MCP Integration: Tool Reference

Available Tools

The Summit53 MCP server is the distribution layer for the Revenue Engine: it lets compatible assistants diagnose, decide, act, and preserve evidence around the Figure-Eight revenue loop. The server registers 116 functions across 18 categories. This reference documents the 102 functions that can be offered to external assistants; the remaining 14 are registered write functions that Summit53 holds back from external clients (listed under “Not available to external assistants” in the Access Model section).

You do not need to call functions by name — describe the outcome you need and the AI assistant can select from the functions available to the current session. Visibility depends on granted scopes, Summit53 organisation permissions, and the type of connection; call list_available_tools for the authoritative session-specific list. The in-product Ask Summit53 assistant is a separate surface and does not mirror this external catalogue.

Read functions never need approval. Functions marked (write · approval) change nothing until the requesting user approves the exact change on a Summit53 review page, and they are only offered when action approvals are enabled and the assistant connects through a trusted connection set up by Summit53. batch_update_outreach_leads stages several lead updates for one review page, and cancel_job is a job-management write that needs no approval. See the Access Model section below and the approval flow in the MCP Setup Guide.

Search

Find or fetch accounts and opportunities by name, filters, or ID.

get_opportunity

Get full details for a specific opportunity or deal by ID, including stage, amount, owner, health score, close date, and more. Use this when you already know the deal ID and need its complete record.

search_opportunities

Search and filter CRM opportunities and deals by name, stage, account, and value range. Useful for finding specific deals or building filtered lists for review.

search_accounts

Search and filter CRM accounts by name, industry, owner, and revenue. Returns account details and key metrics.

Forecasting

Forecast confidence and its drivers, forecast gap diagnosis, forward and reverse capacity planning, the Revenue Engine snapshot and context, segment and ARR operating views, ARR growth, hunter/farmer mix, and ramp cohorts.

forecast_confidence

Forecast data with confidence scores, rep performance metrics, pipeline velocity, and quota coverage. The primary tool for understanding forecast health.

forecast_confidence_decomposition

Explain why forecast confidence moved, using the same previous-versus-current contributor breakdown shown on the Revenue Engine page and based on stored contributor-level evidence.

forecast_gap_diagnosis

Diagnose the forecast gap by decomposing it into ramp capacity, productivity shortfall, and residual, using total-ARR rep quota attainment. Shows where the gap to target actually comes from.

capacity_to_target_planner

Plan whether current quota, ramp, forecast, and pipeline capacity can support a bookings target. Group by all sellers, team, sales role, segment, Sales Org region (sales_region) or Market region (market_region), with level set to leaf or top for the region groupings.

reverse_capacity_planner

Work backwards from a bookings target to the selling capacity and support roles required — AEs, BDRs, SEs, and managers — using configurable ratios, ramp factors, and average quota assumptions. Supports the same groupings as the capacity-to-target planner: all sellers, team, sales role, segment, sales_region or market_region, with leaf or top level for regions.

unified_forecast_summary

Executive and CRO-level view of revenue intelligence: pipeline coverage, team metrics, and quota tracking in a single summary. The go-to tool for leadership briefings.

revenue_engine_snapshot

Get the complete organisation Revenue Engine Figure-Eight snapshot used by the Revenue Engine page, including stage metrics, flow, conversion, velocity, and target context.

revenue_engine_context

Get one read-only bundle of Revenue Engine context in a single call. Choose the sections you need: snapshot, mechanic, engine state, readiness, forecast decomposition, segment performance, ARR operating summary, hunter/farmer mix, ramp cohort, unified forecast, capacity to target, or reverse capacity.

revenue_mechanic_diagnosis

Get the persona-aware Revenue Mechanic diagnosis, or a fast preview, as used by the Revenue Engine page. Includes the ranked drag, benchmark mode, interventions, and action context.

segment_performance_summary

Period-aware performance by segment (the default), Sales Org region or Market region, covering active contract ARR, booked ARR, win rate, pipeline coverage, velocity, and retention. Choose leaf or top level, add a secondary dimension for a cross-tab (for example Sales Org by Market), and pick summary, standard, or full detail.

arr_operating_summary

The canonical read-only ARR operating view: active contract ARR stock, booked ARR flows, retention estimates, coverage, and reconciliation, each tied to explicit period anchors. The by-group breakdown follows the chosen dimension (segment, Sales Org region or Market region) and level (leaf or top).

