Sellfire MCP — Setup & Capabilities Guide

What you get

Coaching, in conversation

The Sellfire MCP server connects Claude to your Sellfire workspace, so the work that used to mean digging through dashboards and call recordings now happens in plain conversation. Ask a question — get the answer, with the calls, metrics, and context already pulled together.

Coach faster

Get a complete coaching read on any rep from a single question — performance against your top performer, the calls that prove the point, and a ready-to-share action plan.

Never miss a struggling rep

Ask for a team health check and Claude surfaces the two or three people who need your attention this week — with the specific metric gap behind each.

Understand any call in seconds

Pull a transcript, a stage-by-stage breakdown, and the AI summary on demand — no scrubbing through recordings to find the moment that mattered.

Onboard clients in minutes

Turn a new client's intake answers into a personalized, ready-to-run cold-call script built on Sellfire's proven template.

Read-only by default. The actions that change your data are adding coaching notes, creating or updating Impact Plans, and — for admins only — creating sales and updating sale line items — and Claude always confirms with you first.

Built for the people who coach and run the floor. This integration is designed for Sales Managers and Admins. Individual reps do their work in the Sellfire app and on the phones — they don't need a Claude connection.

Find your path

Choose your track

What you can do depends on your Sellfire role. Pick the track that fits you — each one lists the tools, playbooks, and everyday workflows you'll reach for most.

Sales Manager

You coach a team of SDRs and AEs.

Calls, transcripts & coaching notes
Rep & team performance
Impact Plans, leads & sales
Org-wide roster, raw metric exports, tiers & trends
All four playbooks and the full reference shelf

Jump to your track →

Admin

You run sales ops across the whole org.

Everything a Sales Manager gets
Sale writes — record & adjust deals
Lead writes — create, update & delete

Jump to your track →

Get connected

Setup & connect flow

Connecting takes about two minutes. You'll add Sellfire as a custom connector in Claude, sign in with your Sellfire account, and approve access. Follow the steps below.

Add Sellfire to Claude One click on Claude web — opens the connector dialog with the Sellfire server URL pre-filled. Prefer to do it by hand? Follow the steps below.

Open the connector settings in Claude

In Claude — on web (claude.ai), Desktop, or Claude Code — go to Settings → Connectors and choose Add custom connector.
Enter the Sellfire server URL

Paste the remote server URL exactly as shown and save:

https://mcp.sellfire-edge.com/mcp
Sign in to Sellfire

Claude opens Sellfire's OAuth flow. Sign in with your Sellfire account — or, if you're already signed in to the Sellfire web app, simply approve the connection.
Approve access

Claude connects securely and acts on your behalf with your permissions — nothing more. The tools your role unlocks appear in Claude right away.
Start asking

You're connected. Try "Show me last week's calls for my team" or "Run a team health check" to confirm everything is working.

Connection details

Server URL

https://mcp.sellfire-edge.com/mcp

Transport

Streamable HTTP (remote, hosted on Cloudflare)

Authentication

OAuth 2.0 — you authenticate as your Sellfire identity

Permissions

The connector acts on your behalf, scoped to your role

Data safety

Read-only by default; writes always require your confirmation

Tool not showing up? If a tool you expected isn't there, your Sellfire role probably doesn't include it — Sales Managers and Admins get the coaching toolkit. Check your role with your Sellfire admin. If your login spans several organizations, a tool appears when any of them allows it; running it against an org where you lack access is refused with a message naming the orgs where it works, so pass that organization instead.

Track · Sales Manager

If you're a Sales Manager

You coach a team of SDRs and AEs. Everything here is built to help you find the moments worth coaching, understand them fast, and turn them into plans your reps can act on.

