docker-mcp AGENTS.md

A repository instruction file for docker-mcp, a Python server that makes Docker functions available to assistants through tools.

In plain words
What is it for?
Use it when reviewing or changing the docker-mcp project, its instructions, commands, or project areas.
Why use it?
It tells contributors what to check first, including inaccurate documentation, ineffective safeguards, and outdated lists.

Instructions file for CodexOpenCode

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add instructions/l337-org/docker-mcp/agents-md
Clone the repo
git clone --depth 1 https://github.com/L337-org/docker-mcp

Made for: Codex, OpenCode.

Per session 7,432 This file is loaded in full into every session.
When invoked 7,432 The same file — it is already loaded in full.
Security scan A 1 finding. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.07432 $0.07432
Opus 5 $0.03716 $0.03716
Sonnet 5 $0.01486 $0.01486
Haiku 4.5 $0.00743 $0.00743

Measured 2d ago against content hash 7e784dca6ef0, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

docker-mcp AGENTS.md scanned grade A with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Runs shell commandslowCapability

Expected in a hook, worth knowing in a rule or an instructions file.

| `_cli.py`, `_ssh_proxy.py`, CLI-backed tools | [architecture/cli-shell-out.md](architecture/cli-shell-out.md) | `subprocess.run` called directly; `shell=True`; a missing `timeout=`; the remote-exec fallback preferred o
AGENTS.md · 394 lines

How it starts

The opening of the file, as written. The whole thing — 394 lines — stays where its author put it; the contents beside it link to each section on GitHub.

AGENTS.md

The shared instruction file for this repository. Every assistant reads this one; CLAUDE.md and .github/copilot-instructions.md are pointers to it.

docker-mcp is a Python MCP server (requires Python >=3.14) managed with uv that exposes the Docker SDK for Python as MCP tools. The entry point is the docker_mcp package, run with python -m docker_mcp or via the installed console script.

Review priorities

Ordered by where real defects in this project have actually come from. Spend review effort in this order.

  1. Claims in prose that nothing type-checks. Docstrings, docs and comments asserting a method, flag, route or identifier semantics that does not exist. Three scout tools shipped a --format flag the subcommand does not define; a docstring claimed buildx resolves --file against the build context when it resolves against the CWD. Verify against the primary source, not recollection.
  2. A guard that silently stops guarding. See the invariant list below. These pass CI, pass review at a glance, and fail in production or never fail loudly at all.
  3. Enumerations that drift. Any list of domains, tools, channels or files that is copied rather than derived. One list was wrong in six places across a single feature branch.
  4. Unbounded reads of anything external. Daemon streams, registry bodies, staged files, CLI output.
  5. Docstring quality on any touched @tool() - the ratchet in the checklist below.
  6. Everything else.

Area checks

Read the linked file before judging a substantive change to that area.

Area Read Watch for
server.py, docker_mcp/tools/ architecture/server.md _slim_schema or _apply_host_schema changing validation rather than display; a disabled capability registered-then-refusing instead of absent
_hosts.py, host selection architecture/hosts.md use_context=False dropped; system_reconnect gaining the ability to retarget an arbitrary URL; a write path that no longer requires an explicit host in multi-host mode
_cli.py, _ssh_proxy.py, CLI-backed tools architecture/cli-shell-out.md subprocess.run called directly; shell=True; a missing timeout=; the remote-exec fallback preferred over a usable local CLI; staging consequences absent from the docstring
any @tool() docstring architecture/tool-descriptions.md the checklist below
code calling docker architecture/docker-sdk.md an unverified method; a hand-built route that is not in the Engine API spec; a reach-in past the public SDK introduced without recorded sign-off
Dockerfile, manifest.json, server.json, release workflows architecture/distribution.md version drift across the four files; a registry ownership marker no longer matching server.json's name
skills/l337-docker/ architecture/agent-skill.md the skill described as equivalent to the server; tool-permission frontmatter added; a hand-edited figure in MCP_VS_SKILLS.md
.github/workflows/ architecture/ci.md a uses: naming a tag or branch; a new job without timeout-minutes where the workflow requires one

Read the full file on GitHub · 394 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 394 lines · 7,432 tokens per session scan A 7e784dca6ef0

Subscribe to this mod's changes

docker-mcp AGENTS.md is an instructions file published in the GitHub repository L337-org/docker-mcp (4 stars, last pushed 5d ago), licensed MIT. It adds 7,432 tokens to every session, about $0.0372 per session on Opus 5. A static security scan graded it A with 1 finding (runs shell commands). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.