Integration walkthrough · Updated 2026-05-22

PandaDoc MCP + Claude: setup walkthrough and the 5 workflows that earn their install

PandaDoc publishes a hosted MCP server documented at developers.pandadoc.com/docs/use-pandadoc-mcp-server. API-key authenticated, write-capable (create, send, update, manage templates), and reachable from any MCP client. This walkthrough covers the setup, the Business-tier requirement that gates API access, the 5 highest-leverage workflows, and the bulk-send gotcha every operator should know before turning the LLM loose on real proposals.

Want to try PandaDoc?

PandaDoc MCP requires Business tier ($65/user/mo, 3-user minimum)

The write surface is real — create, send, update from Claude. API access starts on Business; Free eSign and Essentials are out of reach.

Start with PandaDoc →Affiliate link — StackSwap earns a commission if you sign up for PandaDoc. We only partner with tools we'd recommend anyway.

Step 1: Confirm your PandaDoc plan supports API access

PandaDoc gates API access to the Business plan ($65/user/mo, 3-user minimum) and Enterprise tier. Free eSign and Essentials ($35/user/mo) don't include API — the MCP is literally unreachable on those plans. If you're below Business, you'll need to upgrade before the MCP integration is possible. The Business floor is $195/mo (3 users × $65/user). For most B2B sales teams sending 20+ proposals/month, this earns its cost quickly through the AI-driven proposal-drafting workflow; for solo founders sending 2-3 proposals/month, probably not.

Step 2: Generate a PandaDoc API key

In PandaDoc: Settings → Integrations → API & Webhooks → Create API Key. Name it something explicit — "Claude integration (Nick)" or "AI - scoped writes" — so the activity log is readable later. Copy the token immediately; PandaDoc shows it once.

Operator hygiene: don't use your admin user's key. Create a dedicated PandaDoc user scoped to the templates and workspaces the agent should touch, then generate the API key under that user. The LLM inherits the user's permissions; scope = blast-radius control.

Step 3: Add PandaDoc MCP to Claude

Total setup time: about 5 minutes if you already have an API key, 10 if you need to create the scoped user too.

Step 4: Verify the connection

Smoke test in chat: "Using the PandaDoc MCP, list the templates in my account." If the connector is healthy, Claude responds with your actual template list. If the call fails, check three things: (a) API key copied without truncation, (b) endpoint URL correct, (c) the user behind the API key actually has access to templates in PandaDoc's UI. Fix in that order before deeper troubleshooting.

The 5 workflows that earn their install

1. Draft and send proposals from a template

The flagship workflow. "Draft a proposal for Acme Corp using the SaaS template — pricing is $24K ARR for 3 seats, start date June 1, payment net 30. Send it to jane@acme.example.com." The LLM picks the template, fills the variables, and sends. Replaces the manual template-fill workflow that previously took 15-20 minutes per proposal.

2. Identify stalled deals in viewed-not-signed status

"Show me every proposal sent in the last 30 days that's been viewed but not signed, and sort by deal value." The MCP returns the filtered list; the LLM helps prioritize follow-ups. The visibility into stalled deals is a regular operator pain point that this workflow fixes in one chat.

3. Update variables on in-flight proposals

"The Acme proposal I sent yesterday — update the start date to June 15 and the implementation fee to $5K instead of $7.5K." The MCP updates variables on the existing document without rebuilding it; the recipient sees the corrected version on their next view. No "resend with new doc" awkwardness.

4. Bulk template inventory cleanup

"Walk my template library and tell me which templates haven't been used in 90+ days. Archive the ones tagged 'legacy' or 'archived' that match. Surface anything that looks like a duplicate of an active template." A scheduled 30-minute cleanup that previously required clicking through the template library, compressed into a single chat exchange.

5. KBA / Notary on compliance-grade docs (tier-gated)

For workflows that require knowledge-based authentication or live remote online notarization, the MCP exposes those endpoints — but they're tier-gated to higher plans and charged per use. If your tier supports them, the workflow becomes: prompt Claude to send a document with KBA verification enabled, recipient verifies via security questions, signed copy returns. Validate plan support before building agent loops that depend on these endpoints.

The bulk-send gotcha — confirm before you click

Bulk operations (send 50 proposals, update variables across 100 documents) are where the blast radius lives. Three mitigations:

Where StackSwap MCP fits alongside PandaDoc MCP

PandaDoc MCP exposes your PandaDoc data. StackSwap MCP exposes the cross-vendor GTM catalog — ~400 tools with monthly costs, 104 hand-verified overlap pairs, partner sign-up paths. Load both into the same Claude session: "draft this proposal" (PandaDoc MCP) plus "what should our proposal + e-sign stack look like at our scale" (StackSwap MCP). One chat surface, two layers of decision support.