Your toolkit

Calls, transcripts, AI summaries, and the named stages of any recorded call
Timestamped coaching notes you can add right on a call
Rep and team performance, measured against your top performer or team average
Metric drill-downs — the actual calls behind any number
Impact Plans you can create, update, and track over time
AI coaching suggestions tuned to the metric you want to move
Leads, their full activity timelines, and closed deals
Org-wide analysis — the full user roster, raw metric exports, tier history, and time-series trends (the analysis workflows in the Admin track below work for you too)

Your playbooks

Pre-packaged workflows you launch as slash commands in Claude Code:

/mcp__sellfire__coaching-report /mcp__sellfire__team-health-check /mcp__sellfire__investigate-call /mcp__sellfire__generate-cold-call-script

Your reference shelf

Coaching knowledge Claude can pull into any conversation — just ask for it by name:

Metric definitions and the Sellfire roles reference
The cold-call and demo-call 5-stage coaching guides
The Sellfire cold-call script template
The Sales Lab Scripting evaluation framework

Day-to-day workflows

The most common things coaches do with Claude. Each combines a few tools in sequence — just describe the goal and Claude runs the steps.

Coach a specific rep

When you want a complete coaching read on one rep and an action plan to share.

Find the rep by name and pull their performance snapshot (last week or month) vs. the top performer.
Ask for illustrative calls where they struggled or excelled on your focus metric.
Ask Claude for an AI coaching suggestion tuned to that metric.
Create an Impact Plan with action items and the calls attached as examples.

Review team health

For your regular review of who needs attention.

Ask for the team's performance for the last week or month.
Have Claude flag outliers — Tier 3 reps, the largest negative gaps, or reps in their first 90 days.
Drill into the specific calls behind a metric for anyone who stands out.

Investigate a single call

When something happened on a call and you want to understand it before talking to the rep.

Pull the call by ID or by searching (rep + date + outcome).
Walk through the transcript, stage timeline, and AI summary.
Review existing coaching notes.
Add a new coaching note at the timestamp where you spotted the issue.

Track a lead's journey

When you want to know everything that's happened with a prospect.

Pull the lead's profile.
List the full activity timeline — every call, email, SMS, disposition, and status change in order.
Drill into any individual touchpoint for full context.

Review closed deals & generate scripts

For deal-flow reviews — and for spinning up cold-call scripts during onboarding.

List sales for a rep, lead, or date range, then pull deal-level detail (items, terms, prepayment, CRM links).
To generate a script: paste a client's Q1–Q10 intake answers and ask Claude to build one from Sellfire's template.

Try saying…

"Give me a coaching report on Marcus for last month."
"Run a team health check for last week — who needs my attention?"
"Pull Dana's three lowest set-rate cold calls and walk me through what went wrong."
"Create an Impact Plan for Priya focused on hold rate, with those two no-show calls attached."
"Show me everything that's happened with the Acme Dental lead, in order."

Track · Admin

If you're an Admin

You can do everything a Sales Manager can — all the calls, coaching, Impact Plans, org-wide analysis, and playbooks above — plus the write actions reserved for admins.

Your extra toolkit

Sale writes (admin-only): create a sale and update a sale's line items
Lead writes (admin-only): create one or many leads in a single call (each with its own details, up to 25), update the details of one or many leads in a single call (each with its own changes, up to 25), and delete a lead (soft delete — recycle bin, permanent after ~15 days)
Lead lifecycle (admin-only): assign a lead to a rep, release one lead off a rep (records the release, so that rep won't be handed it again for ~60 days), bulk-unassign up to 200 leads back to the pool, and permanently retire up to 25 leads in one call. Just say what you want to happen — Claude picks the tool and never fabricates a call disposition to do it.

Day-to-day workflows

On top of every Sales Manager workflow, these help you manage the whole org. The analysis workflows here (roster, trends, tiers) are available to Sales Managers too — only the write workflows are admin-only.

Audit the roster

When you need a clean list of who's actually on the floor.

Pull the full user roster with role, team, and status.
Spot Sellfire support accounts (internal staff, not your reps) so they don't skew rankings.
Hand specific reps off to their Sales Managers for coaching.

