Agent skill

Docs Check

by RLinf in RLinf/RLinf

Cross-check RLinf documentation against code, natural explanation flow, and other docs, including English-Chinese parity.

Apache-2.0Auto-check passedAI & LLM Engineering

Install Docs Check

skills CLI
$ npx skills add RLinf/RLinf --skill docs-check -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install RLinf/RLinf docs-check --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/RLinf/RLinf.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.agents/skills/docs-check .claude/skills/docs-check && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
docs-check
GitHub stars
5.5k
Token cost
~2.1k tokens
SKILL.md length
960 words
Files
5
Skills in repo
9
Repo updated
First seen
Licence
Apache-2.0

At a glance

Cross-check RLinf documentation against code, natural explanation flow, and other docs, including English-Chinese parity.

  • Works in 7 steps: Read reference.md and extract the… → Run the build harness for both languages… → Verify doc-to-code correctness → …
  • Reviewing doc PRs
  • SKILL.md covers Quick Start, Inputs, Workflow and Severity Rules, plus 3 more sections
  • Runs Python scripts from its folder; calls python3

What it does

Docs Check is an agent skill from RLinf/RLinf. Cross-check RLinf documentation against code, natural explanation flow, and other docs, including English-Chinese parity. Use when adding or editing docs, reviewing doc PRs, validating commands/config keys/model-env names, or checking EN/ZH readability and consistency.

Its SKILL.md is about 2.1k tokens, which your agent loads only when the skill is triggered. The skill folder holds 4 other files (for example `build_docs.py`, `check_doc_symbols.py` and `check_rst_markup.py`).

It sits in AI & LLM Engineering, covering Plain language and style rules. The repository describes itself as: RLinf: Reinforcement Learning Infrastructure for Embodied and Agentic AI. The licence is Apache-2.0.

When your agent uses it

  • Reviewing doc PRs
  • Validating commands/config keys/model-env names
  • Checking EN/ZH readability and consistency

Example prompts

  • “/docs-check”

Requirements

  • Python 3

Workflow steps

7 steps, taken from the first numbered list in SKILL.md.

  1. Read reference.md and extract the relevant checklist items.
  2. Run the build harness for both languages and fix every warning
  3. Verify doc-to-code correctness
  4. Verify doc-to-doc consistency within one language
  5. Verify EN-ZH parity
  6. Verify natural-language flow
  7. Report findings with severity and concrete fixes.

What it can do on your machine

Read from SKILL.md and the folder at commit e27e631. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Ships script files (Python), which the agent can run.

    Shell commands in SKILL.md call:

    • python3

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Docs Check loads about 2.1k tokens when it runs. Until then it costs about 70 tokens; SKILL.md has 960 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~70
When it runs · the whole SKILL.md, loaded when a task matches
~2.1k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from RLinf/RLinf at commit e27e631, republished under its Apache-2.0 licence (© RLinf). 960 words, ~2,146 tokens.

Download SKILL.mdSave it as .claude/skills/docs-check/SKILL.md (or your agent's skills folder). This skill also uses 4 other files; get the full folder from GitHub.
name
docs-check
description
Cross-check RLinf documentation against code, natural explanation flow, and other docs, including English-Chinese parity. Use when adding or editing docs, reviewing doc PRs, validating commands/config keys/model-env names, or checking EN/ZH readability and consistency.

Docs Check

Quick Start

Use this skill when documentation changes may introduce mismatches with:

  • Code and config source of truth
  • Other existing docs in the same section
  • English and corresponding Chinese docs

Always read reference.md first, then run the workflow below.

Two harnesses in this folder do the mechanical part; run them before reasoning about content:

bash
# Does the page still build? Both Read the Docs projects use fail_on_warning.
python3 .agents/skills/docs-check/build_docs.py            # en + zh

# Does CJK punctuation break inline markup, loudly or silently?
python3 .agents/skills/docs-check/check_rst_markup.py      # needs docutils

# Does the page still name things the code has?
python3 .agents/skills/docs-check/check_doc_symbols.py

Inputs

Collect these inputs before reviewing:

  • Changed doc files (or target docs to validate)
  • Corresponding EN and ZH files for the same topic
  • Related code/config files referenced by the docs

If scope is unclear, default to checking:

  • docs/source-en/ and docs/source-zh/ counterparts
  • rlinf/config.py (SupportedModel)
  • rlinf/envs/__init__.py (SupportedEnvType)
  • Referenced scripts under examples/, toolkits/, ray_utils/, and requirements/