arr_growth_analytics

ARR growth metrics including total ARR, net retention rate, and growth stage classification. Supports trend analysis over configurable time periods.

arr_growth_narrative

AI-generated ARR growth narrative with recommendations and stage transition hints. Designed for board presentations and QBR preparation.

hunter_farmer_mix

Per-rep deal-mix classification (hunter / farmer / hybrid / unknown) over the rolling closed-won window. Manager-gated — individual contributors receive an empty payload with a manager-role warning.

ramp_cohort

Per-cohort and per-rep ramp progress for reps still in their ramp window: expected vs actual attainment and status (ahead / on track / behind / no quota). Manager-gated.

Pipeline & Risk

Portfolio and deal-level risk, heatmaps, and deal-drag diagnosis.

pipeline_risk_summary

Pipeline risk analysis including high-risk deals, risk distribution across stages, and overall confidence. Ideal for pipeline review meetings and forecast calls.

risk_heatmap

Paginated deal-level heatmap built on healthScore, which runs from 0 to 1 where higher means healthier. The riskiest deals (lowest health) are listed first by default, with risk factors for each deal. Filter by risk level, rep, stage, forecast category, close date, and amount.

deal_drag_summary

Summary metrics for Deal Drag analysis: stalled deal counts, average waste scores, risk distribution, and top qualification gaps across the pipeline. A quick health check on deal velocity.

deal_drag_dashboard

Full Deal Drag dashboard with paginated stalled opportunities, waste score breakdowns, drag patterns, and recovery strategies. Filter by score range, stage, activity level, or specific reps.

RevOps

CRM data quality, revenue data readiness, and sync health for revenue operations leaders.

data_quality_score

Organisation-wide CRM data quality score: overall completeness, the top issues, and the opportunities needing attention. Mirrors the RevOps data quality home card.

organization_revenue_readiness

Get the exact organisation deal-history counts and machine-learning readiness flags behind the Revenue Engine readiness indicators.

crm_sync_health

Per-CRM sync health for the org’s enabled CRMs (Salesforce, HubSpot, JobAdder): last successful sync, 7-day error rate, average latency, and a rolled-up health status.

Activity

Track engagement, identify neglected accounts, and get AI-prioritised action plans for individuals and teams.

activity_gaps

Find accounts with no recent engagement — accounts that have not been touched in a configurable number of days. Essential for maintaining healthy account coverage.

weekly_action_plan

Prioritised weekly action plan with tasks, risk summaries, and suggested actions. The AI analyses your pipeline and generates the highest-impact activities for the week ahead.

team_action_plan

Team-wide action plan overview for sales leaders. Shows each rep’s completion rate, high-priority items, flagged deals, and missed actions. Filter by week, rep, completion rate, or urgency level.

Opportunity Strategy

Strategic narrative artefacts for individual opportunities — the signals, rescue packets, assertive hypotheses, discovery questions, value framing, and stakeholder maps that drive customer-facing meetings.

get_opportunity_signals

Get the Revenue Signal Layer summary for one deal: the first constraint to fix, ranked signals with their evidence, confidence, and severity, win-probability movement, and signal coverage. Available to leaders and the owning rep only.

prepare_dark_deal_rescue

Prepare a read-only, decision-ready rescue packet for one opportunity that has gone quiet, using CRM activity, stakeholder roles, close timing, stored risk and signals, and published Knowledge Base assets. Returns draft CRM updates and re-engagement proposals in shadow mode: nothing is saved, sent, scheduled, queued, or executed. Available to leaders, admins, and the owning rep.

opportunity_three_whys

Get the Three Whys strategic narrative for an opportunity: executive summary, integrated narrative covering Why Anything / Why Us / Why Now, and a set of power phrases the rep can drop into a customer conversation.

opportunity_discovery_planner

Get the Discovery Planner for an opportunity: executive summary, structured discovery guide, and the priority questions to ask on the next call. Designed for first-meeting prep and sales coach hand-offs.

opportunity_value_pyramid

Get the Value Pyramid analysis for an opportunity: executive summary, layered value pyramid (functional → financial → strategic), and stakeholder-specific messaging that ladders up from features to business outcomes.

opportunity_stakeholder_intelligence

Get stakeholder intelligence for an opportunity: stakeholder profiles, org structure, recommended engagement moves, coverage gaps, and a buying-committee risk summary. The single view of who’s on the deal and where the air-cover is missing.

Revenue Loop

Diagnose the Figure-Eight revenue loop across its pipeline and coaching lanes, read the stage overlay, and track interventions in the Intervention Ledger from preparation to outcome.

revenue_loop_diagnosis

Find the first failing Figure-Eight stage and the first failing coaching layer for the highest-priority manager-scoped rep, with persisted evidence and safe next-inspection functions. The two lanes are reported separately — no invented cross-lane score. Requires team or all-metrics application permission. In assistants that support MCP Apps, this opens the interactive Figure-Eight Revenue Constraint view.

revenue_engine_state

Get the exact persona-scoped overlay used on the Revenue Engine canvas: which stages are at risk, recovering, or building, plus the Signal Layer state for each stage.

list_interventions

List a bounded page of Intervention Ledger records you are allowed to see, optionally filtered by entity, status, or outcome label. Pending, unknown, missing baseline or window, did-not-work, and worked states are all kept as they are — no outcome evidence is created.

get_intervention_outcome

Get one visible Intervention Ledger record with its explicit baseline, outcome window, and correlational outcome evidence. An ID that does not exist, belongs to another organisation, or is outside your access all return the same result.

get_revenue_loop_digest

Summarise the outcomes of visible interventions created within a bounded window (90 days by default). Pending and unknown outcomes are reported separately and excluded from the changed-after rate, and observed movement is described as correlational rather than causal.

prepare_intervention

Validate and fill out an intervention draft without writing anything. Resolves the baseline evidence and proposed owner so you can review the draft before an approval-gated create_intervention.

create_intervention (write · approval)

Create a tracked Intervention Ledger record for an entity and signal. The first call returns an approval request; no record is created until you approve it on the Summit53 review page and the assistant repeats the call with the approval receipt.

update_intervention_status (write · approval)

Change the status, due date, or completion of one visible Intervention Ledger record, after you approve the change on the Summit53 review page.

Notes & Activities

Opportunity-level notes, activities, and AI-generated summaries.

get_opportunity_notes_summary

Get the AI-generated notes summary for a specific opportunity or deal. Returns a synthesised overview of all logged notes, highlighting key themes, decisions, and next steps.

get_opportunity_activities

List all activities (calls, emails, meetings, notes) for a specific opportunity or deal. Optionally filter by activity type to focus on a specific channel.

get_opportunity_notes

List all notes for a specific opportunity or deal. Returns the raw note entries with timestamps, authors, and content.

create_opportunity_note (write · approval)

Add one note (a summary and optional details) to an opportunity or deal, after you approve it on the Summit53 review page. Nothing is sent or published, and a note added this way does not start any background processing — the AI notes summary is not refreshed by this call.

Account Intelligence

Account-level intelligence, execution health, and qualification framework scores.

account_intelligence_summary

Dashboard-level metrics including health scores, at-risk accounts, and expansion candidates. Provides a bird’s-eye view of your entire account portfolio. Supports filtering by industry, territory, owner, and revenue range.

account_execution_health

Execution health score and deal-level health breakdown for a specific account. Shows how well the sales team is executing on each opportunity within the account.

opportunity_framework_summary

Sales qualification framework scores for a specific opportunity. Supports MEDDPIC, BANT, SPICED, and THREE_WHYS frameworks. Returns component scores, overall score, and gaps to address.

Customer Success

Renewal exposure and account-health movement for customer success and retention work. Both functions are manager-team scoped, read-only, and require team or all-metrics application permission.

renewal_forecast_summary

Summarise active primary contract ARR renewing in the next 30 to 180 days, including at-risk causes and existing renewal-opportunity commit and best-case forecasts. Accepts a window_days parameter (default 90).

account_health_movement

Compare persisted current and previous account-health snapshots, identify manager-visible movers, and explain the existing evidence behind each delta. Reports exact snapshot ages and insufficient history explicitly — it never creates a new score or infers a prior one. Accepts optional account_id and limit (default 5) parameters.

Research

Queue fresh external account research for meeting, opportunity, and target-account preparation.

external_account_research

Queue external web and news research for a CRM account (by account_id or account_name) or for an outreach target account (by target_account_id). Results arrive asynchronously; for a target account, the latest results appear as external research on get_target_account. Use this to gather fresh intelligence before important meetings.

Outreach

Search and update outreach leads, draft sequences, and run account-centric outreach with target accounts and buying committees. External assistants can prepare and, with your approval, apply CRM-style changes; they cannot publish sequences, enrol leads, process due steps, or contact anyone.

search_outreach_leads

Search and filter outreach leads by query, status, owner, lead list, stage, tier, fit rating, touchpoint stage, LinkedIn connection status, next action, engagement signal, and target account (a specific target account, or whether a lead has one at all). Each result includes its target-account context — committee role, angle, other contacts, and coordination. Supports pagination for large result sets.

get_outreach_lead

Get full details for a specific outreach lead including list memberships, sequence enrolments, recent activity, and its target-account context: committee role, angle, thread order, other contacts at the account, and coordination. Coordination is a recommendation: if another contact at the same account was touched recently, Summit53 suggests waiting until the next allowed touch date, and if an earlier thread has not been opened yet it suggests opening that first. You can still choose to send anyway.

update_outreach_lead (write · approval)

Update fields on one existing outreach lead after you approve the change. Only the fields you provide are changed. It can also attach the lead to a target account or detach it, and set its buying-committee role, angle, thread order, and the lead who referred them. For two or more leads, use batch_update_outreach_leads so every change is approved on one review page.

batch_update_outreach_leads (write · batch approval)

Stage several outreach lead updates (up to 25 by default, one item per lead) for a single review page. You approve or reject each item, then the assistant repeats the call with the approval receipt and every item is reported as applied, rejected, conflict, or failed. Use it for bulk CRM hygiene (notes, stage, tier, next action) and for mapping a buying committee across several leads. It never contacts or enrols a lead.

list_outreach_lead_lists

List outreach lead lists with lead counts. Optionally filter by name or owner.

list_outreach_lead_events

List outreach lead activity events or replies across the outreach book. Filter by lead, sequence, event type, or time window.

outreach_funnel_summary

Funnel summary of outreach leads with breakdowns by status, stage, fit rating, tier, engagement metrics, enrolment counts, and conversions. The bird’s-eye view of outreach pipeline health.

list_outreach_sequences

List outreach sequences with published and draft version metadata plus enrolment counts. See which sequences are active and how many leads are enrolled.

get_outreach_sequence

Get full details for a specific outreach sequence including individual steps, timing, and version history.

create_outreach_sequence (write · approval)

Create a new draft outreach sequence with ordered email, manual, and wait steps, after you approve it. Sequences created by an external assistant are always drafts: they cannot be published, send anything, or enrol leads. Publish the sequence in Summit53 when it is ready.

update_outreach_sequence (write · approval)

Update a draft sequence’s name, description, or steps after you approve the change. Only the fields you provide are changed. External updates only apply to draft sequences, and publishing remains blocked.

list_outreach_starter_templates

List the predefined outreach starter templates available for assistants and operators. Use these as a starting point when creating new sequences.

search_target_accounts

Search outreach target accounts for account-centric outreach, filtered by query, stage, tier, engagement model, or owner. Each result shows buying-committee coverage and gaps, engaged threads, and coordination state, including the next allowed touch date. Single-threaded accounts are worked through one contact by design and report no committee gaps.

get_target_account

Get one target account: its thesis, buying committee (each lead’s role, angle, thread order, referral, and last touch), coverage by role (missing, mapped, or engaged), coordination state, the latest external research, and the opportunity once converted. Check coordination here before drafting or scheduling a touch to anyone at the account.

propose_target_accounts

Read-only. Group outreach leads that are not yet attached to a target account by company (email domain, website, or company name) into proposed target accounts, with note excerpts and title-based role hints. Nothing is created — confirm each proposal with create_target_account.

create_target_account (write · approval)

Create a target account (name, website, stage, tier, fit, engagement model, thesis, and minimum gap between touches) and attach unattached leads to it in the same approved action. Committee roles are then assigned with batch_update_outreach_leads. Never contacts anyone.

update_target_account (write · approval)

Update a target account’s name, website, stage, tier, fit rating, engagement model, owner, minimum touch gap, or thesis (the problem hypothesis), after you approve the change. Only the fields you provide change. Committee membership is changed on the leads with update_outreach_lead or batch_update_outreach_leads.

convert_target_account (write · approval)

Convert a target account into one CRM opportunity that carries its whole buying committee. Links (or creates) the CRM account, converts every committee lead to a contact, gives each contact its committee role on the opportunity, and seeds the opportunity’s stakeholder intelligence. Link an existing opportunity instead of creating one by passing its ID. Requires your approval and never contacts anyone.

get_stakeholder_research_brief

The starting point for researching a target account’s buying committee on the web. Returns the account’s identity and thesis, your organisation’s ICP and positioning context, missing roles, the current committee, saved suggestions (including people not to resubmit), existing leads and contacts at the company, and the evidence rules that submit_stakeholder_suggestions enforces.

submit_stakeholder_suggestions (write · approval)

Save up to six researched people at a target account as pending stakeholder suggestions for you to review. Each needs at least one piece of evidence with a source URL and a verbatim excerpt naming the person and the company; emails and phone numbers are never guessed. Suggested roles are hypotheses (economic buyer, technical buyer, user, or influencer), and suggestions only join the committee once accepted. Never contacts anyone.

accept_stakeholder_suggestion (write · approval)

Add one pending stakeholder suggestion to the target account’s buying committee, creating or reusing a lead with the committee role, angle, and thread order you choose. If a possible duplicate lead exists, the assistant asks you whether to reuse it or treat the person as distinct. A champion or blocker role needs you to confirm the evidence first. Never contacts anyone.

dismiss_stakeholder_suggestion (write · approval)

Dismiss a pending stakeholder suggestion so it is not suggested again for that target account.

target_account_funnel_summary

The account-level outreach funnel: which target accounts have stalled and the next move for each, which multithreaded accounts to widen next, and this week’s touches against the team’s target, plus accounts by stage, engaged threads, and committee coverage.

icp_fit_evidence

Is the ICP working? Engagement, conversion, and win rates by lead fit tier, and win rates by account ICP fit, each with a sample-size flag so a thin bucket is never read as a real rate, plus how much fit scoring used the currently published ICP. The results are correlational: fit tiers that win more don’t prove the ICP caused it.

Value Impact

Read the journey from promised value to proven impact: contracts, promised outcomes, post-sale evidence, fulfilment scores, and expansion readiness. Changes to contracts and impact evidence are made in Summit53 itself — those functions are not available to external assistants.

get_value_impact_card

Get the Value→Impact diagnostic card for an account: fulfilment score, category, expansion readiness, evidence confidence, and recommended actions. The single-screen view of whether you’re delivering on your promises.

get_value_contract

Get the value contract with promised outcomes for a specific opportunity. Shows what was committed to the customer at close — the baseline against which impact is measured.

list_impact_entries

List post-sale impact entries (evidence) for a specific contract. Each entry represents proof that value is being delivered — metrics improvements, customer feedback, usage milestones, etc.

list_account_contracts

List all contracts for an account with dates, ARR, status, and primary flag. Useful for understanding the full commercial relationship.

Coaching

Read stored diagnostics, agendas, development history, and meeting quality for the reps you are authorised to coach, then record completed sessions and later outcomes with your approval.

coaching_diagnostic

Read the latest stored performance diagnostic for an authorised rep or your coaching roster. Shows the diagnosed coaching layer (Results / Effort / Knowledge / Skills), the single focus metric with the highest improvement leverage, its metric-selection score, and the gap percentage against the team P75 benchmark. The score only chooses between one rep’s own metrics — it must not be used to rank reps against each other.

render_team_framework_heatmap

Show the latest stored framework-component scores for your authorised team as an interactive, read-only heatmap. Missing component evidence stays unknown (never zero) and is excluded from averages. In assistants that support MCP Apps it opens the Team Framework Heatmap; it also returns an image of the heatmap and the underlying data.

coaching_agenda

Read a stored, pre-populated 1:1 coaching agenda for an authorised rep, including the cadence manager. Returns the week type (Strategic / Skills / Career / Execution / Close), recommended coaching model, diagnostic summary, agenda sections with data-driven discussion prompts, and previous action items. This read never generates an agenda and never counts as evidence that coaching took place.

coaching_development

Read an authorised rep’s stored development tracker: 13 weeks of diagnostic history, a 26-week focus-metric trend, coaching-model usage, the current quarter, and development goals with progress.

coaching_history

Read a rep’s durable coaching timeline, keeping recommendations, plans, schedules, coaching received, acknowledgements, and outcomes distinct. Intervention Ledger assignments are returned separately and never count as coaching received.

coaching_meeting_analytics

Read roster-scoped meeting-quality metrics for a rep or team: talk ratio, question rate, interactivity, monologue length, interruptions, trends, and flagged coaching signals.

coaching_intervention_plan

Read the leader-only coaching priority queue, which joins stored rep diagnostics to at-risk deals, with intersection counts and recommended deal actions. Optionally pick the week.

rep_coaching_context

Get bounded coaching context for one authorised rep in a single call, choosing from the stored diagnostic, development, coaching history, meeting analytics, and agenda sections. Coaching history follows the same access rules as coaching_history; it returns 20 records by default and up to 50.

complete_coaching_session (write · approval)

Complete a coaching session exactly once, after you approve it, with a manager attestation, when the session occurred and was completed, notes, action items, and optional evidence, provenance, and linked records (development goal, intervention, meeting, opportunity, or account). Creates an auditable lifecycle event. Missing evidence stays unverified, and a plan or agenda is never treated as proof that coaching was delivered.

record_coaching_outcome (write · approval)

Record a later, coaching-specific outcome assessment for a completed session, after you approve it. Unknown evidence is kept as unknown, and the assessment is stored as a separate audit event.

Meetings

Read captured meetings and their intelligence. These functions always require the meeting:read scope.

meeting_capture_history

List the Meeting Capture sessions you are authorised to see, with the same filters as the Meeting Capture page (search, status, opportunity, and date range), pagination, processing status, linked opportunity or meeting, summary counts, and topic tags. The team view is leader-only and limited to your roster.

meeting_context

Read bounded context for one authorised meeting capture: processing state, summary, structured insights, conversation analytics, speaker and timeline evidence, sentiment, topics, and entities. Transcript text is only included when requested and is capped in length; audio URLs are never returned.

Knowledge Base

Search the organisation’s Knowledge Base — positioning, ICP, brand voice, product, battlecards, case studies, methodology, and reactive plays.

search_knowledge

Search the Knowledge Base across positioning, ICP, brand voice, product, battlecards, case studies, methodology, and snippets. Optionally restrict to specific asset types. Only published assets in your organisation are returned, ranked by relevance.

get_knowledge_asset

Fetch a single Knowledge Base asset by ID, including its full markdown body, tags, and structured payload.

list_methodology_playbooks

List the org’s published methodology assets (e.g. MEDDPICC, BANT, SPICED definitions). Each asset carries framework key and component metadata when present.

get_battlecard

Find the best-matching competitive battlecard in the Knowledge Base by competitor name. Returns up to three ranked candidates across battlecard title, summary, and body.

list_reactive_plays

List the org’s published reactive plays. Optionally filter by trigger type from the asset’s structured payload.

Job Lifecycle

Inspect and manage durable work returned by long-running Revenue Engine functions. These functions always require their scope: jobs:read to inspect, jobs:manage to cancel. Job results and errors are retained for 30 days by default.

get_job

Get the status, progress, result, error, timestamps, and expiry for one job you can access. Results and errors are kept for the configured retention window (30 days by default).

list_jobs

List the jobs you can access, newest first, with pagination (page, page_size) and an optional status filter. Members see their own jobs; organisation owners and admins see all of the organisation’s jobs.

cancel_job (write)

Request cancellation of a pending job you can access. Cancellation is only confirmed when the queued work is removed or skipped before it runs. This job-management action requires jobs:manage, needs no approval, and does not bypass job ownership or organisation boundaries.

Discovery

Confirm who and where you are connected, and discover the function surface available to the current session.

current_context

Confirm the connected environment, organisation, user, role, and how the organisation membership was selected. Read-only, and never returns credentials or token material.

list_available_tools

List only the functions you can actually run in this session, with their categories, access classification, required scope, whether approval is required, and behaviour annotations. Approval tiers are applied, so approval-gated writes only appear when approvals are enabled for your connection. Optionally filter by category. Use this instead of assuming every client or user sees the same catalogue.

Prompts, Resources and Apps

Alongside functions, the server publishes standard MCP prompts, read-only resources, and interactive MCP Apps. All of them are read-only: none creates or changes a CRM record, message, intervention, job, or approval.

Prompts

• Prepare Deal Review (deal_review_preparation) — an evidence-backed review of one opportunity. Takes opportunity_id.
• Prepare Weekly Revenue Engine Brief (weekly_revenue_engine_brief) — a weekly brief across the Revenue Engine. Takes lookback_days for the intervention digest, from 30 to 365 (default 30).
• Diagnose Forecast Gap (forecast_gap_diagnosis) — diagnoses the gap for the current single period. Takes period_type: month, quarter, or year (default quarter).

