has delivered false and deliveredAt null; its nextSendAt is when it will send (null while it waits for a review).

Inputs: ids, limit, cursor, search, +4 more. Returns: items, slots, warnings, available, +6 more

campaignstack_get_craft_config_treeA
Read-onlyIdempotent
Inspect

Everything a craft prompt is built from, in one read: the workspace's five playbook sections with their caps, offerContext, capabilities and outreach intents; every campaign's guidance, offer addendum and intents; every workflow's craft nodes with the text a human wrote into them; every signal agent's script; and one voice profile summary per person and language. It also returns warnings: tone language in an agent script, a section written as an explanation of itself, the same sentence sitting at two levels, a campaign goal promising something the offer never states, a greeting that is really a whole message, a voice built on almost no real writing. Each warning names the tool and field that fixes it. Call this before editing craft configuration, or when the user asks to change how CampaignStack writes messages. It does not show one node's actual rendered prompt: campaignstack_preview_craft_prompt does that for a specific lead. Exemplars and message bodies are never returned.

Inputs: include, campaignId, workspaceId, resolveForWorkflowId, +1 more. Returns: voice, warnings, campaigns, workflows, +4 more

campaignstack_get_exclusionsA
Read-onlyIdempotent
Inspect

Get exclusion policy rows with a stable {items,pagination} shape. Target workspace returns level-0 policy; target signal_agent or workflow returns the effective cascade (workspace policy plus that consumer's additive exclusions). Use include.manualProfiles with a cap to inspect bounded manual profile samples.

Inputs: ids, limit, cursor, filters, +2 more. Returns: items, warnings, missingIds, pagination

campaignstack_get_identitiesGet identitiesA
Read-onlyIdempotent
Inspect

Read sender identities and optional account or voice details in one stable collection envelope. Use this when the user needs to resolve an identity, LinkedIn account, or Google account to the human sender and voice profile. Omit ids to list matching records, pass one id for one record, or pass several ids for several records. The response still uses the same collection envelope. Use include only for extra details the user asked for. Large, computed, historical, or external includes must use their documented caps and must report when data is truncated or paginated. This read-only tool does not change data. Instead use campaignstack_get_campaigns for campaign guidance, and campaignstack_get_workspaces for team or workspace settings. The result returns items, pagination, missingIds when requested ids are unavailable, and warnings for capped includes.

Inputs: ids, limit, cursor, filters, +2 more. Returns: items, warnings, missingIds, pagination

campaignstack_get_integrationsGet integrationsA
Read-onlyIdempotent
Inspect

Read connected integrations plus optional bounded provider resources and import progress. Secrets and OAuth tokens are never returned. Use this before provider export/import calls; it does not connect, disconnect, or send leads to a provider. Results are one page. When pagination reports more, say so rather than presenting the page as the whole set, and page through before counting anything.

Inputs: limit, providers, resources, workspaceId. Returns: items, warnings, resources, pagination, +1 more

campaignstack_get_job_applicationsGet job applicationsA
Read-onlyIdempotent
Inspect

Read job watcher results through one selector. selector.kind='matches' lists postings a watcher found, including found rows nobody acted on yet. selector.kind='applications' lists acted-on applications across the workspace. selector.kind='application' reads one full review payload with letter, screening answers, pending questions, account, resume, and confirmation. Use status filters for found/needs_answer/pending_review/submitted/etc. Results are one page. When pagination reports more, say so rather than presenting the page as the whole set, and page through before counting anything.

Inputs: selector (required), workspaceId. Returns: items, watcherId, missingIds, pagination

campaignstack_get_job_watchersGet job watchersA
Read-onlyIdempotent
Inspect

Read live job watchers for a workspace: connected LinkedIn account, saved searches, apply mode, application profile completeness, resume, and last search time. Pass ids to select specific watchers from the bounded live set; use this before editing, pausing, resuming, or archiving a watcher. This read does not include matched jobs; use campaignstack_get_job_applications for matches and applications.

