FAQs

Browse frequently asked questions about the Pliant MCP, grouped by topic.

Permissions & Access
Data & Security
Capabilities & Use Cases
Setup & Compatibility
Pricing
Troubleshooting

Permissions & Access

What permissions does the Pliant MCP have?

Actions taken through the Pliant MCP follow the same permissions as the connected member. The agent can never do more in Pliant than the member who connected it can do. See Access & Permissions for details.

Can I enable the Pliant MCP for only specific members?

Not currently. MCP access is enabled or disabled for an entire organization at once, and results are then scoped per member by their existing role. Per-member enablement controls may come later.

How do I revoke access for an agent or MCP client?

See Disconnect Pliant MCP for the steps, both from inside Pliant and from your AI client.

Data & Security

How is data processed between Pliant and my AI tool?

Pliant makes data available through the Pliant MCP, and any data read through the MCP is passed to your connected agent, AI tool, and model provider. Pliant also has access to everything sent to Pliant through the MCP's write tools.

Could the Pliant MCP accidentally change my organization's setup?

No. The Pliant MCP cannot change organization configuration, card programs, or user access. It only reads data and, for a narrow set of write tools, updates a transaction's comment, team, project, category, subcategory, VAT rate, or supplier. Every change is visible in Pliant's activity logs.

Can the Pliant MCP move money or change a card?

No. It cannot move money, top up an account, issue or lock a card, or change a card limit. See What It Doesn't Do for the full list.

Capabilities & Use Cases

What use cases are covered by the Pliant MCP?

The Pliant MCP can search transactions, cards, members, and teams, get exact counts for review/receipt/export status, and read or update a transaction's coding. See Pliant MCP Overview for the full list of what's supported today.

What's the difference between the Pliant MCP and the Pliant API?

The Pliant MCP is optimized for AI agents: connect it directly to a client like Claude, and the agent decides which tools to call and when. The Pliant API is designed for code to call directly with explicit, predetermined logic.

Can I use the Pliant MCP to find missing receipts?

Yes. Ask for transactions missing a receipt to get an exact count and list, and ask the agent to check or view a specific transaction's receipt. See Example Use Cases.

Setup & Compatibility

Can I use the Pliant MCP with a custom LLM or agent?

Yes. Any MCP-compatible client that supports a remote Streamable HTTP server can connect. See Any Other MCP Client. Your organization may limit which MCP servers you're allowed to connect.

Is the Pliant MCP listed in Claude's, ChatGPT's, or Codex's connector directory?

Not yet - it's currently added as a custom connector in each. See Set Up Pliant MCP.

What do I need to add Pliant as a custom connector?

The Pliant MCP server URL, and a Client ID and Client Secret. Ask Pliant support for the Client ID and Secret for your organization and AI tool - see Set Up Pliant MCP for exactly where to enter them.

Troubleshooting

Sign-in fails immediately when I try to connect

You may already have an active Pliant session in your browser from something else. Log out of Pliant, then try connecting again.

My agent says a capability isn't available

Check that your organization admin has enabled the Pliant MCP module (see Access & Permissions), and that the action you're asking for is in scope (see What It Doesn't Do). If both check out, contact Pliant support with the exact prompt and error.