Easymcp is a fast efficient docker image to turn any rest api into a MCP server, AI ready.
480
Turn any OpenAPI service into MCP tools your AI agents can actually use.
EasyMCP is a Docker-packaged OpenAPI-to-MCP runtime with an integrated easymcp CLI for setup, lifecycle management, tool discovery, and agent installation. The image runs the server. The CLI makes it usable by developers and platform teams.
docker pull ab0tcom/easymcp:latest
Most internal APIs already have OpenAPI specs. Most AI agents speak MCP. EasyMCP connects those two worlds without requiring every service team to hand-write an MCP server.
Use it to:
EasyMCP has two parts that are designed to work together:
The Docker image runs a configured MCP server:
easymcp CLIThe CLI is integral to the workflow. It is the control surface for real usage:
Install the CLI from the public GitHub installer:
curl -fsSL https://raw.githubusercontent.com/ab0t-com/easymcp/main/cli/install.sh | bash
Pinned install:
curl -fsSL https://raw.githubusercontent.com/ab0t-com/easymcp/main/cli/install.sh | EASYMCP_VERSION=v0.1.0 bash
Dry run:
curl -fsSL https://raw.githubusercontent.com/ab0t-com/easymcp/main/cli/install.sh | EASYMCP_DRY_RUN=1 bash
easymcp create auth-service \
--openapi https://auth.service.ab0t.com/openapi.json \
--port 8091
easymcp start auth-service
easymcp check auth-service
easymcp discover refresh auth-service
easymcp find "create a user token"
easymcp agent install codex auth-service
That creates a Docker-backed EasyMCP instance, validates the MCP endpoint, indexes its tools, and installs it into an agent config.
You can also run the image directly when you already have a config file.
Example auth-service.yaml:
version: "1.0"
server:
name: auth-service
description: Auth service MCP server
openapi:
url: https://auth.service.ab0t.com/openapi.json
api_auth:
type: none
mcp_auth:
enabled: false
type: none
transport:
type: http
host: 0.0.0.0
port: 8000
Run:
docker run --rm \
--name easymcp-auth-service \
-p 8000:8000 \
-v "$PWD/auth-service.yaml:/app/config.yaml:ro" \
ab0tcom/easymcp:v0.1.0 \
/app/config.yaml
MCP endpoint:
http://localhost:8000/mcp
Health endpoint:
http://localhost:8000/health
EasyMCP is designed so production secrets do not need to be baked into images or committed into config files.
For upstream APIs that require bearer tokens, store only the environment variable name:
api_auth:
type: bearer
token_env: EASYMCP_AUTH_SERVICE_API_TOKEN
Run with the secret supplied by the environment:
docker run --rm \
-p 8000:8000 \
-e EASYMCP_AUTH_SERVICE_API_TOKEN \
-v "$PWD/auth-service.yaml:/app/config.yaml:ro" \
ab0tcom/easymcp:v0.1.0 \
/app/config.yaml
For enterprise workflows, use the CLI profile system to keep customer, tenant, environment, and credential references explicit:
easymcp profile create acme-prod --customer acme --environment prod
easymcp profile bind acme-prod auth-service
easymcp profile credential add-env acme-prod mcp_access_token EASYMCP_ACME_MCP_TOKEN --required
easymcp profile doctor acme-prod
v0.1.0: first public pinned release.latest: most recent published image; use intentionally.For production and shared environments, prefer immutable tags such as v0.1.0.
EasyMCP is more than a single Docker image. The CLI is designed to act like a local MCP hub for operators who need to manage many MCP servers across services, customers, environments, and agents.
Register EasyMCP-generated servers and generic MCP servers side by side:
easymcp create payment-service \
--openapi https://payment.example.com/openapi.json \
--group finance \
--port 8092
easymcp instance add external-search \
--kind remote_http \
--transport http \
--url https://search.example.com/mcp \
--group productivity \
--auth-mode none
easymcp ps
easymcp group ls
Use groups to operate on related MCP instances:
easymcp group start finance
easymcp group stop finance
easymcp group disable finance
easymcp group enable finance
After MCP servers are running, refresh the discovery cache:
easymcp discover refresh --all
easymcp discover ls
easymcp find "create a payment plan"
easymcp discover inspect create_payment_intent_payments
This helps humans and agents find the right tool by intent instead of memorizing tool names.
Profiles are optional, but useful when the same workstation manages multiple customers, tenants, or environments.
easymcp profile create acme-prod \
--customer acme \
--environment prod \
--display-name "Acme Production"
easymcp profile bind acme-prod payment-service
easymcp profile credential add-env acme-prod mcp_access_token EASYMCP_ACME_MCP_TOKEN \
--purpose mcp_client_bearer \
--required
easymcp profile doctor acme-prod
Profiles do not silently change global behavior. Use them explicitly:
easymcp ps --profile acme-prod
easymcp find "refund a customer" --profile acme-prod
easymcp agent install codex payment-service --profile acme-prod
This keeps account switching visible and reduces the risk of sending data to the wrong tenant or using the wrong credentials.
Tenant metadata records how a profile selects tenant context without storing the tenant secret value.
easymcp profile credential add-env acme-prod tenant_id EASYMCP_ACME_TENANT_ID \
--purpose tenant_selector \
--required
easymcp profile tenant set acme-prod \
--mode header \
--value-ref tenant_id \
--header-name X-Tenant-ID \
--expected "Acme Production"
For high-risk production access, prefer separate instances per tenant or customer. Tenant metadata is useful for visibility and verification, but separate instances are easier to audit.
Render before installing when you want to inspect exactly what will be written:
easymcp agent render codex payment-service --profile acme-prod
easymcp agent render claude-code payment-service --profile acme-prod --scope project
Install when ready:
easymcp agent install codex payment-service --profile acme-prod
easymcp agent install claude-code payment-service --profile acme-prod --scope project
The CLI writes agent configs using environment variable references where possible. It does not write raw profile secrets into agent config.
The CLI keeps its owned state under ~/.easymcp:
~/.easymcp/
instances.yaml
profiles.json
audit.jsonl
configs/
runtime/
cache/
Profile mutations and profile-aware agent installs are recorded in audit.jsonl. The audit log stores action metadata, profile names, instance names, and changed field names, not raw credential values.
https://github.com/ab0t-com/easymcphttps://raw.githubusercontent.com/ab0t-com/easymcp/main/cli/install.shhttps://hub.docker.com/r/ab0tcom/easymcpContent type
Image
Digest
sha256:7a16ddd7f…
Size
69 MB
Last updated
4 months ago
docker pull ab0tcom/easymcp:v0.5.3