Inputs: ids, status, workspaceId. Returns: items, watcherId, missingIds, pagination

campaignstack_get_lead_commentsGet lead commentsA
Read-only
Inspect

Freshly read recent LinkedIn comments for one lead/profile with a hard maxRequests cap. Nothing is written to the other platform and nothing is stored from this read. Charges recent_comments_read credits per call; campaignstack_estimate_cost gives the number before you commit. Charged per Voyager call the runner makes, not per comment returned. The workspace's own connected account reads free.

Inputs: untilDate (required), profileUrl (required), accountId, maxRequests, +1 more. Returns: ok, items, failed, results, +1 more

campaignstack_get_lead_enrichmentsGet lead enrichmentsA
Read-onlyIdempotent
Inspect

List recent workspace lead enrichment rows for async audit: type enrichment_queue (default) covers profile/company refresh jobs, type contact_finder covers built-in email/phone lookups (pass requestId for one specific lookup). The two job kinds have different status vocabularies and are never merged into one list. Use when queue counts are not enough and the user needs bounded row-level evidence. It reads audit state only and never retries work. Results are one page. When pagination reports more, say so rather than presenting the page as the whole set, and page through before counting anything.

Inputs: type, limit, cursor, status, +2 more. Returns: items, warnings, missingIds, pagination

campaignstack_get_lead_listsGet lead listsA
Read-onlyIdempotent
Inspect

Read lead lists by ids, or list a campaign's or the workspace's lead lists, narrowed by origin (source) or by LinkedIn account (accountId: that account's connections, invited and post engagers lists). Optionally includes a bounded page of members. Use this when the user asks which lists exist or who is in them; it does not add or remove members. Results are one page. When pagination reports more, say so rather than presenting the page as the whole set, and page through before counting anything.

Inputs: ids, limit, cursor, source, +4 more. Returns: items, warnings, missingIds, pagination

campaignstack_get_lead_refresh_statusGet lead refresh statusA
Read-onlyIdempotent
Inspect

Read workspace lead-enrichment queue status counts, optionally for one campaign. Use this when the user asks whether lead refresh/enrichment is moving; it does not enqueue refreshes. completed and failed cover about the last 6 hours, not all time.

Inputs: campaignId, workspaceId. Returns: items, total, failed, pending, +7 more

campaignstack_get_leadsGet leadsA
Read-onlyIdempotent
Inspect

Read leads by ids or search filters in one bounded V2 envelope. Includes are capped; campaign-specific scoreBreakdown requires campaignId. Use this instead of legacy search/get pairs when the user needs leads plus selected context. It does not modify leads or enqueue outreach. Results are one page. When pagination reports more, say so rather than presenting the page as the whole set, and page through before counting anything.

Inputs: ids, limit, query, cursor, +4 more. Returns: items, warnings, missingIds, pagination

campaignstack_get_notificationsGet notificationsA
Read-onlyIdempotent
Inspect

Read notifications in a stable envelope with the unread count for the same caller. Use this when the user asks what needs attention; it does not change read state. Results are one page. When pagination reports more, say so rather than presenting the page as the whole set, and page through before counting anything.

Inputs: type, limit, cursor, workspaceId. Returns: items, warnings, resources, pagination, +1 more

campaignstack_get_platform_capabilitiesA
Read-onlyIdempotent
Inspect

Lists every platform action (like_post, comment_on_post, follow_profile, send_message, send_connection_request, send_inmail, extract_post_engagements) with the platforms it supports and, per platform, the accepted target fields and optional params. Returns platforms, which says for each platform whether accounts can be connected, how many this workspace has, and which actions exist there: a platform with no actions is tracking-only by design, not unfinished. connectedPlatforms is the older shape and lists only platforms with a connected account. Call this when an action tool returns wrong_target_for_platform or unsupported_capability.

Inputs: workspaceId. Returns: actions, platforms, connectedPlatforms

campaignstack_get_reviewsGet reviewsA
Read-onlyIdempotent
Inspect

Read pending review drafts and optional pending count in one stable envelope. Pass entryId to inspect an approved review and see its send state and wait reason. Each item carries what a decision needs: the draft text and the key it sits under, the lead's name, headline, company and profile URL, the channel, the workflow or campaign, the sending account and any flag with its reason. Use include only for extra details the user asked for. Large, computed, historical, or external includes must use their documented caps and must report when data is truncated or paginated. Includes add the lead's full profile, the thread, what the draft answers, the stored draft with its reasoning, and the critic history. Items include kind: workflow for drafted workflow touches and kind: job_application for job watcher cover letters; pass that kind back to campaignstack_decide_review or campaignstack_edit_draft. Use before deciding a review; a page is not the whole queue.

