What does this skill do, and when should you use it?
This skill ships inside the marmite repository and targets engineers who want to contribute to marmite's own codebase — not users building sites with marmite (that is the separate `marmite` skill). It codifies code quality rules (formatting, clippy, no unwrap/panic), a two-layer testing system (unit tests plus subprocess-based integration tests), backward-compatibility principles, documentation requirements, and a full contribution checklist. It also teaches agents to use marmite's built-in --show-urls, --init-site, and --new commands for development verification. The project is AGPL-3.0-or-later, Rust edition 2021, and uses the mask task runner.
Once read, it directs a developer or agent through: running mask fmt / mask check / mask pedantic for formatting and linting; running mask test across unit tests in src/tests/ and integration tests in tests/; adding config fields per the prescribed flow (serde default in config.rs + CLI flag in cli.rs + override wiring); running marmite --show-urls to compare URL output before and after refactors; scaffolding with marmite --init-site and marmite --new; keeping documentation in sync across example/content/ blog posts, example/ai/llms.txt, .agents/skills/marmite/references/ files, and draft release notes; and applying template/static changes to both the main theme and theme_template.
- A Rust developer preparing a PR to marmite who needs to know the project's formatting, clippy, testing, and documentation gates
- An AI agent tasked with implementing a new marmite feature that must update code, tests, and multiple docs in one pass
- A contributor fixing a bug who must add a reproducing test and update the draft release notes
- A developer adding a marmite.yaml config option who needs to also add the CLI flag, defaults, and reference docs
- A maintainer changing Tera templates who must keep the main theme and theme_template in sync
How do you install this skill?
- This is a static source review; no commands or tests were executed, confidence is low.
- The skill assumes marmite binary, Rust toolchain and mask runner are installed, but states no prerequisites.
- Referenced files (references/*.md) are absent from the evidence; their accuracy and sync status are unverified.
- The repo involves remote theme downloads and a curl|sh install script; mind supply-chain and network provenance.
- SKILL.md appears truncated at the end (cut off mid-checklist); verify against the full file.
- Builds depend on crates.io and GitHub resources, which may be constrained from mainland-China networks.
- Shell / CLI
- Network access
- Local filesystem
Rust (edition 2021)mask task runnermarmite CLI
The skill is bundled in the marmite repository at .agents/skills/marmite-development/. Clone the repo and point any Agent Skills-compatible client at the SKILL.md in that folder. The marmite binary also offers --skill / --skill-install style commands for embedded skills (the README confirms the mechanism, though the source does not state whether this development skill is embedded).
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/rochacbruno/marmite.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/.agents/skills/marmite-development" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
Once installed, send your agent any of these to trigger it:
- Add a --foo CLI flag to marmite and complete tests and docs per SKILL.md.
1) Install the mask task runner: cargo install mask. 2) Install the pre-commit hook: mask install_hook. 3) Build and preview the example site: mask serve (http://localhost:8000, WebSocket live reload). 4) After every change run mask fmt, mask check, mask test; before a PR also run mask pedantic and mask serve_theme. Example agent prompt: 'Add a --foo CLI flag to marmite and complete tests and docs per SKILL.md.'
What are this skill's strengths and limitations?
- Comprehensive: code standards, two test layers, compatibility principles, documentation requirements, and checklists all in one place
- Copy-paste-ready commands and code patterns (config fields, template functions, integration test templates)
- Explicitly teaches agents to verify refactors with built-in tools like --show-urls
- Strong backward-compatibility emphasis (feature flags, serde defaults, no field/flag removal) reduces breakage risk
- Only useful for contributing to marmite's code; worthless for users building sites (that is the sibling marmite skill's scope)
- Hard dependency on the mask task runner and a Rust toolchain adds setup cost
- Pedantic clippy issues require maintainer confirmation, limiting full agent autonomy
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| Marmite Development Guide this page | 58 · Recommended | ★ 886 | 26d ago | AGPL-3.0 |
| Marmite Static Site Generator Skill | 55 · Use with care | ★ 886 | 26d ago | AGPL-3.0 |
| Code Review and Testing Guidance ✓ OpenAI · Official | 43 · Not recommended | ★ 129k | 11d ago | Apache-2.0 |
| Test-Driven Development Workflow | 57 · Use with care | ★ 107k | 12d ago | MIT |
| Remote Executor Integration Tests ✓ OpenAI · Official | 50 · Use with care | ★ 129k | 11d ago | Apache-2.0 |
SKILL.md itself defines the split of labor with the site-building marmite skill: this one covers contributing code, the other covers building sites. The README names Zola, Cobalt, and Zine as alternative SSGs, but that is a product-level comparison irrelevant to this development skill.
How did FollowSkills review this skill?
Pure documentation skill with no destructive commands, credential handling, or covert data flows; recommends cargo audit and dependency minimization; provenance and license clearly attributed. Deducted for: remote theme download (--set-theme) and curl|sh install referenced without risk/confirmation discussion, and unverified publisher identity.
Instructions are self-consistent: toolchain commands (mask/cargo), two-layer testing, code patterns and error-handling guidance align. Deducted for: no executed verification possible in static review; maskfile not in evidence; sample --show-urls JSON contains trailing commas (invalid JSON); failure-feedback quality unverified.
Audience is explicit (contributors to the marmite codebase) and non-fit boundary is declared up front (use the marmite skill for site building). Deducted for: no Chinese-language support notes, builds depend on overseas crates.io access, and no declared environment prerequisites (Rust version, mask availability).
Well-layered structure (tools, code quality, testing, architecture, checklists); AGPL-3.0-or-later consistent across SKILL.md, Cargo.toml and LICENSE; references maintainer sign-off. Deducted for: no skill-level version/changelog, SKILL.md text appears truncated mid-checklist, and dependency on mask without install instructions.
Covers the core contribution workflow (fmt/lint/test, architecture, step-by-step feature recipes, breaking-change policy) with concrete, actionable content and clear marginal value over source-diving. Deducted for: steps unverifiable without execution; referenced files (references/*.md) absent from evidence so completeness unconfirmed.
Claims are cross-checkable against source, Cargo.toml and CI workflows; structural claims (test inventory, module table) are auditable. Deducted for: no execution evidence in static review; the listed test files themselves are not in evidence; fact/inference separation left to the reader.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →