Harbor
Read-only MCP server for Harbor container registries: projects, scans, robots, replication
- Pricing
- Open source
- Platforms
- CLI, self-hosted
- Runs
- Locally
- Version
- 0.2.0
- Listing updated
- 3 days ago
About Harbor
Harbor is an open-source MCP server for developer tools. It runs locally on your machine and is started by your AI client.
This is an independent community project, not affiliated with the Harbor project or the CNCF. It connects an assistant to a running Harbor instance and answers questions about quotas, artifacts, scan results, access and replication health, using fourteen tools that only read.
It authenticates with a Harbor user or robot account, and the README recommends a read-only system robot, with a ready-made request body for creating one. Harbor credentials, robot secrets and webhook authorization headers are kept out of tool results, logs and errors. No tool deletes anything; write tools such as starting a scan or replication are described as coming later and only with explicit opt-in.
A distinctive part is support for agent skills stored in Harbor as OCI artifacts: the server can list and show them, and when HARBOR_MCP_SKILLS_DIR is set it can install one locally, accepting signed skills only by default. It runs over stdio for desktop clients, or streamable HTTP with a bearer token inside a cluster, and ships as a signed multi-arch image.
Install
Run in a terminal
claude mcp add harbor-mcp --env HARBOR_URL='<HARBOR_URL>' --env HARBOR_USERNAME='<HARBOR_USERNAME>' --env HARBOR_PASSWORD='<HARBOR_PASSWORD>' -- docker run -i --rm -e HARBOR_URL -e HARBOR_USERNAME -e HARBOR_PASSWORD ghcr.io/oreondev/harbor-mcp:0.2.0Add to claude_desktop_config.json
{
"mcpServers": {
"harbor-mcp": {
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"-e",
"HARBOR_URL",
"-e",
"HARBOR_USERNAME",
"-e",
"HARBOR_PASSWORD",
"ghcr.io/oreondev/harbor-mcp:0.2.0"
],
"env": {
"HARBOR_URL": "<HARBOR_URL>",
"HARBOR_USERNAME": "<HARBOR_USERNAME>",
"HARBOR_PASSWORD": "<HARBOR_PASSWORD>"
}
}
}
}Add to ~/.cursor/mcp.json
{
"mcpServers": {
"harbor-mcp": {
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"-e",
"HARBOR_URL",
"-e",
"HARBOR_USERNAME",
"-e",
"HARBOR_PASSWORD",
"ghcr.io/oreondev/harbor-mcp:0.2.0"
],
"env": {
"HARBOR_URL": "<HARBOR_URL>",
"HARBOR_USERNAME": "<HARBOR_USERNAME>",
"HARBOR_PASSWORD": "<HARBOR_PASSWORD>"
}
}
}
}Add to .vscode/mcp.json
{
"servers": {
"harbor-mcp": {
"type": "stdio",
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"-e",
"HARBOR_URL",
"-e",
"HARBOR_USERNAME",
"-e",
"HARBOR_PASSWORD",
"ghcr.io/oreondev/harbor-mcp:0.2.0"
],
"env": {
"HARBOR_URL": "<HARBOR_URL>",
"HARBOR_USERNAME": "<HARBOR_USERNAME>",
"HARBOR_PASSWORD": "<HARBOR_PASSWORD>"
}
}
}
}Add to ~/.codeium/windsurf/mcp_config.json
{
"mcpServers": {
"harbor-mcp": {
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"-e",
"HARBOR_URL",
"-e",
"HARBOR_USERNAME",
"-e",
"HARBOR_PASSWORD",
"ghcr.io/oreondev/harbor-mcp:0.2.0"
],
"env": {
"HARBOR_URL": "<HARBOR_URL>",
"HARBOR_USERNAME": "<HARBOR_USERNAME>",
"HARBOR_PASSWORD": "<HARBOR_PASSWORD>"
}
}
}
}Add to your client's MCP settings (most use this shape)
{
"mcpServers": {
"harbor-mcp": {
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"-e",
"HARBOR_URL",
"-e",
"HARBOR_USERNAME",
"-e",
"HARBOR_PASSWORD",
"ghcr.io/oreondev/harbor-mcp:0.2.0"
],
"env": {
"HARBOR_URL": "<HARBOR_URL>",
"HARBOR_USERNAME": "<HARBOR_USERNAME>",
"HARBOR_PASSWORD": "<HARBOR_PASSWORD>"
}
}
}
}Configuration
HARBOR_URLHarbor base URL, for example https://harbor.example.comRequiredHARBOR_USERNAMEHarbor user or robot account; a read-only system robot is recommendedRequiredHARBOR_PASSWORDPassword or robot secretRequiredHARBOR_INSECURE_SKIP_VERIFYSet to true to skip TLS verification of HarborHARBOR_MCP_RATERequests per second to Harbor
Tools this server provides (15)
Taken from the project documentation; the running server may differ.
- harbor_system_info
- Harbor version, authentication mode and the acting user or robot.
