MCPInstall

mcp2cli-runtime-api-tooling

by reason-machinesBackend & APIs

Turn MCP, OpenAPI, or GraphQL servers into CLIs at runtime with zero codegen, saving 96-99% of tokens on tool schemas

Installs
223
Last 24h
—
Trending rank
—
Security
Not audited

Install

Works with every agent the skills CLI supports.

npx skills add https://github.com/reason-machines/devtools-skills --skill mcp2cli-runtime-api-tooling
  1. 01

    Install

    Run the command in your project. The CLI asks which agents to install for.

  2. 02

    Describe the task

    Ask your agent for something this skill covers. It loads mcp2cli-runtime-api-tooling automatically when relevant.

  3. 03

    Review the result

    The agent follows the SKILL.md steps, scripts and references bundled with the skill.

SKILL.md

The instructions your agent reads when the skill is triggered.

mcp2cli — Runtime API to CLI

Skill by ara.so — Devtools Skills collection.

mcp2cli dynamically converts MCP servers, OpenAPI specs, and GraphQL endpoints into CLIs at runtime, eliminating the need for code generation and reducing token waste by 96-99% when used with AI agents. It introspects APIs, generates subcommands, handles authentication (including OAuth), and caches schemas intelligently.

Installation

# Run directly without installing
uvx mcp2cli --help

# Or install globally
uv tool install mcp2cli

# Install as an AI agent skill
npx skills add knowsuchagency/mcp2cli --skill mcp2cli

Core Concepts

mcp2cli operates in four modes:

  • MCP HTTP/SSE: Connect to MCP servers over HTTP with SSE or streamable HTTP transport
  • MCP stdio: Launch and communicate with MCP servers via stdin/stdout
  • OpenAPI: Convert REST APIs with OpenAPI specs into CLIs
  • GraphQL: Introspect GraphQL endpoints and generate query/mutation commands

All modes support authentication, caching, and can be "baked" into named configurations for reuse.

MCP HTTP/SSE Mode

# List available tools from an MCP server
mcp2cli --mcp https://mcp.example.com/sse --list

# Call a specific tool
mcp2cli --mcp https://mcp.example.com/sse search --query "rust packages"

# With authentication
mcp2cli --mcp https://mcp.example.com/sse \
  --auth-header "x-api-key:env:MCP_API_KEY" \
  query --sql "SELECT * FROM users LIMIT 10"

# Force specific transport (skip auto-detection)
mcp2cli --mcp https://mcp.example.com/sse --transport sse --list

# Search for tools by name or description
mcp2cli --mcp https://mcp.example.com/sse --search "database"

MCP stdio Mode

# Launch an MCP server and list its tools
mcp2cli --mcp-stdio "npx @modelcontextprotocol/server-filesystem /tmp" --list

# Call a tool with arguments
mcp2cli --mcp-stdio "npx @modelcontextprotocol/server-filesystem /tmp" \
  read-file --path /tmp/data.json

# Pass environment variables to the server process
mcp2cli --mcp-stdio "node mcp-server.js" \
  --env "DATABASE_URL=env:DATABASE_URL" \
  --env "LOG_LEVEL=debug" \
  fetch-records --limit 50

OpenAPI Mode

# List all endpoints from a remote spec
mcp2cli --spec https://petstore3.swagger.io/api/v3/openapi.json --list

# Call an endpoint with parameters
mcp2cli --spec https://api.example.com/openapi.json \
  --base-url https://api.example.com \
  list-pets --status available --limit 20

# Use a local spec file
mcp2cli --spec ./openapi.yaml --base-url http://localhost:8000 --list

# With authentication
mcp2cli --spec ./api-spec.json \
  --base-url https://api.example.com \
  --auth-header "Authorization:Bearer env:API_TOKEN" \
  create-item --name "Test Item" --price 99.99

# POST with JSON body from stdin
echo '{"name": "Fido", "species": "dog", "age": 3}' | \
  mcp2cli --spec ./spec.json create-pet --stdin

# Pretty-print JSON output
mcp2cli --spec ./spec.json --pretty list-users

GraphQL Mode

# List all queries and mutations
mcp2cli --graphql https://api.example.com/graphql --list

# Execute a query
mcp2cli --graphql https://api.example.com/graphql \
  users --limit 10

# Execute a mutation
mcp2cli --graphql https://api.example.com/graphql \
  create-user --name "Alice" --email "alice@example.com" --role "admin"

# Override auto-generated field selection
mcp2cli --graphql https://api.example.com/graphql \
  users --fields "id name email createdAt"

# With authentication
mcp2cli --graphql https://api.example.com/graphql \
  --auth-header "Authorization:Bearer env:GRAPHQL_TOKEN" \
  list-repositories --owner "knowsuchagency"

OAuth Authentication

mcp2cli supports OAuth flows across all modes with automatic token caching and refresh.