Spot trends over time

For seeing where a rep or the floor is heading, not just where they are today.

Choose the metrics, reps, and time frame you care about.
Get time-series trend data — one point per week or month.
Compare trajectories to find who's accelerating and who's slipping.

Track tier movement

For quarterly reviews and promotion decisions.

Pull tier history for your reps, month by month.
See who moved up or down between Tier 1, 2, and 3.
Pair it with a team health check to explain the movement.

Record & adjust a sale WRITE

Admin-only — for logging a closed deal or correcting a line item without leaving the conversation.

Create a sale: give Claude the lead, the closing rep's email, the amount, the channel, and who was spoken to (plus any line items).
Adjust a line item later by product ID — change quantity, price override, prepayment, or apply a concession.
Claude confirms the full payload with you before writing either one.

Export the raw numbers

For reporting and data work outside the coaching frame.

Pull raw performance aggregation records for any set of reps and periods.
Resolve names and teams from the roster.
Hand the numbers off to your own reporting.

Try saying…

"List all active reps grouped by team, and flag any Sellfire support accounts."
"Chart close rate by rep for the last three months, weekly."
"Which reps moved up or down a tier over the last quarter?"
"Run a team health check across the whole org — which teams are trending down?"

Claude Code only

Built-in playbooks

These are pre-packaged workflows the server ships with — think of them as MCP-hosted skills. In Claude Code, invoke them as slash commands using the pattern /mcp__sellfire__<prompt-name>.

/mcp__sellfire__coaching-report /mcp__sellfire__team-health-check /mcp__sellfire__investigate-call /mcp__sellfire__generate-cold-call-script

Playbook	You provide → you get
`coaching-report`	A full coaching report for one rep. Provide: rep ID, optional period (`week`/`month`). Get: metric gaps, current & historical tier, AI suggestion, recommended action items.
`team-health-check`	A team-wide pulse check. Provide: optional team ID & period. Get: a short list of 2–3 reps to focus on with the specific metric gaps driving each.
`investigate-call`	A deep-dive review of one call. Provide: the call ID. Get: metadata, transcript, stage timeline, AI summary, and existing notes — top to bottom.
`generate-cold-call-script`	A personalized script. Provide: the client's Q1–Q10 intake answers. Get: a full script tailored to their industry, DM, competitors, and value props.

Claude Code only

Built-in coaching knowledge

Claude can pull these reference materials into the conversation on demand. You don't need to memorize their names — just ask for "the Sellfire metric definitions" or "the cold call stages guide" and Claude loads them.

Resource	What it covers
Roles reference	Every Sellfire role — what each does, which metrics to coach against, and how to recognize internal support accounts.
Metric definitions	The full formula and value type for every metric Sellfire tracks — win, DM connect/set rate, hold rate, close rate, revenue, dials, plus supporting activity metrics.
Cold call 5-stage guide	Theory and target KPI for each stage: Gatekeeper Intro, DM Intro, 30-Second Pitch, Setting the Demo, Confirming the Demo.
Demo call 5-stage guide	Theory for each demo stage: Cover/Agenda/Needs Analysis, Credibility, Product, Timeline/Trial Close, Pricing/Transition to Close.
Cold-call script template	The verbatim Sellfire cold-call script — all eight sections including objection rebuttals — used as the base for personalized scripts.
Sales Lab Scripting framework	The evaluation framework from Sales Lab Scripting (Hodson & Mahar): three pillars, the two-call structure, prospect types A/B/F, and a scoring rubric.

Compatibility

What works in which client

The server offers three kinds of capability — tools, resources (the reference shelf), and prompts (the playbooks). Which ones you can use depends on the Claude client you connect from.

Capability	Claude chat (claude.ai, Desktop)	Claude Code
Tools	Yes	Yes
Resources (reference shelf)	Not supported	Yes
Prompts (playbooks)	Not supported	Yes — slash commands `/mcp__sellfire__<name>`

In the Claude chat interface you work entirely through tools — just describe what you want and Claude calls the right one. In Claude Code you additionally get the coaching reference shelf and the built-in playbooks.

