CLI Reference

Global Options

api2mcp [--log-level LEVEL] [--version] COMMAND [ARGS]...

Option	Default	Description
--log-level	warning	Root logging level: debug, info, warning, error, critical
--version	—	Show version and exit

---

api2mcp generate

Generate an MCP server from an API spec file.

api2mcp generate SPEC [OPTIONS]

Arguments

Argument	Description
SPEC	Path to the API spec file (OpenAPI YAML/JSON, GraphQL SDL, Postman Collection)

Options

Option	Default	Description
--output, -o	./generated	Output directory for generated files
--base-url	from spec	Override the server base URL
--config, -c	auto-detect	Path to .api2mcp.yaml config file
--format	auto-detect	Force input format: openapi, graphql, postman, swagger

Examples

# Generate with default output dir
api2mcp generate openapi.yaml

# Custom output directory
api2mcp generate openapi.yaml --output ./my-server

# Override base URL
api2mcp generate openapi.yaml --base-url https://staging.example.com/api

# Use explicit config file
api2mcp generate openapi.yaml --config ./prod.api2mcp.yaml

---

api2mcp serve

Start a generated MCP server.

api2mcp serve SERVER_DIR [OPTIONS]

Arguments

Argument	Description
SERVER_DIR	Directory containing the generated server (must have spec.yaml)

Options

Option	Default	Description
--host	127.0.0.1	Bind address
--port	8000	Port number
--transport	http	Transport: http (Streamable HTTP) or stdio
--config, -c	auto-detect	Config file path

Examples

api2mcp serve ./generated
api2mcp serve ./generated --host 0.0.0.0 --port 9000
api2mcp serve ./generated --transport stdio

---

api2mcp dev

Start a hot-reload development server that watches the spec file for changes.

api2mcp dev SPEC [OPTIONS]

Arguments

Argument	Description
SPEC	Path to the API spec file to watch

Options

Accepts all options from api2mcp serve, plus:

Option	Default	Description
--output, -o	./generated	Output directory (re-generated on change)
--watch-dir	spec parent dir	Additional directory to watch

Examples

api2mcp dev openapi.yaml
api2mcp dev openapi.yaml --port 9000 --output ./dev-server

---

api2mcp validate

Validate an API spec without generating any output.

api2mcp validate SPEC [OPTIONS]

Options

Option	Default	Description
--strict	false	Treat warnings as errors

Exit codes

Code	Meaning
0	Valid (or valid with warnings when not --strict)
1	Validation errors found

---

api2mcp wizard

Interactive step-by-step MCP server creation wizard.

api2mcp wizard [OPTIONS]

Options

Option	Description
--spec	Pre-fill the spec file path
--output	Pre-fill the output directory
--no-confirm	Skip confirmation prompts

---

api2mcp template

Manage MCP server templates from the community registry.

api2mcp template search

api2mcp template search [QUERY] [--verbose]

Search the registry. Leave QUERY empty to list all templates.

api2mcp template list

api2mcp template list [--verbose]

List all available templates.

api2mcp template install

api2mcp template install NAME [--version TAG] [--dest DIR]

Argument/Option	Default	Description
NAME	required	Template slug
--version, -V	latest	Git tag to install
--dest, -d	./<NAME>	Installation directory

api2mcp template update

api2mcp template update NAME [--version TAG] [--dest DIR]

Update an installed template to a newer version.

---

api2mcp orchestrate

Run a LangGraph orchestration workflow against one or more MCP servers.

api2mcp orchestrate PROMPT [OPTIONS]

Arguments

Argument	Description
PROMPT	Natural-language task description

Options

Option	Default	Description
--server NAME=URL	—	MCP server to connect (repeatable)
--graph TYPE	reactive	Workflow type: reactive, planner, conversational
--mode MODE	sequential	Execution mode for planner graph: sequential, parallel, mixed
--api-name NAME	—	API name to expose from --server (repeatable; defaults to all)
--model ID	claude-sonnet-4-6	Claude model ID
--thread-id ID	—	Thread ID for checkpointing / conversation memory
--stream	off	Stream output tokens in real-time
--checkpoint PATH	in-memory	SQLite DB path for checkpointing
--output-format	text	Output format: text or json

Graph types

Type	Description
reactive	create_react_agent loop — best for single-API tool use
planner	Sequential / parallel / mixed multi-step planner
conversational	Human-in-the-loop conversational agent

Examples

# Run a reactive agent against a GitHub MCP server
api2mcp orchestrate "List open PRs in anthropics/claude" \
  --server github=http://localhost:8001 \
  --api-name github \
  --graph reactive

# Multi-API planner in parallel mode
api2mcp orchestrate "Sync Jira issues from GitHub" \
  --server github=http://localhost:8001 \
  --server jira=http://localhost:8002 \
  --graph planner \
  --mode parallel

# Conversational agent with SQLite checkpointing
api2mcp orchestrate "Help me debug this issue" \
  --graph conversational \
  --checkpoint ./sessions.db \
  --thread-id my-session

---

api2mcp diff

Compare two API specifications and report added, removed, and changed endpoints.

api2mcp diff SPEC_A SPEC_B [OPTIONS]

Arguments

Argument	Description
SPEC_A	Path to the first (baseline) spec file
SPEC_B	Path to the second (updated) spec file

Options

Option	Default	Description
--output-format FORMAT	text	Output format: text or json

Exit codes

Code	Meaning
0	No differences found
1	Differences found
2	Parse error

Examples

# Compare two OpenAPI specs
api2mcp diff openapi-v1.yaml openapi-v2.yaml

# Machine-readable output for CI
api2mcp diff old.yaml new.yaml --output-format json

---

api2mcp export

Export a generated MCP server in various distribution formats.

api2mcp export SERVER_DIR [OPTIONS]

Arguments

Argument	Description
SERVER_DIR	Directory containing the generated server

Options

Option	Default	Description
--format FORMAT	zip	Export format: zip, tar, docker
--output PATH	./export	Output path

Examples

# Export as zip archive
api2mcp export ./generated --format zip --output ./my-server.zip

# Export as Docker image tarball
api2mcp export ./generated --format docker --output ./my-server.tar