PeerPanda MCP for Claude, ChatGPT, Copilot and AI Assistants

Introduction

The PeerPanda MCP Server allows AI assistants to securely access data from the PeerPanda Competitive Intelligence platform on behalf of a signed-in user.

Key Use Cases

  • Preparing monthly or quarterly competitive reports
    An AI assistant can summarize recent competitor activity tracked in PeerPanda and turn it into a structured report that follows your provided format.
  • Deep-diving on competitor activity
    When an activity is spotted in PeerPanda, the assistant can investigate further by combining PeerPanda data with relevant information from the public web.
  • Answering competitive questions with internal context
    The assistant can combine PeerPanda intelligence with your internal documents (for example, strategy notes or past analyses) to give more complete, context-aware answers.

How AI assistants can use PeerPanda

The MCP server works with multiple AI assistants including Claude (Desktop and Web), ChatGPT (Plus and Pro subscriptions), Mistral Le Chat, Dust.tt, and Microsoft Copilot Studio.

It allows the AI assistant to access PeerPanda data as the signed-in user, respecting that user’s permissions.

Important notes:

  • The set of tools available through MCP may change over time, and the MCP server is meant for interactive AI usage only.
  • It does not replace the PeerPanda API and should not be used as a programmatic or backend integration.

How Authentication works

  • When a user connects, they are asked to sign in to their PeerPanda account.
  • Once signed in, a secure connection is created between the AI tool and PeerPanda.
  • Access is user-scoped:
    • One PeerPanda user = one authenticated connection

General Instructions to setup PeerPanda MCP

  1. First, make sure that you are logged-in to your PeerPanda Account
  2. Add a new MCP server in your AI Assistant of choice.
  3. Enter the server URL: https://mcp.getpeerpanda.com/mcp
  4. Follow the on-screen sign-in prompts to authorize the PeerPanda MCP access to your competitor data. You might need to select OAUTH autoconfiguration or a similar setting. There is no need to specify a Client ID nor a Client Secret
  5. When the MCP is setup, adding the Competitive Intelligence Skill (at the bottom) to your agent yields the best results

ChatGPT Instructions

For adding PeerPanda MCP, you need to have an active ChatGPT Plus, or ChatGPT Pro subscription.

  • First, make sure that you are logged-in to your PeerPanda Account
  • In ChatGPT, click your name in the bottom left-hand corner of the screen
  • Select Settings in the drop down list
  • Open the Apps panel, then in the Advanced Settings, click Create App
  • Fill the form with the example below
  • Leave OAuth Client ID and OAuth Client Secret blank
Adding PeerPanda MCP in ChatGPT

Adding PeerPanda MCP in ChatGPT

Dust.tt Instructions

For adding PeerPanda MCP, you need to have an active Dust subscription, and you need to be an administrator of your workspace.

You might find the Dust documentation useful.

  • First, make sure that you are logged-in to your PeerPanda Account
  • In Dust, click Spaces in the top left-hand corner of the screen
  • In the Administration section click Tools.
  • In the right part of the screen, click + Add Tools, then + Add MCP Server
  • Paste the PeerPanda MCP URL : https://mcp.getpeerpanda.com/mcp
  • Keep Authentication to Automatic, and Custom Headers Disabled
  • Click Save
Adding PeerPanda MCP to Dust

Adding PeerPanda MCP in Dust

  • If people who will be using PeerPanda MCP in Dust have PeerPanda accounts, then select Personal accounts
  • Otherwise, Shared accounts will enable Dust users to access PeerPanda information through your own account
PeerPanda MCP Account Sharing in Dust

Select Personal or Shared Account

Finally, follow the prompts to finish connecting your PeerPanda account.

PeerPanda MCP Dust Tools Settings

Dust PeerPanda MCP Tool Configuration

You may finish the setup by setting all PeerPanda tools as Never Ask.

PeerPanda tools cannot delete or modify data, so it's safe to do so, and will make the experience smoother for your users.

Microsoft Copilot Instructions

Use PeerPanda’s MCP server to let Microsoft Copilot agents search competitors and retrieve recent competitive activity directly from your PeerPanda workspace.

This integration is read-only and can use per-user permissions: Copilot can only access data the signed-in user is allowed to see.

You might find the Microsoft Copilot Studio Documentation useful

Add PeerPanda as an MCP tool in Copilot Studio

  1. Open Microsoft Copilot Studio
  2. Open your agent
  3. Go to Tools
  4. Select Add a tool
  5. Choose New tool
  6. Select Model Context Protocol (MCP)


MCP server configuration

Fill in the MCP server fields as follows:

Server name: PeerPanda Competitive Intelligence

Server description: Search competitors and retrieve recent competitor activity tracked in PeerPanda.”

Server URL: https://mcp.getpeerpanda.com/mcp


Authentication setup

When prompted for authentication:

  1. Choose OAuth 2.0
  2. Select Dynamic discovery
  3. Do not enter a Client ID or Client Secret
  4. Continue and complete the sign-in flow when prompted

You will be asked to authenticate with PeerPanda during setup. This links the MCP connection to your user account and permissions.


Save and test

Save the MCP tool and open the Test chat for your agent

Try a prompt like: "List my PeerPanda Competitors"

If configured correctly, Copilot will automatically call PeerPanda’s MCP tools when relevant.

Mistral Le Chat

For adding PeerPanda MCP, you need an active Le Chat Free, Le Chat Pro, or Le Chat Student account.

This feature requires Admin privileges on Le Chat, which all account owners have by default.

This integration is read-only and can use per-user permissions: Le can only access data the signed-in user is allowed to see.

You might find the Le Chat Custom Connector documentation useful

Steps to Configure PeerPanda in Mistral Le Chat

First, make sure that you are logged in to your PeerPanda Account.

  • In Le Chat, click the toggle panel button to reveal the side panel if it's not already visible.
  • Expand the Intelligence menu and click Connectors.

Le Chat Connectors

Connectors are available in the left panel

Click the + Add Connector button on the right side of the page.

This opens the MCP Connectors directory.

At the top of the MCP Connectors directory pop-up, click the Custom MCP Connector tab to switch to the Custom MCP Connector creation form.


MCP server configuration

Fill in the MCP server fields as follows:

Server name: PeerPanda Competitive Intelligence

Server description: Search competitors and retrieve recent competitor activity tracked in PeerPanda.”

Server URL: https://mcp.getpeerpanda.com/mcp

Authentication method: OAuth 2.1

Leave the Client ID and Client Secret fields blank if they appear.

Setting up PeerPanda MCP in Le Chat

Setting up PeerPanda MCP in Le Chat

Click the Connect button and follow the on-screen prompts to authorize PeerPanda access to your account through the OAuth authentication flow.

Setting Function Permissions (Recommended)

After successfully connecting PeerPanda, you can configure function-level permissions to avoid repetitive permission prompts.

  • Click your PeerPanda Connector card to open its details.
  • Go to the Functions tab.
  • For each function listed, toggle the Always Allow setting to enabled.

Since PeerPanda tools are read-only and cannot modify or delete data, it is safe to enable Always Allow for all PeerPanda functions.

Setting up PeerPanda tools permissions in Le Chat

Setting up PeerPanda tools permissions in Le Chat


Save and test

Once you've configured PeerPanda, you can test the connection to make sure everything is working correctly.

Open a new conversation in Le Chat and try a simple prompt such as "list my PeerPanda competitors"

If configured correctly, Le Chat will automatically call PeerPanda's MCP tools and return information from your PeerPanda workspace.

Enable the PeerPanda tool when talking to Le Chat

Enable the PeerPanda tool when talking to Le Chat

Competitive Intelligence Skill

This skill teaches your agent to use the PeerPanda MCP.

You may add it to Claude directly, or integrate it in the instructions of your agents.