Playbooks behave like skills, but they're served by the MCP server and are reachable in Claude Code via slash commands of the form /mcp__sellfire__<name> — for example /mcp__sellfire__generate-cold-call-script.

Reference

Tool reference

Every individual capability the server exposes, grouped by what it helps you do. Tools marked WRITE change data; all others are read-only.

Account

Tool	What it does	Who
`list_accessible_orgs`	See every org your login can access, your role in each, and the exact value to pass as the organization parameter when working across orgs.	Everyone

Calls & transcripts

Tool	What it does	Who
`list_calls`	Search and filter calls by rep, team, date, outcome, who was reached, channel, or duration.	Sales Manager · Admin
`get_call`	Core metadata for one or several calls: lead, rep, disposition, duration, recording link.	Sales Manager · Admin
`get_call_transcript`	Full transcript with speaker names and timestamps for a call (or several at once).	Sales Manager · Admin
`get_call_summary`	Sellfire's AI-generated summary points for a call (or several at once).	Sales Manager · Admin
`get_call_segments`	The named timeline stages of a call with start/end timestamps.	Sales Manager · Admin
`get_call_theory_analysis`	Pre-computed pass/fail evaluations against Sellfire's methodology, with coaching notes and transcript pointers.	Sales Manager · Admin
`get_call_notes`	Lists existing coaching notes on a call with timestamps.	Sales Manager · Admin
`add_coaching_note` WRITE	Add a new timestamped coaching note onto a call, or update an existing one.	Sales Manager · Admin

Performance & dashboards

Tool	What it does	Who
`get_rep_performance`	One rep's metric snapshot (week/month) vs. team average, top performer, or prior period.	Sales Manager · Admin
`get_team_performance`	Side-by-side metrics for every rep on a team or org — current week/month or a custom date range — with tier info and optional top-performer gaps.	Sales Manager · Admin
`get_dashboard_calls`	The actual calls that fed into a specific metric — the dashboard drill-down.	Sales Manager · Admin

Impact Plans (coaching plans)

Tool	What it does	Who
`list_impact_plans`	List existing plans, filtered by rep, team, focus metric, or creation date.	Sales Manager · Admin
`get_impact_plan`	Full detail on one plan — focus, action items, attached calls, notes, completion status.	Sales Manager · Admin
`create_impact_plan` WRITE	Create a new plan for a rep. Claude confirms before saving.	Sales Manager · Admin
`update_impact_plan` WRITE	Update an existing plan — focus, action items, attached calls, notes. Confirms before saving.	Sales Manager · Admin
`get_coaching_suggestion`	AI-generated coaching narrative for a rep, focused on a specific metric.	Sales Manager · Admin