- harbor_list_projects
- Lists projects.
- harbor_get_project
- Quota, CVE allowlist, retention and immutability of one project.
- harbor_list_repositories
- Lists repositories.
- harbor_list_artifacts
- Artifacts with tags, size, push and pull times and scan summary.
- harbor_get_vulnerabilities
- Vulnerabilities of one artifact, worst first.
- harbor_list_members
- Members with roles.
- harbor_list_robots
- Robot accounts with permissions and expiry.
- harbor_list_registries
- Replication endpoints.
- harbor_list_replications
- Replication policies with the last run.
- harbor_list_webhooks
- Webhook policies.
- harbor_get_gc
- Garbage collection schedule.
- harbor_list_skills
- Agent skills stored in a project as OCI artifacts.
- harbor_get_skill
- One skill's SKILL.md and files, digest and signature status.
- harbor_install_skill
- Installs a skill into HARBOR_MCP_SKILLS_DIR when that is set.
What it does
- Vulnerability reports
- Returns the vulnerabilities of one artifact, worst first.
- Project details
- Shows quota, CVE allowlist, retention and immutability settings.
- Access review
- Lists project members with roles and robot accounts with permissions and expiry.
- Replication status
- Lists registries and replication policies along with the last run.
- Agent skills
- Lists and reads skills stored in Harbor, with digest and signature status.
- Opt-in skill install
- Writes a skill into a local directory only when HARBOR_MCP_SKILLS_DIR is set.
- Secret hygiene
- Keeps credentials and webhook headers out of results, logs and errors.
Good for
- Ask which images in a project carry the worst vulnerabilities
- Audit which robot accounts exist and when they expire
- Check whether replication jobs ran successfully
- Pull agent skills from Harbor into a local skills folder
Pricing
Open source
Prices change. Check oreonit.com before you buy.
Harbor FAQ
Can it change anything in Harbor?
No. The README says the tools are read-only and none deletes anything; write tools may come later behind an explicit opt-in.
Which Harbor versions are tested?
The README lists tests against Harbor 2.15 on every pull request and 2.12 to 2.15 nightly.
What credentials does it need?
HARBOR_URL, HARBOR_USERNAME and HARBOR_PASSWORD; a read-only system robot is recommended.
How is it transported?
Over stdio for desktop clients, or streamable HTTP with a bearer token for running in a cluster.
Is Harbor free?
Yes. Harbor is open source, so you can use and self-host it for free.
How do I add Harbor to Claude Code?
Run this command in your terminal: claude mcp add harbor-mcp --env HARBOR_URL='<HARBOR_URL>' --env HARBOR_USERNAME='<HARBOR_USERNAME>' --env HARBOR_PASSWORD='<HARBOR_PASSWORD>' -- docker run -i --rm -e HARBOR_URL -e HARBOR_USERNAME -e HARBOR_PASSWORD ghcr.io/oreondev/harbor-mcp:0.2.0
What are the best alternatives to Harbor?
The closest alternatives to Harbor are REA, Codebase Memory, GitHub, Serena MCP: the IDE for your agent and XERJ.
Alternatives to Harbor

REA
MCP serverCLI and MCP server that lets coding agents reverse engineer apps and binaries
Runs locally- Open source
- 464K GitHub stars
- Developer tools

Codebase Memory
MCP serverLocal MCP server that builds a code knowledge graph for AI coding agents
Runs locally- Open source
- 48.7K GitHub stars
- Knowledge and memory

GitHub
MCP serverOfficial MCP server connecting AI assistants to GitHub repos, issues and workflows
Not checked yet- Remote
- Open source
- 34.5K GitHub stars
- Developer tools

Serena MCP: the IDE for your agent
MCP serverMCP toolkit giving coding agents symbol-level code retrieval and editing
Runs locally- Open source
- 31.5K GitHub stars
- Developer tools

XERJ
MCP serverLocal search engine for AI agents with cited answers over your code and docs
Runs locally- Open source
- 19.2K GitHub stars
- Knowledge and memory

Ouroboros
MCP serverSpec-first layer that pins requirements and verifies results of AI coding agent work
Runs locally- Open source
- 6.4K GitHub stars
- Developer tools