REST API: tool management and invocation

List, fetch, invoke, enable, and disable tools through the MCPJungle HTTP API. Also covers prompt listing, fetching, and rendering endpoints.

Tools are the callable functions that registered MCP servers expose. MCPJungle aggregates them into a single proxy surface and lets you invoke them over HTTP without speaking the MCP protocol directly. Prompts are reusable message templates — they follow the same listing and fetch patterns as tools.

Tools

GET /api/v0/tools

Returns all tools registered across all enabled servers. Pass the optional server query parameter to filter by a single server.

Access: admin and user

server · string

Name of a registered MCP server. When provided, only tools from that server are returned.

List all tools

curl http://localhost:8080/api/v0/tools \
  -H "Authorization: Bearer YOUR_TOKEN"

Response 200

[
  {
    "name": "filesystem__read_file",
    "enabled": true,
    "description": "Read the contents of a file",
    "input_schema": {
      "type": "object",
      "properties": {
        "path": { "type": "string", "description": "Absolute path to the file" }
      },
      "required": ["path"]
    }
  }
]

name · string · required

Fully qualified tool name in the format server__tool.

enabled · boolean · required

Whether the tool is currently active and callable through the MCP proxy.

description · string

Human-readable description of what the tool does.

input_schema · object

JSON Schema describing the tool’s input parameters.

Show input_schema properties

type · string

Always "object" for tool input schemas.

properties · object

Map of parameter names to their JSON Schema definitions.

required · string[]

Names of parameters that must be provided.

annotations · object

Optional metadata attached to the tool by the upstream server.

---

GET /api/v0/tool

Returns a single tool by name. The name must be passed as a query parameter (not a path segment) because tool names contain slashes and double underscores.

Access: admin and user

name · string · required

Fully qualified tool name, e.g. filesystem__read_file.

cURL

curl "http://localhost:8080/api/v0/tool?name=filesystem__read_file" \
  -H "Authorization: Bearer YOUR_TOKEN"

Response 200 — A single tool object with the same shape as items returned by GET /api/v0/tools.

---

POST /api/v0/tools/invoke

Invokes a tool and returns its output. The request body is a JSON object containing the tool name and all tool-specific arguments at the top level.

Access: admin and user

name · string · required

Fully qualified tool name to invoke, e.g. filesystem__read_file.

* · any

All remaining fields are passed as arguments to the tool. The exact set of accepted fields depends on the tool’s input_schema.

cURL

curl -X POST http://localhost:8080/api/v0/tools/invoke \
  -H "Authorization: Bearer YOUR_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "filesystem__read_file",
    "path": "/tmp/notes.txt"
  }'

Response 200

content · object[] · required

Array of content blocks returned by the tool. Each block is a map with at least a type field ("text", "image", etc.) and type-specific data.

isError · boolean

true when the tool itself reported an error (distinct from an HTTP-level error). The response still returns 200 in this case.

structuredContent · any

Optional structured data returned by the tool alongside the content blocks.

_meta · object

Optional MCP metadata attached to the response.

---

POST /api/v0/tools/enable

Enables one or more tools. Pass either a fully qualified tool name (e.g. filesystem__read_file) or a server name to re-enable all tools from that server.

Access: admin only

entity · string · required

A tool name or server name. When a server name is given, all tools belonging to that server are enabled.

cURL

# Enable a single tool
curl -X POST "http://localhost:8080/api/v0/tools/enable?entity=filesystem__read_file" \
  -H "Authorization: Bearer YOUR_TOKEN"

# Enable all tools from a server
curl -X POST "http://localhost:8080/api/v0/tools/enable?entity=filesystem" \
  -H "Authorization: Bearer YOUR_TOKEN"

Response 200 — Array of tool names that were enabled.

["filesystem__read_file", "filesystem__write_file"]

---

POST /api/v0/tools/disable

Disables one or more tools, removing them from the MCP proxy surface. They remain registered and can be re-enabled.

Access: admin only

entity · string · required

A tool name or server name. When a server name is given, all tools belonging to that server are disabled.

cURL

curl -X POST "http://localhost:8080/api/v0/tools/disable?entity=filesystem__write_file" \
  -H "Authorization: Bearer YOUR_TOKEN"

Response 200 — Array of tool names that were disabled.

---

Prompts

Prompts are message templates registered by upstream MCP servers. They can accept arguments and are rendered server-side before being returned.

GET /api/v0/prompts

Returns all prompts across all enabled servers. Pass the optional server query parameter to filter by a single server.

Access: admin and user

server · string

Name of a registered MCP server. When provided, only prompts from that server are returned.

cURL

curl http://localhost:8080/api/v0/prompts \
  -H "Authorization: Bearer YOUR_TOKEN"

---

GET /api/v0/prompt

Returns metadata for a single prompt by name. The name must be passed as a query parameter.

Access: admin and user

name · string · required

Fully qualified prompt name.

cURL

curl "http://localhost:8080/api/v0/prompt?name=myserver__summarize" \
  -H "Authorization: Bearer YOUR_TOKEN"

---

POST /api/v0/prompts/render

Renders a prompt template with the provided arguments and returns the resulting messages.

Access: admin and user

name · string · required

Fully qualified prompt name to render.

arguments · object

Key-value map of argument names to string values. The accepted keys depend on the prompt’s own argument schema.

cURL

curl -X POST http://localhost:8080/api/v0/prompts/render \
  -H "Authorization: Bearer YOUR_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "myserver__summarize",
    "arguments": {
      "text": "MCPJungle is a self-hosted MCP gateway.",
      "language": "English"
    }
  }'

Response 200

description · string

Optional description of the rendered prompt.

messages · object[] · required

Rendered message sequence. Each message has a role ("user" or "assistant") and a content map.

Show message properties

role · string · required

"user" or "assistant".

content · object · required

Content block from the MCP prompt result. Shape varies by content type.

meta · object

Optional MCP metadata attached to the response.

{
  "description": "Summarize the provided text",
  "messages": [
    {
      "role": "user",
      "content": {
        "type": "text",
        "text": "Please summarize the following in English:\n\nMCPJungle is a self-hosted MCP gateway."
      }
    }
  ]
}

---

Prompt enable / disable

The enable and disable endpoints for prompts follow the same pattern as tools.

Endpoint	Method	Description
/api/v0/prompts/enable	POST	Enable a prompt or all prompts from a server. Pass ?entity=<name>.
/api/v0/prompts/disable	POST	Disable a prompt or all prompts from a server. Pass ?entity=<name>.

Both endpoints are admin-only and return an array of prompt names that were affected.