# Authorization code + PKCE (opens browser)
mcp2cli --mcp https://mcp.example.com/sse --oauth --list

# Client credentials flow (machine-to-machine)
mcp2cli --spec https://api.example.com/openapi.json \
  --oauth-client-id "env:OAUTH_CLIENT_ID" \
  --oauth-client-secret "env:OAUTH_CLIENT_SECRET" \
  list-resources

# With specific scopes
mcp2cli --graphql https://api.example.com/graphql \
  --oauth --oauth-scope "read:users write:posts" \
  create-post --title "Hello" --body "World"

# Local spec with OAuth requires --base-url for discovery
mcp2cli --spec ./openapi.json \
  --base-url https://api.example.com \
  --oauth --list

Tokens are cached in ~/.cache/mcp2cli/oauth/ and automatically refreshed when expired.

Bake Mode — Named Configurations

Save connection settings and filters to avoid repeating flags.

# Create a baked configuration from OpenAPI spec
mcp2cli bake create petstore \
  --spec https://api.example.com/openapi.json \
  --auth-header "Authorization:Bearer env:PETSTORE_TOKEN" \
  --exclude "delete-*,update-*" \
  --methods GET,POST \
  --cache-ttl 7200

# Create from MCP stdio server
mcp2cli bake create github \
  --mcp-stdio "npx @modelcontextprotocol/server-github" \
  --env "GITHUB_TOKEN=env:GITHUB_TOKEN" \
  --include "search-*,list-*" \
  --exclude "delete-*"

# Create from MCP HTTP server with OAuth
mcp2cli bake create analytics \
  --mcp https://analytics.example.com/mcp \
  --oauth --oauth-client-id "env:ANALYTICS_CLIENT_ID" \
  --oauth-client-secret "env:ANALYTICS_CLIENT_SECRET"

# Use a baked tool with @ prefix
mcp2cli @petstore --list
mcp2cli @petstore list-pets --limit 10
mcp2cli @github search-repositories --query "rust mcp"
mcp2cli @analytics query-metrics --start-date "2026-01-01"

# Manage baked tools
mcp2cli bake list
mcp2cli bake show petstore
mcp2cli bake update petstore --cache-ttl 3600
mcp2cli bake remove petstore

# Install wrapper script to ~/.local/bin
mcp2cli bake install petstore
# Now you can run: petstore --list

# Install to custom directory
mcp2cli bake install github --dir ./scripts/

Configurations are stored in ~/.config/mcp2cli/baked.json.

Filtering and Discovery

# Include only specific tools (glob patterns)
mcp2cli bake create myapi \
  --spec ./spec.json \
  --include "list-*,get-*,search-*"

# Exclude specific tools
mcp2cli bake create myapi \
  --spec ./spec.json \
  --exclude "delete-*,admin-*"

# Filter by HTTP methods (OpenAPI only)
mcp2cli bake create readonly-api \
  --spec ./spec.json \
  --methods GET

# Search tools by name or description
mcp2cli @myapi --search "user"
mcp2cli --spec ./spec.json --search "database query"

Usage-Aware Tool Ranking

mcp2cli tracks tool usage locally and ranks output to reduce token costs.

# List tools sorted by call frequency (default when usage data exists)
mcp2cli @myapi --list

# Top 10 most-used tools, compact format (~20 tokens vs ~1400)
mcp2cli @myapi --list --top 10 --compact

# Sort by most recently used
mcp2cli @myapi --list --sort recent

# Alphabetical sort
mcp2cli @myapi --list --sort alpha

# Show full descriptions (unwrapped)
mcp2cli @myapi --list --verbose

Usage data is stored in ~/.cache/mcp2cli/usage.json.