Inputs: limit, cursor, entryId, include, +3 more. Returns: items, warnings, missingIds, pagination, +1 more

campaignstack_get_review_timeGet review timeA
Read-onlyIdempotent
Inspect

Get the calling user's time-in-review history: an 8-week series of minutes spent in the inbox review queue, this week's decisions split by who made them (app, your agent via MCP, automatic), and whether minutes per decision fell from last week. Call this when the user asks how much time they're spending reviewing drafts, or whether the queue is getting faster. Renders no series (empty weeks, distinctWeeksWithData omitted) until at least 2 weeks of data exist. Requires a USER API key (csu_); workspace keys have no user identity.

Inputs: no named parameters. Returns: weeks, decisionsThisWeek, distinctWeeksWithData, minutesPerDecisionTrend

campaignstack_get_signal_agentsA
Read-onlyIdempotent
Inspect

Get signal agents with optional ids, filters, pagination, and bounded stats/leads includes. Account-maintenance watchers stay hidden unless includeAccountMaintenance is true.

Inputs: ids, limit, cursor, filters, +2 more. Returns: items, warnings, missingIds, pagination

campaignstack_get_signal_watchesA
Read-onlyIdempotent
Inspect

Get person signal watches with optional ids, filters, and pagination. Returns only watches inside the resolved workspace; missing requested ids are reported as missingIds.

Inputs: ids, limit, cursor, filters, +2 more. Returns: items, warnings, missingIds, pagination

campaignstack_get_workflow_node_typesGet workflow node typesA
Read-onlyIdempotent
Inspect

Discover supported workflow node types before graph edits. Includes graph shape docs, examples, and webhook templates for integration:webhook nodes.

Inputs: includeWebhookTemplates. Returns: nodeTypes, graphShape, exampleGraph, webhookTemplates

campaignstack_get_workflowsGet workflowsA
Read-onlyIdempotent
Inspect

Read one or many workflows in a stable envelope. Omit ids to list matching records, pass one id for one record, or pass several ids for several records. The response still uses the same collection envelope. Use include only for extra details the user asked for. Large, computed, historical, or external includes must use their documented caps and must report when data is truncated or paginated. Includes can add graph, stats, bounded split optimization history, node activity, and leads at nodes.

Inputs: ids, limit, nodes, cursor, +3 more. Returns: items, warnings, missingIds, pagination

campaignstack_get_workspace_memoriesA
Read-onlyIdempotent
Inspect

List workspace memories: active notes already used in prompts first, then candidate notes awaiting confirmation, each newest first. Each item carries status, observations (a candidate goes live at 3, from three different leads), direction (include: asks drafts to contain something; avoid: asks them to leave it out) and source (agent, user, user_edit for app edits, or mcp_edit for MCP edits). Page with pagination.nextCursor while pagination.hasMore is true.

Inputs: limit, cursor, category, workspaceId. Returns: data, pagination

campaignstack_get_workspacesGet workspacesA
Read-onlyIdempotent
Inspect

Read workspaces and optional workspace details in one stable collection envelope. Use this when the user needs workspace identity, settings, members, billing, credits, metrics, setup, branding, compliance, playbook, playbook history, or the pending playbook proposal. Omit ids to list matching records, pass one id for one record, or pass several ids for several records. The response still uses the same collection envelope. Use include only for extra details the user asked for. Large, computed, historical, or external includes must use their documented caps and must report when data is truncated or paginated. This read-only tool does not change data. Instead use campaignstack_get_campaigns for campaign setup, and campaignstack_get_identities for account or voice ownership. The result returns items, pagination, missingIds when requested ids are unavailable, and warnings for capped includes.

