v2.1.3

Main change here are improved docstrings on some tools. I've codified the standards for docstrings so all tools from now on should be compliant and automated code reviews will hold them to that standard.

Also some plumbing work in case I resurrect the homebrew tap.

What's Changed

Changes

• Point the (paused) Homebrew channel at L337-org/homebrew-tap by @GavinLucas in #140
• Use canonical L337-org casing for the homebrew-tap repo path by @GavinLucas in #141
• Point the Glama badge and listing link at the new L337-org listing by @GavinLucas in #142
• Codify the Glama TDQS docstring standard and lift the 20 lowest-scoring tools by @GavinLucas in #143

Full Changelog: v2.1.2...v2.1.3

v2.1.2

First release from our new organisational home - L337.org

What's Changed

Dependencies

• Build(deps): Bump the actions group with 4 updates by @dependabot[bot] in #135

Changes

• Sweep repository identity to L337-org ahead of the org transfer by @GavinLucas in #136
• 2.1.2: first post-transfer release; narrow stack_ps test retry by @GavinLucas in #138

Full Changelog: v2.1.1...v2.1.2

v2.1.1

Patch release with one small fix.

Found an issue where, when using the ssh transport, an alternate port specified in .ssh/config was not being picked up. It's actually an issue in docker_py that needed a bit of a work around, but with that in place it's now working correctly.

What's Changed

Changes

• Fix ssh:// daemons on a non-default SSH port by @GavinLucas in #134

Full Changelog: v2.1.0...v2.1.1

V2.1.0

Couple of larger changes here.

One is a tool for looking up documentation since not all models will have access to resources in a way that is useful. Tools with slightly complicated parameters have instructions in the docstrings to prompt them to look up the docs using this tool so hopefully this will increase the usability of the tools and reduce hallucination.

Speaking of resources, I've also added some resources for swarm.

I've also made some improvements and additions to enable the model to wait for specific events, e.g. when a given pattern appears in a container log, e.g. indicating that a service is actually up and running. That should be faster and more robust than constantly polling the logs. The added/improved wait tools include service_wait, node_wait, container_wait, and registry_tag_wait.

What's Changed

Changes

• 2.1.0: swarm observability resources + wait tools by @GavinLucas in #132
• Add docs_lookup: tool-callable mirror of docker-docs:// resources by @GavinLucas in #133

Full Changelog: v2.0.1...v2.1.0

V2.0.1

Main change in this release is a pass of improvements for some of the tool docstrings.

Split the 'contributing' section out of the README into it's own file and made other documentation improvements for the benefit of anyone who wishes to help contribute.

What's Changed

Dependencies

• Build(deps): Bump the actions group with 7 updates by @dependabot[bot] in #128

Changes

• Improve docstrings for 17 low-scoring tools (Glama quality, round 3) by @GavinLucas in #129
• Split the Contributing section out into CONTRIBUTING.md by @GavinLucas in #130
• Pin canary's uv version; bump to 2.0.1 by @GavinLucas in #131

Full Changelog: v2.0.0...v2.0.1

v2.0.0

Although the "what's changed" section might look small, this is a major revision.

Quite a lot of work has gone into ensuring that the tools are as consistent as possible. This means consistency in tool naming, in parameter naming, filters, etc.

This has also allowed us to do a bit of reconciling tools and deduplication so the tool count has gone down slightly but without sacrificing functionality or too discoverability.

This does mean that many tools have been renamed which may have an impact if you've customised tool approval settings. For reference we've created a document detailing the modified tools. See MIGRATION-2.0.md for more information.

The tool count is still quite high, but that works very well with clients that do lazy or deferred tool loading. I'm not planning to combine tools in order to reduce the tool count further, although that may be of benefit to clients that don't lazy load yet, at the cost of tool discoverability and clarity. Lazy loading is the way MCP clients are heading so I think this is the correct approach.

What's Changed

Changes

• Silence setup-uv warnings in the canary install-smoke job by @GavinLucas in #118
• Run the canary's repo-resolution leg on PRs into main by @GavinLucas in #125
• docker-mcp-server 2.0: tool-surface coherence overhaul by @GavinLucas in #127

Full Changelog: v1.9.3...v2.0.0

v1.9.3

Mainly an internal release here although it also updates dependencies.

