skills
AGENTS.md1 starsChanged 42 days ago0BSDView on GitHub
Multi-module Go repo of stdio MCP servers. Remote: git@github.com:Quad4-Software/ai.git, branch master. Module prefix: github.com/Quad4-Software/ai/. From repo root: Other root targets: go-fix (modernizers), install, mcp-config and server-json (client config generation via scripts/gen-configs.py), inspector (scripts/mcp-inspector.py), links (scripts/link-check.py), release (GoReleaser plumbing, see .agents/skills/release). Per server: cd mcp/<name> && make test, or go test ./.... - Stdlib-first: no third-party dependencies unless vendored or justified. - Read-only by default: servers never mutate host state. Exception: meshchatx ships opt-in GitHub issue tools (create, update, close, comment,…
- Reads credentials
Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.
Of the AGENTS.md files we read, the 31,542 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.
In this repository75 other files it gives its agents.
- agentsSkill
- android-connectivitySkill
- android-devSkill
- android-mediaSkill
- android-performanceSkill
- anti-hallucinationSkill
- blue-teamingSkill
- bug-huntingSkill
- cdnSkill
- ci-securitySkill
- contextSkill
- coolifySkill
- dep-versionsSkill
- devopsSkill
- devsecopsSkill
- dockerSkill
- electronSkill
- firecrackerSkill
- gatewaySkill
- ghSkill
- goSkill
- i18nSkill
- k3sSkill
- lxmfSkill
- lxmfySkill
- mcp-toolkitSkill
- media-stackSkill
- mediawikiSkill
- memorySkill
- meshchatxSkill
- mesh-vpnSkill
- micronSkill
- microservices-haSkill
- nebulaSkill
- no-slopSkill
- no-trickerySkill
- opensubsonicSkill
- osintSkill
- owaspSkill
- pnpmSkill
- podmanSkill
- project-scaffoldSkill
- python-packagingSkill
- qaSkill
- releaseSkill
- research-methodsSkill
- reticulum-goSkill
- reticulumSkill
- rngitSkill
- rnsSkill
- rustSkill
- sastSkill
- shipped-textSkill
- stylometrySkill
- surrealdb-cliSkill
- surrealdb-docsSkill
- surrealdb-jsSkill
- surrealdb-pythonSkill
- surrealdbSkill
- surrealdb-vectorSkill
- The first 60 shown.
The file
We split this AGENTS.md into the rules it gives an agent
6 rules. About 0 are common, 3 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 28
Read-only by default: servers never mutate host state.
- Very specific: line 38
Keep main.go thin, domain logic lives in internal/ packages.
- Distinctive: line 68
Keep it stdlib-only, read-only, offline-capable.
- Distinctive: line 70
Add README.md and follow SECURITY.md rules.
- Distinctive: line 79
All actions pinned to full SHAs.
- Very specific: line 103
Use --skill <name> to install only one, -g for global installation, or --list to preview.
≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
AGENTS.md
Multi-module Go repo of stdio MCP servers. Remote: git@github.com:Quad4-Software/ai.git, branch master. Module prefix: github.com/Quad4-Software/ai/.
Commands
From repo root:
make all # fmt + go-fix + vet + test + build for every mcp/* dir
make gosec # gosec security scan per server
make golangci # golangci-lint run ./... per server
make build test vet fmt cleanOther root targets: go-fix (modernizers), install, mcp-config and server-json (client config generation via scripts/gen-configs.py), inspector (scripts/mcp-inspector.py), links (scripts/link-check.py), release (GoReleaser plumbing, see .agents/skills/release).
Per server: cd mcp/<name> && make test, or go test ./....
Conventions
- Stdlib-first: no third-party dependencies unless vendored or justified.
- Read-only by default: servers never mutate host state. Exception: meshchatx ships opt-in GitHub issue tools (create, update, close, comment, search) that only activate when a token env var is set (MESHCHATX_GITHUB_TOKEN, GITHUB_TOKEN, GH_TOKEN). Tokens are never returned in tool output and the target repo is owner/name validated.
- Path jailing: all file access resolved under an allowed root, traversal rejected.
- Secrets: never return key material or config secrets, redact them.
- #nosec annotations require an inline justification comment.
- Offline builds and tests: no network fetches in build or test paths.
- Keep main.go thin, domain logic lives in internal/ packages.
Toolchain
Go 1.27 with a custom toolchain go1.27.1-X:nodwarf5. golangci-lint is incompatible with this toolchain (it fails on the custom version string / nodwarf5 build), so make golangci may not work locally. Rely on make vet, make gosec, and CI instead. Do not switch go.mod toolchain lines to fix it.
Vendored Micron parser
micron vendors micron-parser-go under mcp/micron/third_party/micron-parser-go via:
require micron-parser-go v1.1.4
replace micron-parser-go => ./third_party/micron-parser-goTo update: replace the vendored tree, bump the require version to match, keep the replace line, run go mod tidy and go test ./... inside micron. Never go get the parser from the network. Lint targets exclude /third_party/. Read the vendored CHANGELOG.md when bumping (NomadNet 1.4.0 fold headings landed in v1.1.4).
Adding an MCP server
- Copy mcp/scaffold/ to mcp/
<name>/. - Set module to github.com/Quad4-Software/ai/mcp/
<name>in its go.mod. - Register tools in main.go via mcp.NewServer(name, version, tools, nil).
- Keep it stdlib-only, read-only, offline-capable.
- The root Makefile discovers it automatically (wildcard mcp/*, excluding scaffold).
- Add README.md and follow SECURITY.md rules.
CI
Workflows in .github/workflows/: ci.yml, dependency-review.yml, gosec.yml, race.yml, fuzz.yml, bench.yml, leak.yml, release.yml, scorecard.yml, badges.yml, docker.yml, links.yml, mcp-inspector.yml, zizmor.yml (zizmor static analysis of the workflow files themselves, gates on findings; local equivalent: make zizmor). All actions pinned to full SHAs. Every job begins with step-security/harden-runner (pinned). No arbitrary shell on untrusted input. Dependabot handles gomod weekly.
See .agents/skills/ for detailed guides: mcp-toolkit, reticulum, lxmfy, micron, ci-security, release, nebula. Style skills: no-slop (prose linting), shipped-text (docs, comments, and debug text must describe the software only, never the conversation or design debate that produced it). Documentation-only reference skills (no MCP servers) cover self-hosted infrastructure: media-stack, docker, mesh-vpn, xmpp, coolify, opensubsonic. App platform skills: wails, android-connectivity, android-dev, android-media, android-performance. Language/toolchain skills: go, typescript, rust (edition 2024 reference).
Agent Skills distribution
The .agents/skills/ directory follows the Agent Skills specification. Once
this repo is pushed to GitHub, anyone can install the skills with:
npx skills add Quad4-Software/aiUse --skill <name> to install only one, -g for global installation, or
--list to preview. The skills.sh.json at the repo root lists the skill
groups for the skills.sh marketplace.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