Leads & activities

Tool	What it does	Who
`get_lead`	Full profiles for one or more leads (up to 25 per call) — business name, contact, industry, source, phase, owning rep, lifecycle dates.	Sales Manager · Admin
`search_leads`	Find leads by attributes (source, industry, location, channel, phase, ownership, owning rep, brand), free-text on business name / email, or created/updated date ranges. Searching by phone number matches every phone slot — primary plus the mobile / home / alternate phone book — so it can confirm a number is free before you create a lead.	Sales Manager · Admin
`list_lead_activities`	The activity timeline for one or more leads, in reverse chronological order. Can export the combined timeline of up to 10 leads as one CSV.	Sales Manager · Admin
`get_activity`	Full context for one or more activities (up to 25 per call) — channel, disposition, notes, email/SMS body, attached call.	Sales Manager · Admin
`add_lead_activity` WRITE	Log a new touchpoint on a lead — call (with optional disposition), email, SMS, or other contact. Claude confirms before saving.	Sales Manager · Admin
`create_lead` WRITE	Create one or many leads in a single call (up to 25), each with its own details — business name and phone number required per lead; optional contact, location, industry, source, and rep assignment. Assigning a rep also enqueues that lead into the rep's dialer. Duplicate phone numbers are rejected. An optional note rides the same request and is saved as a lead activity (shows up in the lead's Notes tab — the lead itself has no notes field). Each lead is written independently, so Claude reports what was created and what failed. Claude confirms first.	Admin
`update_lead` WRITE	Update one or many leads (up to 25 per call), each with its own field changes. Only the fields you name change; fields can't be cleared to empty. Each lead is written independently, so Claude reports what was updated and what failed. Claude confirms the exact diff for every lead first.	Admin
`delete_lead` WRITE	Delete a lead — soft delete to the recycle bin, gone from Sellfire immediately and permanently erased after ~15 days. Claude confirms the lead's identity first.	Admin
`assign_lead` WRITE	Assign one or many existing leads (up to 25 per call) to a single rep with full in-app mechanics — assignment history, claim date, and a dialer-queue entry so each lead is dialable today. Leads a rep already owns need an explicit override; an optional inbound-activity flag ranks a lead that replied ahead of ordinary cold calls in the rep's queue. Claude confirms first.	Admin
`unassign_leads` WRITE	Release up to 200 leads back to the unassigned pool — not retirement: no disposition is written and the leads stay alive for future assignment. The sales cycle is not reset unless you explicitly ask (reset_sales_cycle). Retired, already-unassigned, and future-booked leads are skipped with a reason. Claude confirms first.	Admin
`release_lead` WRITE	Release one lead from the rep who holds it, the way that rep would from their dashboard. Costs nothing in call stats. Unlike bulk unassign it records the release against that rep, enforcing your org's same-rep cooldown (~60 days by default) and feeding the Release metric. The lead stays alive and assignable. Claude confirms the lead and its current owner first.	Admin
`retire_lead` WRITE	Permanently retire one or many leads (up to 25 per call) — the terminal state, via the same machinery as the in-app admin bulk retire. Detaches the rep, exits any sequence, clears dialer entries. Fully metric-neutral: no activity, dial, or connect is recorded for anyone. Already-retired leads are skipped and anything the platform refuses is reported, rather than failing the whole batch. Leads with an upcoming booked demo/callback are refused unless you explicitly approve deleting the event. Claude confirms each lead's identity first.	Admin
`list_industries`	List the org's industry options — the exact, case-sensitive labels a lead's industry must match on create/update. Use it to find valid values.	Admin
`list_lead_fields`	List the org's lead custom-field definitions — the exact names, types, and allowed values the custom_fields map on create/update lead must match.	Admin
`create_industry` WRITE	Add a new industry option to the org so it becomes a valid lead industry. Duplicate labels are rejected. Claude confirms first.	Admin
`create_lead_field` WRITE	Define a new lead custom field so its name can be set on leads. Names can't collide with a built-in lead field, and Dropdown fields need their full option list up front — there's no edit or delete from here. Claude confirms first.	Admin
`list_lead_sources`	List the org's lead sources and show which are actually registered. Only registered ones can be used in rep assignment filters — sources that arrived on leads via the API, an integration, or a CSV column are listed separately with a lead count, and are invisible to those filters until registered. Unregistered values are capped at the busiest 500, with a has_more flag when the list is truncated.	Admin
`create_lead_source` WRITE	Register a lead source so it becomes selectable in lead filters and rep assignment. Doesn't change any lead — leads already carrying the name become filterable right away. Claude confirms first.	Admin

Nightly assignment (dial queue)