---
name: competitive-intelligence
description: >
  Competitive intelligence research using PeerPanda and web search.
  Use whenever the user asks about competitors, a competitive landscape, battlecards,
  "what's [competitor] up to?", "any news on [company]?", or "how does X compare to us?"
  Trigger on competitor-tracking or competitive-intel intent even if PeerPanda isn't named.
---

# PeerPanda Competitive Intelligence Skill

You are a competitive intelligence analyst working on top of **PeerPanda's MCP server**.

PeerPanda is a competitive intelligence platform. Its MCP exposes the customer's **curated**
competitor database: tracked competitors, an activity timeline (news, product updates, posts,
press releases) that is AI-summarized with relevance scores and tags, and full-text search over it all.
Treat it as proprietary signal the open web doesn't have. Always check PeerPanda first.

## Tool Selection

- **`get_competitors`** — Call first. Returns competitor **slugs** and **groups** you need for filtering.
- **`get_account_info`** — Returns account metadata and the available **activity tag** names. Call before
  filtering activities by tag.
- **`get_competitor_activities`** — Time-based questions ("what happened recently", "briefing").
  - Set `since` explicitly — it defaults to **7 days**. Use **60 days** when no timeframe is given.
  - `until` bounds the upper end for fixed windows. `competitors` / `competitor_groups` / `activity_tags`
    (comma-separated) scope it. `min_relevance` 0.6+ for focused work, 0.4 for broad (API default 0.4).
  - `page_size` maxes at 50; the response paginates. Page through for a full briefing.
- **`search_competitor_data`** — Topic questions ("anyone mention pricing?"). Pure full-text search:
  boolean operators work as expected — **AND** strictly narrows (all terms must match), **OR** unions,
  **NOT** excludes, and parentheses group. Quoted phrases for exact matches; `sort` = relevance
  (default) / newest / oldest; optional `since`. Searches text across all competitors (no ownership
  filter — see ownership vs. mentions below). Start broad with OR synonyms, then narrow with AND.
- **`get_user_info`** — Rarely needed; only to confirm who's authenticated.

## Example calls

```
# "What has Acme been up to lately?" → that competitor's activity, recent first
get_competitor_activities(since="2025-06-12", competitors="acme-corp", min_relevance=0.6)

# "Any acquisitions or funding across our market?" → topic sweep, all competitors
search_competitor_data(q="acquisition OR merger OR funding OR buyout", sort="newest")

# "Who's been talking about Acme?" → mentions of that name anywhere in the database
search_competitor_data(q="acme-corp AND (partnership OR launch OR pricing)", sort="newest")
```
## If a query comes back empty

Empty usually means filters are too tight, not that nothing is happening. Broaden before concluding:
lower `min_relevance` (e.g. 0.6 → 0.4), widen `since` (e.g. 60 → 180 days), drop `competitors` /
`competitor_groups` / `activity_tags` filters one at a time, and for `search_competitor_data` loosen
the query — add OR synonyms, remove AND clauses, and include local-language terms. Only after a genuinely
broad query still returns nothing should you tell the user there's no coverage for that topic.


## Output Rules
  
- **Show relevance as the indicator, not the number.** Display PeerPanda's `relevance_score_icons`
  (⚡⚡⚡ / ⚡⚡) rather than the raw `relevance_score`.
  Put the indicator **before** the activity title (e.g. `⚡⚡⚡ [Title](peerpanda_url)`) so importance
  reads at a glance. Use the numeric score internally to rank, threshold, and decide what to surface.
- **Cite every PeerPanda claim** as `⚡⚡⚡ [Activity title](peerpanda_url)` Never invent URLs.
  `peerpanda_url` = the in-app reference
- **Date every claim**
- **Match output to question size** — quick question, quick answer; briefing request, structured report.
- **Match the user's language** in your answer. For tool queries, English works when the corpus or
  PeerPanda's AI summaries are English; for a non-English corpus, add local-language synonyms
  (e.g. `pricing OR tarif OR prix`) or you'll miss source-text matches.
- Use web search only to supplement (breaking news, pricing, untracked competitors) **if it's available**.