Inputs: ids, limit, cursor, filters, +2 more. Returns: items, warnings, missingIds, pagination

campaignstack_import_external_leadsImport external leadsAInspect

Import leads from a provider list, campaign, or portal into CampaignStack. The job may continue server-side; read progress through campaignstack_get_integrations resources.importProgress.

Inputs: provider (required), sourceId (required), sourceKind (required), campaignId, +2 more. Returns: ok, failed, result, results, +1 more

campaignstack_import_leadsImport leadsAInspect

Import parsed CSV lead rows into private workspace leads with CSV provenance, optionally attaching successful rows to one lead list. Max 100 rows per call; per-row failures are returned without aborting the whole import.

Inputs: rows (required), leadListId, workspaceId, importedFrom. Returns: ok, items, failed, results, +1 more

campaignstack_like_postA
Destructive
Inspect

Ask CampaignStack to like a lead's most recent post from a connected LinkedIn account, as a light way to be seen before writing. Call this when the user asks to like or engage with a specific lead's post before an outreach step. platform is required (currently "linkedin"). Name the person with leadId, which is the supported form; profileUrl is accepted for someone already in the workspace. The workspace's only connected LinkedIn account is used automatically, or pass accountId (campaignstack_get_accounts). A like is public and puts a notification on that person's account under the owner's name. It does not comment; use campaignstack_comment_on_post for that. This reaches a real person. Before calling it, tell the user exactly who it goes to and read the wording back to them, and wait for an explicit yes. One person per call. There is no bulk form, and calling this in a loop over a list is not the supported way to reach a list: add the leads with campaignstack_edit_lead_list and run a workflow over it with campaignstack_trigger_workflow, which paces the sends and applies the review step the workspace configured. Business hours do not apply. CampaignStack's daily limit for this action (and the weekly limit for invitations) still applies and counts sends already queued, so a run of calls stops at the limit; the owner sets those limits in the app. A paused account, or an action LinkedIn has challenged, refuses the send whatever the limit.

Inputs: platform (required), leadId, accountId, profileUrl, +2 more. Returns: jobId, status, success

campaignstack_lookup_companyLook up a company on LinkedInAInspect

Call this when the user asks for a company that campaignstack_get_companies with a query does not find. With query, CampaignStack searches the web for matching LinkedIn company pages and returns up to 5 (companyId is set when CampaignStack already holds one). With linkedinUrl, CampaignStack reads that public company page and adds the company to the shared company database; the answer is ready with a companyId, or queued (call again with the same linkedinUrl to check), or failed. Pass exactly one of the two. Free, up to 50 searches and 50 new pages per workspace per day. It does not add the company to a campaign or reach anyone at it.

Inputs: query, linkedinUrl, workspaceId. Returns: ok, items, failed, results, +1 more

campaignstack_onboard_workspaceAInspect

Set up a CampaignStack workspace: who the user works for, what that company sells, why it exists, how they write, and who they sell to.

Call this with no arguments FIRST. It returns what CampaignStack already knows, including the employer company resolved from the user's LinkedIn profile and any facts already gathered on it, plus every field you can write and its current value. Read that before you propose anything: your job is to correct and complete what is there, not to describe the company from scratch.

Then call it again with the fields you are confident about. Confidence means you can point to something specific: their profile, their company page, their site, something they told you. When you are not confident about a field, leave it out. An empty field is fixed in ten seconds by the user. A plausible sentence that is wrong gets read into every message the system writes for them, and nobody catches it until a prospect does.

Report back in plain language what you set and what you left empty, so the user can correct you before any of it reaches a message.

Inputs: voice, firstIcp, whatWeSell, companyName, +2 more. Returns: next, note, state, written

campaignstack_preview_craft_promptA
Read-onlyIdempotent
Inspect

Render the exact prompt a craft node would send for one lead, without sending it. Call this when the user asks to see what a draft would look like, or to debug why a node's wording keeps going wrong. Runs the real craft path and stops immediately before the model call, so what comes back is the same bytes the model would read. index lists the named blocks in the system prompt with their character offsets and sizes (COMPANY CONTEXT, WHAT WE WANT, the playbook sections, RED LINES, the voice), which is how you find out WHICH piece of configuration produces a phrase you do not want. Writes no audit row, records no AI usage, assigns no voice experiment arm, fetches no web context, and costs nothing. Pair it with campaignstack_get_craft_config_tree, which shows where each block comes from.