Output Control

# Pretty-print JSON (auto-enabled for TTY)
mcp2cli @myapi list-records --pretty

# Raw response body (no JSON parsing)
mcp2cli @myapi get-binary-data --raw > output.bin

# Limit output to first N records
mcp2cli @myapi list-users --head 5

# TOON format — token-efficient encoding (40-60% fewer tokens)
mcp2cli @myapi list-tags --toon

# Pipe-friendly (compact JSON when not a TTY)
mcp2cli @myapi list-users | jq '.[] | select(.active == true) | .email'

Caching

Specs and MCP tool lists are cached with a 1-hour TTL by default.

# Force refresh (bypass cache)
mcp2cli --spec https://api.example.com/openapi.json --refresh --list

# Custom TTL in seconds
mcp2cli --mcp https://mcp.example.com/sse --cache-ttl 86400 --list

# Custom cache key
mcp2cli --spec https://api.example.com/openapi.json \
  --cache-key my-api-prod --list

# Override cache directory
MCP2CLI_CACHE_DIR=/tmp/my-cache mcp2cli --spec ./spec.json --list

Local file specs are never cached.

Secrets Management

Avoid passing secrets in CLI arguments (visible in process lists) using env: or file: prefixes.

# Read from environment variable
mcp2cli --mcp https://mcp.example.com/sse \
  --auth-header "Authorization:env:MCP_TOKEN" \
  --list

# Read from file
mcp2cli --mcp https://mcp.example.com/sse \
  --oauth-client-secret "file:/run/secrets/oauth_secret" \
  --oauth-client-id "env:OAUTH_CLIENT_ID" \
  --list

# Works with secret managers that inject env vars
vault kv get -field=token secret/mcp | \
  MCP_TOKEN=$(cat) mcp2cli --mcp https://mcp.example.com/sse \
  --auth-header "Authorization:env:MCP_TOKEN" \
  search --query "logs"

Common Patterns

Create a Project-Specific CLI Wrapper

#!/usr/bin/env python3
"""Project-specific API wrapper using mcp2cli."""

import subprocess
import sys
import os

def main():
    """Run mcp2cli with project defaults."""
    cmd = [
        "mcp2cli",
        "--spec", "https://api.myproject.com/openapi.json",
        "--auth-header", f"Authorization:Bearer {os.environ['PROJECT_API_KEY']}",
        "--cache-ttl", "3600",
        *sys.argv[1:]
    ]
    subprocess.run(cmd, check=True)

if __name__ == "__main__":
    main()

Batch Operations with Shell

#!/bin/bash
# Batch create resources from CSV

while IFS=, read -r name email role; do
  echo "Creating user: $name"
  mcp2cli @myapi create-user \
    --name "$name" \
    --email "$email" \
    --role "$role"
done < users.csv

Combine with jq for Processing

# Extract specific fields and transform
mcp2cli @analytics get-metrics --period "2026-05" | \
  jq '[.data[] | {date: .timestamp, value: .metric_value}]' | \
  mcp2cli @warehouse import-timeseries --stdin

Create a Skill from an API

If you're building AI agent skills for APIs, use mcp2cli to generate the skill:

# Generate a skill from an OpenAPI spec
mcp2cli --spec https://api.example.com/openapi.json --list > API_COMMANDS.txt

# Use the command list as reference in your SKILL.md
# Document the baked configuration approach for the skill

Configuration Locations

  • Baked configs: ~/.config/mcp2cli/baked.json (override with MCP2CLI_CONFIG_DIR)
  • Cache: ~/.cache/mcp2cli/ (override with MCP2CLI_CACHE_DIR)
  • OAuth tokens: ~/.cache/mcp2cli/oauth/
  • Usage tracking: ~/.cache/mcp2cli/usage.json

Troubleshooting

OAuth flow fails to open browser

# Ensure DISPLAY is set (Linux) or browser is available
echo $DISPLAY

# Use client credentials flow instead if available
mcp2cli --spec ./spec.json \
  --oauth-client-id "env:CLIENT_ID" \
  --oauth-client-secret "env:CLIENT_SECRET" \
  --list