Connect StackSwap MCP free →

Want to try PandaDoc?

PandaDoc MCP + Claude is the SMB proposal-tool default in 2026

Business tier real, write surface real, 5-minute setup. The 5 workflows above pay for the install on the first proposal.

Start with PandaDoc →Affiliate link — StackSwap earns a commission if you sign up for PandaDoc. We only partner with tools we'd recommend anyway.

FAQ

First, confirm you're on the Business plan ($65/user/mo, 3-user minimum) or Enterprise — API access isn't included on Free eSign or Essentials. Generate an API key in PandaDoc (Settings → Integrations → API & Webhooks → Create API Key). In Claude Desktop or claude.ai: Settings → Connectors → Add custom MCP, paste the PandaDoc MCP endpoint and your API key. In Claude Code: add the server to your workspace .mcp.json. Reference docs at developers.pandadoc.com/docs/use-pandadoc-mcp-server document the schema. Smoke test in chat: 'List my PandaDoc templates' — if you see your templates, the connection is good.

Yes, and this is the major cost-floor consideration. The MCP requires API access, which starts at the Business plan: $65/user/mo with a 3-user minimum, so the effective floor is $195/mo just to use the MCP. Free eSign and Essentials don't include API access. Enterprise tier adds advanced compliance features (KBA, Notary) and content-library depth. For most operator stacks, Business is the right tier — only step up to Enterprise if compliance posture or volume gates push you there.

Five workflows: (1) draft and send a proposal from a template, with the LLM pulling variables from CRM context or chat conversation — collapses the 'paste customer info into template builder' workflow into one chat; (2) search documents by status, recipient, or content to identify stalled deals in viewed-not-signed status; (3) update document variables on in-flight proposals when pricing or terms change — no rebuild required; (4) bulk template-inventory management — archive stale templates, surface usage stats, rename for consistency; (5) initiate KBA-authenticated or Notary workflows on documents that require compliance-grade identity verification (tier-gated, validate your plan first).

PandaDoc enforces standard rate limits on the API and therefore on MCP calls. For interactive chat, you almost never hit the ceiling — the rate limits are designed for sustained API usage, not for human-typing-speed chat conversations. For agent loops that fan out (bulk send, bulk update, full template inventory walks), you can hit limits. Mitigation: batch operations into groups of 10-20 documents with a short wait between batches, and watch PandaDoc's activity log during the first heavy session to calibrate. Claude surfaces rate-limit errors in chat with the wait window so you can adjust pacing.

Yes — same operator hygiene as every MCP credential. Create a dedicated API key for Claude in PandaDoc, scoped to the workspaces and templates the agent should touch. Don't paste your admin-scoped key into a chat connector; admin scope gives the LLM full workspace control, which is overkill for most workflows and creates a meaningful blast radius if something goes wrong. The activity log shows every action under the key that issued it, so isolating AI-driven activity from production workflows pays off the first time you need to audit.

Bulk operations are where the write surface earns its name and where the blast radius lives. Sending 50 contracts via an unattended agent loop is a different risk shape than searching for documents. Mitigations: (1) configure your MCP client's confirmation UX so bulk-send operations require explicit human approval — Claude Desktop, claude.ai, and Claude Code handle this differently, so test before relying on it; (2) start agent loops with a system prompt that requires explicit confirmation before any send operation touching more than 5 documents; (3) for high-stakes legal documents, keep human-in-the-loop review regardless of how confident the LLM seems.

Zapier routes through a workflow runner — Claude calls Zapier, Zapier calls PandaDoc. That adds latency, an extra failure point, and a separate cost line ($19.99-$199/mo on the volumes that matter). Direct PandaDoc MCP is structurally simpler: Claude → PandaDoc, one hop. For interactive chat-driven proposal drafting and sending, native MCP wins. Zapier still earns its place for event-driven workflows ('when document is signed, post to Slack and create Asana task') — different problem, different tool. Most operator stacks run both, MCP for chat-driven querying and Zapier for event-driven destination plumbing.

Yes to all three. Claude Desktop and claude.ai both support custom MCP connectors — Settings → Connectors → Add custom MCP, paste endpoint + API key. Claude Code reads MCP server config from your workspace .mcp.json or project-level config. The same MCP works across all three Claude surfaces with identical config; that's the design intent of the protocol. The behavioral difference between surfaces is the confirmation UX — Claude Desktop tends to be the most explicit about confirming write operations; claude.ai batches confirmations; Claude Code varies by version. Test your bulk-send flow on the surface you'll actually use.

Related reading

Canonical URL: https://stackswap.ai/pandadoc-mcp-claude-integration. Disclosure: StackSwap is a PandaDoc affiliate. Setup steps above are the same ones we use internally.