PUBLIC TOOL REGISTRY

Deterministic tools AI agents can call instantly

This registry is built for direct agent use. Every tool publishes inputs, outputs, manifests, source paths, sample payloads, auditable work steps, and a stateless execution route. Agents can call one tool or chain multiple tools together by mapping outputs to downstream inputs.

Start with Quickstart Open Chain Studio

28tools

9categories

JSONstateless requests

MCPcompatible access

QUICKSTART

How an agent uses this site

1. Discover tools

Read the registry index, filter by category, or open a manifest for a single tool.

GET /data/tools.json
GET /data/manifests/arithmetic-calculator.json

2. Execute a tool

Send a stateless JSON request. The response includes structured output plus work steps for audit and reasoning support.

POST /api/execute
{
  "tool": "arithmetic-calculator",
  "input": { "expression": "(12000 * 0.18) + 950" }
}

3. Use MCP style calls

List tools or call them through a compact MCP style route.

POST /api/mcp
{ "method": "tools/list" }

POST /api/mcp
{
  "method": "tools/call",
  "params": {
    "name": "date-normalizer",
    "arguments": { "value": "Jan 5th 24" }
  }
}

4. Chain tools

Run multiple tools in sequence and map prior outputs into later inputs with JSON path references like $0.text or $1.records.

[
  {
    "tool": "url-content-fetcher",
    "input": { "url": "https://example.com", "max_chars": 800 }
  },
  {
    "tool": "regex-extraction-engine",
    "input": {
      "text": "$0.text",
      "patterns": ["Example Domain"]
    }
  }
]

REGISTRY

Available tools

Every tool is deterministic and stateless.

WORKSPACE

Select a tool

Manifest

deterministic stateless MCP

Choose a tool from the registry table to inspect its contract and run it.

Contract

Inputs

Outputs

Input JSON

Output JSON

Work steps

Manifest

Audit

Execution flow
Deterministic source path
Agent usage notes: Run with stateless JSON. Review work steps before using outputs in downstream reasoning.

CHAIN STUDIO

Compose chained tool calls

Define a chain as a JSON array. Any string value that starts with $0., $1., or $2. will be resolved against the output of a prior step before execution. This lets an agent map outputs into downstream inputs without custom server state.

Chain spec

Chain results

Chain work

API REFERENCE

Execution surfaces

Execute endpoint

Runs one deterministic tool.

POST /api/execute
Content-Type: application/json

{
  "tool": "tool-id",
  "input": { ... }
}

MCP endpoint

Lists tools and routes a named tool call.

POST /api/mcp
Content-Type: application/json

{ "method": "tools/list" }
{ "method": "tools/call", "params": { "name": "tool-id", "arguments": { ... } } }

Stateless rules

No file uploads
No server-side job state
No session memory between requests
Pass URLs, raw text, JSON, or base64 in the body

Auditability

Visible work steps in responses
Per-tool manifests
Deterministic source paths
Consistent structured outputs