Inputs: leadId (required), nodeId (required), workflowId (required), include, +2 more. Returns: chars, index, caller, nodeType, +2 more

campaignstack_propose_workspace_playbook_changePropose workspace playbook changeBInspect

Ask CampaignStack to draft a proposed playbook change for review instead of directly overwriting the live playbook.

Inputs: instruction (required), sectionKey, workspaceId. Returns: ok, results, warnings, workspaceId

campaignstack_recover_reviewRecover reviewAInspect

Recover one review with mode retry, restore_superseded_draft, reopen, or cancel_scheduled. To stop an approved scheduled workflow draft before sender handoff, use mode cancel_scheduled to cancel one draft, or mode reopen for a conversation draft: it returns the same text to pending review and releases its slot. Then decide_review reject can discard it. Once handed to the sender, both modes return an error. Retry may spend a fresh model generation; reopen validates downstream safety.

Inputs: mode (required), entryId (required), confirmed (required), workspaceId. Returns: ok, results, warnings, workflowId, +1 more

campaignstack_refresh_companiesRefresh companiesAInspect

Start a full company refresh for up to 25 companies. There is no sections argument: the backend schedules all available company-refresh components and returns per-company partial status including gates and already-queued rows. Charges company_refresh credits per call; campaignstack_estimate_cost gives the number before you commit.

Inputs: companyIds (required), workspaceId. Returns: ok, items, failed, results, +1 more

campaignstack_refresh_company_profileRefresh company profileAInspect

Use this when the user asks to read the workspace's own website into its company profile again, the facts every message is grounded on. It reads up to six pages of the workspace website and writes what they say into the profile fields nobody typed by hand; fields a person or an agent wrote stay as written. It returns the fields updated, the fields kept, or why it did not run. When what the site says would not fit the 2,500 character company context, nothing is written and the error names the fields that did not fit. Instead use campaignstack_edit_workspace to set profile fields directly, or campaignstack_analyze_company_website to read another site without storing anything. Charges website_intel credits per call; campaignstack_estimate_cost gives the number before you commit. Charged once per workspace and domain per day, and only when the profile is written. Refresh observes current first-party or public facts again. Say what was refreshed and how fresh the result is; do not imply it found contact data unless this tool says so. This can spend credits, paid provider quota, or model-backed work. Tell the user the cost or budget meaning first when the tool reports one.

Inputs: workspaceId. Returns: ok, results, warnings, workspaceId

campaignstack_refresh_identity_voiceRefresh identity voiceBInspect

Queue voice re-extraction for all LinkedIn accounts linked to an identity visible in the workspace.

Inputs: identityId, workspaceId, googleAccountId, linkedinAccountId. Returns: ok, results, warnings, workspaceId

campaignstack_refresh_inboxAInspect

Ask CampaignStack to poll its connected LinkedIn accounts for new messages now. Call this when the user wants the inbox brought up to date before you read it. Respects a 20-second freshness dedup window per account. Returns { refreshed, skipped, errors }. It does not fetch one thread's older history: use campaignstack_get_conversations with ids and include.messages for that.

Inputs: workspaceId. Returns: errors, skipped, refreshed

campaignstack_refresh_leadsRefresh leadsAInspect

Queue up to 50 LinkedIn lead profiles for fresh enrichment. normal joins the standard queue, which reads during the account's business hours. high with up to 5 profiles reads them right away, even outside business hours, like opening the profiles yourself; high with more goes to the front of the queue and still waits for business hours. Daily limits, pauses and a disconnected account still apply. Enrichment from an existing LinkedIn URL costs no credits; identity finding is charged only when found.

Inputs: results (required), priority, accountId, campaignId, +1 more. Returns: ok, items, failed, results, +1 more

campaignstack_refresh_workspace_playbookRefresh workspace playbookAInspect

Regenerate the workspace playbook from current campaigns, ICPs, personas, and the company profile. User-authored passages are preserved and prior text is snapshotted. A refresh that did not run (no live campaign, out of credits, a managed playbook) comes back with ok false and the reason; a section left empty because it stated company facts the profile does not hold comes back as a warning.

Inputs: workspaceId. Returns: ok, results, warnings, workspaceId

campaignstack_retry_workflow_entriesRetry workflow entriesAInspect

Retry one or many failed workflow entries by explicit entry ids, or retry all failed entries at one node. Use selector type retry_waits after deploying a failure fix to wake auto_retry or condition_retry waits at one node, optionally matching errorMessageContains. Dispatch safety gates still apply. Follow the returned cursor until remaining is 0; remaining is a continuation indicator, not a total. Returns per-entry partial results for explicit ids and node batch counts for node selectors.

Inputs: selector (required), workspaceId. Returns: ok, results, warnings, workflowId, +1 more

campaignstack_send_connection_requestA
Destructive
Inspect

Ask CampaignStack to send one connection request from a LinkedIn account the workspace owner connected. Call this when the user has named one specific person and confirmed sending the request. platform is required ("linkedin"; connection requests are a LinkedIn-only concept, and on other networks you follow instead with campaignstack_follow_profile). Name the person with leadId, which is the supported form; profileUrl is accepted for someone already in the workspace. An optional note is capped at 300 characters by LinkedIn. The workspace's only connected LinkedIn account is used automatically, or pass accountId (campaignstack_get_accounts). This reaches a real person. Before calling it, tell the user exactly who it goes to and read the wording back to them, and wait for an explicit yes. It goes out as soon as this call succeeds: there is no draft state and nothing to recall. One person per call. There is no bulk form, and calling this in a loop over a list is not the supported way to reach a list: add the leads with campaignstack_edit_lead_list and run a workflow over it with campaignstack_trigger_workflow, which paces the sends and applies the review step the workspace configured. Business hours do not apply. CampaignStack's daily limit for this action (and the weekly limit for invitations) still applies and counts sends already queued, so a run of calls stops at the limit; the owner sets those limits in the app. A paused account, or an action LinkedIn has challenged, refuses the send whatever the limit.

Inputs: platform (required), note, leadId, accountId, +3 more. Returns: jobId, status, success

campaignstack_send_inmailA
Destructive
Inspect

Ask CampaignStack to send one InMail to someone the owner is not connected to, from a connected LinkedIn account. Call this when the user has confirmed subject and body for a named person the owner is not connected to; for a connected lead use campaignstack_send_message instead. platform is required ("linkedin"; InMail is a LinkedIn-only premium product and needs Premium, Sales Navigator or Recruiter on the sending account, so a free account has a budget of zero here). Name the person with leadId, which is the supported form; profileUrl is accepted for someone already in the workspace. subject is capped at 200 characters and messageText at 1900. The workspace's only connected LinkedIn account is used automatically, or pass accountId (campaignstack_get_accounts). This reaches a real person. Before calling it, tell the user exactly who it goes to and read the wording back to them, and wait for an explicit yes. It goes out as soon as this call succeeds: there is no draft state and nothing to recall. One person per call. There is no bulk form, and calling this in a loop over a list is not the supported way to reach a list: add the leads with campaignstack_edit_lead_list and run a workflow over it with campaignstack_trigger_workflow, which paces the sends and applies the review step the workspace configured. Business hours do not apply. CampaignStack's daily limit for this action (and the weekly limit for invitations) still applies and counts sends already queued, so a run of calls stops at the limit; the owner sets those limits in the app. A paused account, or an action LinkedIn has challenged, refuses the send whatever the limit.

Inputs: subject (required), platform (required), messageText (required), leadId, +4 more. Returns: jobId, status, success

campaignstack_send_messageA
Destructive
Inspect

Ask CampaignStack to send one message into a LinkedIn conversation the owner's connected account is already part of, optionally with attachments given as public URLs. Takes a conversationRef { platform, id } from campaignstack_get_conversations. Read the thread with campaignstack_get_conversations using ids and include.messages first. If the last message is the owner's and the other person has not answered since, do not send: say so, and send only if the user confirms they want to follow up again. This reaches a real person. Before calling it, tell the user exactly who it goes to and read the wording back to them, and wait for an explicit yes. The send is dispatched asynchronously and the response carries a messageEntityUrn; campaignstack_get_conversations reports delivery. Pass sendAt to hold the message and send it later instead of now; it is cancelled if the other person replies first, so read the wording and the time back and wait for a yes. campaignstack_get_conversations with include.scheduledMessages lists pending ones; campaignstack_edit_conversations cancels or edits one. One person per call. There is no bulk form, and calling this in a loop over a list is not the supported way to reach a list: add the leads with campaignstack_edit_lead_list and run a workflow over it with campaignstack_trigger_workflow, which paces the sends and applies the review step the workspace configured. Business hours do not apply. A message past the account's daily message limit is still sent, and the result carries a warning to pass on to the user. A paused account, or messaging LinkedIn has challenged, refuses the send.

Inputs: text (required), conversationRef (required), sendAt, attachments, +2 more. Returns: sendAt, status, warning, messageEntityUrn, +1 more

campaignstack_submit_feedbackAInspect

Submit a feedback report or bug report to the CampaignStack team. Call this when the user asks to report a bug, request a feature, or send feedback from this conversation. Mirrors the in-app feedback widget: same store, same email notification. Categories: 'bug', 'feature', 'question', 'other'. It does not open a support ticket or get a reply back through this connection.

Inputs: route (required), message (required), category (required), workspaceId. Returns: reportId, submitted

campaignstack_suggest_job_application_profileSuggest job application profileA
Read-onlyIdempotent
Inspect

Propose answers for a job watcher's application profile from the person's own LinkedIn profile and resume: searches from their current titles, the country they live in, the languages their profile and resume are written in or name, the local currency, and years per listed skill (added up from the dated positions judged to use that skill; a skill with no years was not matched to any position). Nothing is saved. Call this before filling a profile, read the proposal back to the person, and save only what they confirm with campaignstack_edit_job_watcher (op update). Living in a country is not a work permit, so ask. Salary, notice period and relocation are never proposed: ask for them.

Inputs: watcherId (required), workspaceId. Returns: skills, currency, searches, languages, +4 more

campaignstack_trigger_workflowTrigger workflowAInspect

Confirmed consequential execution: adds the selected leads to a workflow, or with no leadIds runs its sourcing now. A lead enters a workflow once: leads already in it, in any status, and leads an exclusion matches are skipped and counted in skipped. It joins the workflow queue and obeys pacing, review, account, and safety limits. To retry failed entries, use campaignstack_retry_workflow_entries.

Inputs: confirmed (required), workflowId (required), leadIds, workspaceId. Returns: ok, results, warnings, workflowId, +1 more

campaignstack_whoamiA
Read-onlyIdempotent
Inspect

Who this connection is and what it may do: the workspace it is bound to, the permissions the owner approved, and whether it is an API key or a connected app. Call it when a tool asks which workspace a call is for (a person in several workspaces names one; with only one, workspaceId defaults to it), and use campaignstack_explain_capabilities when the question is which tool to reach for rather than what is permitted. It reads nothing about leads or campaigns.

Inputs: include. Returns: user, scopes, keyType, keyPrefix, +6 more

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      Not graded
      quality
      C
      maintenance
      Enables AI assistants to run an entire B2B prospecting workflow from one request: searching people and companies, enriching verified emails and phone numbers, monitoring buying signals, researching accounts, and building lists. It also generates and runs multi-step email, LinkedIn and calling campaigns, with everything created as drafts until explicitly launched.
      MIT
    • A
      license
      Not graded
      quality
      B
      maintenance
      Enables AI agents to find intent-signaling buyers and run LinkedIn, email, and X outreach from connected accounts while respecting platform limits. Every reply is handed back to a human for approval or decision.
      4
      MIT
    • A
      license
      Not graded
      quality
      C
      maintenance
      Enables AI agents and Google Sheets to verify emails, enrich company and person data from web and LinkedIn sources, score ICP fit, run scheduled multi-step lead workflows, and publish personalized prospect microsites at scale.
      1
      MIT
    Try in Browser

    Glama MCP Gateway

    Add one secure layer between your agents and this server.

    Resources