Resources

Each resource returns JSON:

• summit53://engine/state — the current Revenue Loop engine-state diagnosis for your organisation.
• summit53://revenue-engine/state — the current persona-based Revenue Engine stage overlay.
• summit53://revenue-engine/readiness — deal-history counts and Revenue Engine machine-learning readiness flags.
• summit53://interventions/{intervention_id} — the stored outcome and evidence for one intervention you can see.

MCP Apps

• Figure-Eight Revenue Constraint (ui://summit53/figure-eight/v1) — opened by revenue_loop_diagnosis. A read-only, interactive view of the manager-scoped Revenue Loop diagnosis.
• Team Framework Heatmap (ui://summit53/coaching-framework-heatmap/v2) — opened by render_team_framework_heatmap. A read-only, interactive heatmap of your authorised roster’s framework-component scores, also returned as an image.

Assistants without MCP Apps support receive the same results as text and structured data instead.

Quick Reference Table

The 102 functions that can be offered to external assistants are listed below. Access describes behaviour, not session availability: Write · approval functions only appear on trusted connections with action approvals enabled, and Write · batch approval also needs batch approvals enabled. cancel_job is a scoped job-management write. Summit53 application permissions remain authoritative.

Supported Clients

Summit53 uses standard MCP transport and can be configured in compatible clients. Exact remote-server, custom-app, MCP Apps, and approval capabilities vary by client plan and workspace:

• Claude.ai, Claude Desktop and Cowork — via the Summit53 plugin or a custom connector in Settings → Connectors
• Claude Code — via the Summit53 plugin or claude mcp add --transport http
• Codex — via the Summit53 plugin or codex mcp add --url
• ChatGPT — via Apps connector settings
• Cursor, Windsurf and other remote-capable clients — by adding the server URL directly; stdio-only clients can use the mcp-remote bridge

These standard connections give you every read function your scopes and permissions allow. Approval-gated writes are only offered on a trusted connection that Summit53 sets up for your organisation; they never appear on the standard, dynamically registered /mcp connection.

Access Model

Interactive clients use OAuth with PKCE. Access tokens last one hour and rotating refresh tokens last seven days; approved service integrations may use scoped static API keys. Summit53 organisation, role, and ownership permissions remain authoritative for every function.

• Read functions never need approval. Job Lifecycle and Meetings functions always enforce their scope (jobs:read / jobs:manage and meeting:read); other read scopes are enforced only when the server’s read-scope enforcement is switched on.
• Approved writes (write · approval, 14 functions) require their write scope and only appear when action approvals are enabled and the assistant connects through a trusted connection set up by Summit53. The first call returns an approval request with a review link to a Summit53 page; you, as the requesting user, approve or deny the exact change there; the assistant then repeats the identical call with the approval receipt. Nothing changes until then, and a single approval expires after 10 minutes by default.
• Batch approval (write · batch approval): batch_update_outreach_leads stages up to 25 outreach lead updates (by default) on one review page, where each item is approved or rejected. It also requires batch approvals to be enabled.
• Job management: cancel_job is a write that requires jobs:manage and needs no approval.

Scope authorisation is not the same as human approval. See Action Approvals in the MCP Setup Guide for the step-by-step flow.

Not available to external assistants

The following 14 write functions are registered on the server but held back from external clients. They are never listed, and calling one fails with “This external action is not enabled for the current approval risk tier”. Use Summit53 directly for these actions:

• rebuild_opportunity_notes_summary
• publish_outreach_sequence
• enroll_outreach_leads_in_sequence
• process_outreach_due_steps
• resolve_outreach_manual_step
• convert_outreach_lead
• bulk_upsert_outreach_leads
• create_impact_entry
• update_impact_entry
• delete_impact_entry
• update_account_contract
• set_primary_contract
• generate_value_contract
• run_value_impact_pipeline

Related Documentation

• MCP Setup Guide — Step-by-step connection instructions for each client
• MCP User Guide — Daily workflows and practical usage patterns
• Integrations Overview — Connect Salesforce, Google Workspace, Slack, and more
• Apply to the Founding 50 — Help shape the Revenue Engine in production