MCP transport auto-detection issues

# Force specific transport
mcp2cli --mcp https://mcp.example.com/sse --transport sse --list

# Try streamable HTTP if SSE fails
mcp2cli --mcp https://mcp.example.com/sse --transport streamable --list

Cache not refreshing

# Force refresh
mcp2cli @myapi --refresh --list

# Clear cache manually
rm -rf ~/.cache/mcp2cli/

# Check cache TTL
mcp2cli bake show myapi  # inspect cache-ttl setting

Secrets not loading from environment

# Verify environment variable is set
echo $MY_API_KEY

# Use explicit env: prefix
mcp2cli --mcp https://mcp.example.com/sse \
  --auth-header "x-api-key:env:MY_API_KEY" \
  --list

# Debug with verbose output (if added in future versions)
# Currently: check that env var name matches exactly

OpenAPI spec with relative servers

# Provide explicit base URL
mcp2cli --spec ./openapi.json --base-url https://api.example.com --list

GraphQL field selection too shallow

# Override auto-generated fields
mcp2cli --graphql https://api.example.com/graphql \
  users --fields "id name email profile { avatar bio } posts { id title }"

Large response truncation

# Limit output size
mcp2cli @myapi list-all-records --head 100 > sample.json

# Use raw mode for binary data
mcp2cli @myapi download-file --file-id 123 --raw > output.pdf

Token Efficiency Analysis

When using mcp2cli with AI agents:

  • Traditional approach: Send full tool schemas every turn (~1,400 tokens for 96 tools)
  • mcp2cli approach: --list --compact (~20 tokens for tool names)
  • Savings: 96-99% reduction in tool schema tokens
  • Use --top N to further reduce token costs by showing only frequently-used tools
  • Use --toon output format for 40-60% smaller responses on large uniform arrays

Development Integration

# Use mcp2cli programmatically via subprocess
import subprocess
import json

def call_api(tool: str, **kwargs):
    """Call mcp2cli and parse JSON response."""
    args = ["mcp2cli", "@myapi", tool]
    for k, v in kwargs.items():
        args.extend([f"--{k.replace('_', '-')}", str(v)])
    
    result = subprocess.run(args, capture_output=True, text=True, check=True)
    return json.loads(result.stdout)

# Use it
users = call_api("list-users", status="active", limit=50)
for user in users:
    print(f"{user['name']} <{user['email']}>")

Real-World Example: GitHub MCP Server

# Set up GitHub MCP server as a baked tool
mcp2cli bake create github \
  --mcp-stdio "npx @modelcontextprotocol/server-github" \
  --env "GITHUB_TOKEN=env:GITHUB_TOKEN"

# List all available GitHub operations
mcp2cli @github --list --compact

# Search repositories
mcp2cli @github search-repositories --query "mcp server" --limit 20 --pretty

# Get repository details
mcp2cli @github get-repository --owner "knowsuchagency" --repo "mcp2cli"

# Create an issue
mcp2cli @github create-issue \
  --owner "myorg" \
  --repo "myrepo" \
  --title "Feature request" \
  --body "Add support for X"

# Install as a standalone command
mcp2cli bake install github
# Now run: github search-repositories --query "rust"

For complete documentation and token savings analysis, see the full writeup.

FAQ

What does the mcp2cli-runtime-api-tooling skill do?

Turn MCP, OpenAPI, or GraphQL servers into CLIs at runtime with zero codegen, saving 96-99% of tokens on tool schemas

How do I install mcp2cli-runtime-api-tooling?

Run npx skills add https://github.com/reason-machines/devtools-skills --skill mcp2cli-runtime-api-tooling in your project. The skills CLI detects the agents you use and copies the skill into the right folder. For Claude Code you can also place the skill folder at ~/.claude/skills/mcp2cli-runtime-api-tooling/ (personal) or .claude/skills/mcp2cli-runtime-api-tooling/ (project).

Which AI agents can use mcp2cli-runtime-api-tooling?

Any agent that supports the open Agent Skills format (a folder with a SKILL.md file), including Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, Windsurf, Cline, OpenCode and more.

Is mcp2cli-runtime-api-tooling safe to install?

No partner security audit is available yet. Skills can run commands and scripts, so read the SKILL.md and any bundled files before installing.

Who maintains mcp2cli-runtime-api-tooling?

It is published by reason-machines in reason-machines/devtools-skills. It has 223 recorded installs.

More from reason-machines/devtools-skills

View all →

cc-switch-cli

reason-machines/devtools-skills

Open the skill to read its SKILL.md instructions.