Lots of work went into trying to introduce a homebrew distribution channel, but issue remained with the way that homebrew builds some python dependencies that we couldn't easily work around. So that work is now on hold.

Also some work done to improve workflows and CI checks which makes the project very robust.

Also added the AAA rating Glama quality score badge to the README.

What's Changed

Dependencies

• Build(deps): Bump actions/setup-python from 6.2.0 to 6.3.0 in the actions group by @dependabot[bot] in #110
• Build(deps): Bump the uv group with 5 updates by @dependabot[bot] in #111

Changes

• Add Glama quality score badge to README by @GavinLucas in #98
• Fix manifest.json long_description newlines; add packages/listings table to README by @GavinLucas in #100
• Add Homebrew tap distribution channel by @GavinLucas in #101
• Ensure latest uv is installed for uv pip download support by @GavinLucas in #102
• Replace uv pip download with pip3 download for wheelhouse steps by @GavinLucas in #103
• Switch to macos-latest runner to fix arm64 wheel resolution by @GavinLucas in #104
• Use Python 3.14 interpreter for pip downloads to fix abi3 tag resolution by @GavinLucas in #105
• Drop uv Python 3.14 install; use runner pip3 with explicit platform flags by @GavinLucas in #106
• Restore scripts/ from main after tag checkout so backfill dispatches work by @GavinLucas in #107
• Add skip_clean libexec to suppress dylib linkage fixer errors by @GavinLucas in #108
• Pause Homebrew channel; remove from README, disable release trigger by @GavinLucas in #109
• Add CI guard for the Intel-macOS cryptography pin by @GavinLucas in #113
• Add premerge quality gates; single-source ruff; enforce lockfile consistency by @GavinLucas in #114
• Consolidate release publishing into one ordered pipeline by @GavinLucas in #115
• Add --version flag and weekly platform canary by @GavinLucas in #116
• Bump version to 1.9.3 by @GavinLucas in #117

Full Changelog: v1.9.2...v1.9.3

v1.9.2

Got some great tool quality feedback from Glama (glama.ai) which I used to improve some of the tool descriptions.

Many thanks to the folks who manage Glama!

What's Changed

• Improve docstrings for 8 low-scoring tools (Glama quality) by @GavinLucas in #97

Full Changelog: v1.9.1...v1.9.2

v1.9.1

Minor patch release.

The most visible change is that the mcpb bundle description is much more readable.

The rest are more internal changes, e.g. improving the reliability of some of the integration tests and adding a helpful packaging script for developers.

What's Changed

• Add newlines to the mcpb description for readability by @GavinLucas in #92
• Skip registry integration tests on a 429 rate-limit by @GavinLucas in #94
• Add scripts/build-mcpb.sh dev helper for local .mcpb bundles by @GavinLucas in #93
• Authenticate registry integration tests to Docker Hub by @GavinLucas in #95
• Add glama.json for Glama MCP server listing by @GavinLucas in #96

Full Changelog: v1.9.0...v1.9.1

v1.9.0

Quite a significant change in this release. You're now able to manage multiple docker hosts, both local and remote (TCP, TLS or SSH) from the same plugin. You can also flag a host as read-only which gives you lots of powerful tools to manage your deployments.

This means that you can do things like manage your local development environment, a read-write staging environment, and a read-only production environment all from one agent. You can ask questions like 'why did my container crash in staging?', 'am I suffering memory pressure in production?', 'what differences are there between my staging and production environments?', 'give me an overview of the status of all my environments'.

It will still basically be plug and play with zero configuration for simple environments where you're just managing the local docker daemon, but also hopefully a game changer for administrators of production environments.

What's Changed

• Harden publish-registry against the image-build race by @GavinLucas in #75
• Build(deps): Bump actions/checkout from 6.0.3 to 7.0.0 in the actions group by @dependabot[bot] in #76
• Build(deps): Bump pydantic-settings from 2.14.0 to 2.14.2 by @dependabot[bot] in #77
• Add PR review instructions to copilot-instructions by @GavinLucas in #78
• Minor changes to README.md by @GavinLucas in #79
• Multi-host daemon support (DOCKER_MCP_SERVER_HOSTS) by @GavinLucas in #91

Full Changelog: v1.8.1...v1.9.0