Tool	What it does	Who
`get_nightly_assignment_config`	Org-level nightly-assignment snapshot — default per-rep allotment, exclusion rules and their master switch, the assignment-segment catalog, and queue-ordering settings.	Admin
`list_rep_assignment_filters`	Every rep's nightly-assignment config — pause state, allotment override, segments-vs-inline mode, and the filters that decide which leads they're handed each night.	Admin
`update_rep_assignment_filters` WRITE	Update a rep's filters, caps, pause/retain flags, and segment membership; optional refill_now tops up their queue immediately. Claude confirms the diff first.	Admin
`update_engagement_rules` WRITE	Update the org's Rules of Engagement — resting windows, ownership rules, claim rules, the default nightly allotment, and the exclusion-rules master switch. Claude confirms first.	Admin
`create_assignment_segment` WRITE	Create a named, reusable nightly-assignment filter set assignable to multiple reps. Claude confirms first.	Admin
`update_assignment_segment` WRITE	Update a segment's name or filters (PATCH semantics). Claude confirms first.	Admin
`delete_assignment_segment` WRITE	Soft-delete a segment; reps whose only segment it was fall back to their inline filters. Claude confirms first.	Admin
`set_segment_reps` WRITE	Replace the set of reps assigned to a segment. Claude confirms the rep list first.	Admin
`list_rule_condition_fields`	The valid fields, operators, and option values an exclusion-rule condition can use — always consulted before writing rules.	Admin
`create_exclusion_rule` WRITE	Create an org-wide rule excluding matching leads from the nightly run. Conditions are validated against the field catalogue. Claude confirms first.	Admin
`update_exclusion_rule` WRITE	Rename, toggle, reprioritize, or replace the conditions of an exclusion rule. Claude confirms first.	Admin
`delete_exclusion_rule` WRITE	Soft-delete an exclusion rule; it stops applying at the next nightly run. Claude confirms first.	Admin

Sequences

Tool	What it does	Who
`list_sequences`	List the org's sequences with status (active / paused / deleted), entry criteria, per-step tasks and delays in plain English, and a one-line behavior summary. include_deleted surfaces deleted sequences.	Admin
`get_lead_sequence_state`	One lead's live sequence enrollment — current step, hold state and release time, plus an anomalies list that names broken states verbatim (paused/deleted enrollment — the lead still advances, deleted/detached current step, orphaned holds).	Admin

Sales & analysis

Tool	What it does	Who
`list_sales`	List closed deals, filtered by closing rep or lead; sortable by created/updated date.	Sales Manager · Admin
`get_sale`	Full deal detail — items, amount, prepayment, contract length, terms, CRM cross-references.	Sales Manager · Admin
`create_sale` WRITE	Create a new sale — lead, closing rep's email, amount, channel, who was spoken to, plus optional line items and terms. Claude confirms first.	Admin
`update_sale_item` WRITE	Update one line item of a sale by product ID — quantity, price override, prepayment, or concession. Claude confirms first.	Admin
`list_users`	Full user roster — name, email, role, team, status.	Sales Manager · Admin
`list_brands`	List the org's brands (id + name) — the IDs to pass in the brand_ids filter on the performance, dashboard, and lead-search tools.	Sales Manager · Admin
`list_rep_metrics`	Raw performance aggregation records across periods — current week/month or a custom date range. Numbers without coaching framing.	Sales Manager · Admin
`get_performance_tiers`	Tier history for reps, with month-by-month detail.	Sales Manager · Admin
`get_data_explorer`	Time-series trend data for any metric by rep — choose granularity, time frame (including custom ranges), and metrics.	Sales Manager · Admin

AI Gym (Sellfire staff only)

Internal tools for the AI Gym — Sellfire's practice environment where staff run cold-call drills against an AI prospect and debrief with an AI coach. Available to any login with a @sellfire.com email, regardless of role.

Tool	What it does	Who
`list_gym_reps`	Roster of everyone who has used the AI Gym — one row per staff member with session counts, drills vs coach debriefs, pass rate, average adherence, and first/last activity.	Sellfire staff
`list_gym_sessions`	List a staff member's AI Gym practice sessions (drills and coach debriefs) with headline scores — adherence %, pass/fail, furthest script stage, disposition, duration.	Sellfire staff
`get_gym_scorecard`	Full scorecard for one gym session — per-dimension scores (tone, speed, script adherence, active listening), lesson outcome, per-stage criteria, vocal delivery, strengths, improvement areas.	Sellfire staff

Reading the answers

Understanding the metrics

A short glossary so you can read Claude's answers correctly. Claude can also pull in the full metric definitions resource on request.

The seven core coaching metrics

win

Appointments booked. A count, not a rate.

dm_connect_rate

Share of outbound effort that reached a decision maker. A rate 0–1.

dm_set_rate

Share of DM connects that resulted in a demo set. A rate 0–1.

hold_rate

Share of set demos that actually held (showed up). A rate 0–1.

close_rate

Share of held demos that converted to a sale. A rate 0–1.

revenue

Dollar value of closed deals in the period.

dials

Outbound dial volume. A count.

Tiers

Tier 1 — top third of the org
Tier 2 — middle third
Tier 3 — bottom third
Pre-90-days — too new to tier (less than 90 days of data)

How comparison gaps work

When Claude reports a "gap" or "percentage change" — for example, a rep at −54% versus the top performer — that number is a relative percentage change, not percentage points.

A rep at 0.23 close rate vs. a top performer at 0.50 has a gap of (0.23 − 0.50) ÷ 0.50 × 100 = −54%. That means the rep is 54% below the baseline — not 54 percentage points behind.

Tips for asking Claude effectively

Mention rep names, not IDs. Claude looks up reps by name and passes the right ID into other tools — no UUIDs to copy.
Be specific about time windows. "Last week" and "last month" are the two periods most tools support; trend data also accepts "last 4 weeks," "last 3 months," "this year," or explicit dates.
Reference call IDs from earlier in the chat. Say "look at the second one" or "the call with the SetDemoWithDM disposition" and Claude uses the right ID.
Confirm writes. Adding notes, creating or updating an Impact Plan, and — for admins — creating a sale, updating a sale line item, or creating, updating, or deleting a lead are the only changes that write to your data — Claude pauses to ask first.
Ask for a playbook by name. "Run the coaching report on Jane" or "do a team health check" triggers the built-in prompts.

Good to know

What this server does not do

To set expectations clearly:

It does not change rep assignments, team membership, or org settings. Those still happen in the Sellfire web app.
It does not initiate calls, send emails, or take actions in the app on your behalf. It only reads — plus adds coaching notes, Impact Plans, and (admins only) sales and lead records.
It is read-only for everything except coaching notes, Impact Plans, and admin-only sale and lead writes. Even within those, Claude confirms before writing.
It cannot bypass your Sellfire permissions. If no organization you can reach grants a tool, it won't appear — and Claude can't do anything in an organization that your role there doesn't allow.

The Sellfire MCPfor Claude

Coaching, in conversation

Coach faster

Never miss a struggling rep

Understand any call in seconds

Onboard clients in minutes

Choose your track

Sales Manager

Admin

Setup & connect flow

Open the connector settings in Claude

Enter the Sellfire server URL

Sign in to Sellfire

Approve access

Start asking

Connection details

If you're a Sales Manager

Your toolkit

Your playbooks

Your reference shelf

Day-to-day workflows

Coach a specific rep

Review team health

Investigate a single call

Track a lead's journey

Review closed deals & generate scripts

Try saying…

If you're an Admin

Your extra toolkit

Day-to-day workflows

Audit the roster

Spot trends over time

Track tier movement

Record & adjust a sale WRITE

Export the raw numbers

Try saying…

Built-in playbooks

Built-in coaching knowledge

What works in which client

Tool reference

Account

Calls & transcripts

Performance & dashboards

Impact Plans (coaching plans)

Leads & activities

Nightly assignment (dial queue)

Sequences

Sales & analysis

AI Gym (Sellfire staff only)

Understanding the metrics

The seven core coaching metrics

Tiers

How comparison gaps work

Tips for asking Claude effectively

What this server does not do

Ready to connect?

The Sellfire MCP
for Claude