315 installs

wx-cli-wechat-local-data

reason-machines/devtools-skills

Open the skill to read its SKILL.md instructions.

284 installs

chrome-devtools-mcp-automation

reason-machines/devtools-skills

Open the skill to read its SKILL.md instructions.

AI & Agents263 installs

officecli-office-automation

reason-machines/devtools-skills

Open the skill to read its SKILL.md instructions.

Productivity & Automation260 installs

autocli-web-scraping

reason-machines/devtools-skills

Open the skill to read its SKILL.md instructions.

Productivity & Automation257 installs

cli-anything-agent-native-software

reason-machines/devtools-skills

Open the skill to read its SKILL.md instructions.

257 installs

Popular backend & apis skills

View category →

find-skills

vercel-labs/skills

Helps users discover and install agent skills when they ask questions like "how do I do X", "find a skill for X", "is there a skill that can...", or express interest in extending capabilities. This skill should be used when the user is looking for functionality that might exist as an installable skill.

OfficialBackend & APIs4.1M installs

entra-app-registration

microsoft/azure-skills

Guides Microsoft Entra ID app registration, OAuth 2.0 authentication, and MSAL integration. USE FOR: create app registration, register Azure AD app, configure OAuth, set up authentication, add API permissions, generate service principal, MSAL example, console app auth, Entra ID setup, Azure AD authentication. DO NOT USE FOR: Key Vault secrets (use azure-keyvault-expiration-audit), general Azure resource security guidance.

OfficialBackend & APIs566.7K installs

google-agents-cli-adk-code

google/agents-cli

This skill should be used when the user wants to "write agent code", "build an agent with ADK", "add a tool", "create a callback", "define an agent", "use state management" — in a project that needs ADK (Agent Development Kit) API patterns and code examples. It provides a quick reference for agent types, tool definitions, orchestration patterns, callbacks, state management, the graph Workflow API, and reference recipes to study. Do NOT use for scaffolding (use google-agents-cli-scaffold) or deployment (use google-agents-cli-deploy).

Backend & APIs514.8K installs

embedded-captions

heygen-com/hyperframes

Add captions or subtitles to an existing single-subject talking-head video without editing the footage. Use for plain verbatim captions, cinematic captions embedded behind the subject, VFX captions, “炸/特效/酷炫字幕,” or a named identity from the 35-style catalog. Route by visual identity, not by backend engine. The quiet anchor rail is the default; embed every word only when the user explicitly wants a fully cinematic treatment. The workflow runs locally end to end, including transcription and subject matting; split multi-shot footage before applying it.

Backend & APIs453.3K installs

entra-agent-id

microsoft/azure-skills

Provision Microsoft Entra Agent Identity Blueprints, BlueprintPrincipals, and per-instance Agent Identities via Microsoft Graph, and configure OAuth 2.0 token exchange (fmipath, OBO, cross-tenant) including the Microsoft Entra SDK for AgentID sidecar. USE FOR: Agent Identity Blueprint, BlueprintPrincipal, agent OAuth, fmipath token exchange, agent OBO, Workload Identity Federation for agents, polyglot agent auth, Microsoft.Identity.Web.AgentIdentities. DO NOT USE FOR: standard Entra app registration (use entra-app-registration), Microsoft Foundry agent authoring (use microsoft-foundry).

OfficialBackend & APIs336.6K installs

neon

neondatabase/agent-skills

Overview of Neon, a complete set of cloud backend primitives around Lakebase Postgres: Auth, Object Storage, Functions, the AI Gateway, and Realtime. Start here to choose Neon for undecided login, files, APIs, and LLM calls, set up the CLI or MCP server, and follow the branch-first workflow. Use when building an app or backend on Neon, or when "Neon" or "Lakebase Postgres" is mentioned. Child skill neon-postgres wins for an existing DATABASEURL, SQL, schema, inspect, or search. Child skill neon-auth wins for login, users, sessions, identity routing, and Managed Better Auth setup. Child skill neon-realtime wins for live queries in reactive apps and agents. Also use for object storage, S3, buckets, serverless functions, function triggers, cron, AI gateway, LLM calls, realtime, logs, Loki, Grafana, observability, postgres, database, backend, Claimable Neon, neon.new, or a no-signup database.

Backend & APIs237.1K installs