Workflow

  1. Read reference.md and extract the relevant checklist items.
  2. Run the build harness for both languages and fix every warning:
    • python3 .agents/skills/docs-check/build_docs.py (add --lang zh to iterate faster on a Chinese-only failure).
    • The two Read the Docs projects build independently with fail_on_warning: true, so an English-clean page can still turn docs/readthedocs.org:rlinf-cn red. Never conclude "docs are fine" from one language.
    • Then python3 .agents/skills/docs-check/check_rst_markup.py for the inline-markup defects that render wrong without warning.
  3. Verify doc-to-code correctness:
    • Commands exist and are runnable in principle.
    • Script/module paths in docs exist.
    • Config keys and values match real code/config names.
    • Model/env names match SupportedModel and SupportedEnvType string values.
  4. Verify doc-to-doc consistency within one language:
    • Terminology is consistent across start/tutorials/examples/API pages.
    • New page is linked in the correct index/toctree.
    • No conflicting instructions between related pages.
    • Internal doc links use stable :doc:/relative links, not hardcoded ReadTheDocs URLs.
  5. Verify EN-ZH parity:
    • Same topic coverage and section structure.
    • Same commands, config keys, and model/env identifiers.
    • Translations preserve technical meaning (do not rename code symbols).
    • Corresponding EN/ZH pages use equivalent stable internal links.
  6. Verify natural-language flow:
    • Treat continuity as a requirement for every article, including landing, recipe, reference, and non-code prose pages. Adapt the depth of the lead to the page type, but do not exempt a page from establishing its purpose and order.
    • The first prose sentence after the title, or after a leading figure, states directly what the page explains, enables, routes, or lets the reader look up. Background must not delay the page's purpose until a later paragraph.
    • The page introduction establishes the reader's starting situation, the promised result, the topic boundary, and the order of the explanation.
    • Read the introduction followed only by each section's opening paragraph. They must form a coherent outline in which every section follows from the state established before it.
    • Each section opens by stating the question it resolves and its connection to the surrounding workflow; it does not begin abruptly with code, a table, an API identifier, or a fact unrelated to the previous section.
    • Paragraphs develop one argument in dependency order rather than forming a reorderable list of correct facts.
    • Every public operation in the primary example is explained in caller order, including the relevant input or return value and lifecycle effect. Each non-trivial code block has a stated purpose and an interpretation.
    • Start from a concrete reader question, explain the idea in ordinary language, then introduce the exact API term and example.
    • Headings, cards, and opening sentences do not introduce unexplained implementation terms. An identifier heading is appropriate only on a lookup-oriented API or reference page, or after the term is established.
    • English reads as direct colleague-to-colleague prose, without canned transitions, uniform paragraph rhythms, or promotional summaries.
    • Chinese follows natural Chinese logic rather than English clause order. Familiar developer terms stay in English when translating them would sound unusual or make the code harder to search.
    • Chinese uses restrained written technical language: neither casual chat nor bureaucratic prose.
    • Chinese prose paragraphs and prose list items stay on one source line. A hard wrap inside prose renders as a visible space between Chinese characters even when the source contains no typed space.
  7. Report findings with severity and concrete fixes.
Show full SKILL.md (349 more words)Show less

Severity Rules

  • Critical: A Sphinx warning in either language — Read the Docs fails the build, so the page does not ship at all.
  • Critical: Wrong command/path/key/value that can break user workflow.
  • Major: Inconsistent docs that likely mislead users.
  • Major: An unexplained implementation term in a heading or opening breaks the reading flow on a concept, guide, or extending page.
  • Major: A page or section lacks a lead, sections do not form a logical progression, or the primary example uses public operations that the prose never explains.
  • Minor: Wording/terminology drift without immediate breakage.
  • Minor: Formulaic or translated prose is understandable but does not read naturally in its language.

Prefer actionable findings with exact file paths and corrected values.

Hardcoded ReadTheDocs links to RLinf docs should be reported as at least Major.

Output Format

Use this format when reporting results:

markdown
## Docs Check Findings

- Critical: <issue>, in `<path>`
  - Why: <impact>
  - Fix: <specific correction>

- Major: <issue>, in `<path>`
  - Why: <impact>
  - Fix: <specific correction>

- Minor: <issue>, in `<path>`
  - Why: <impact>
  - Fix: <specific correction>

## Verified

- <what was checked and confirmed>

If no issues are found, explicitly state:

No doc-code or EN-ZH consistency issues found in checked scope.

Guardrails

  • Do not invent model/env/config names; verify against source files.
  • Do not change code to match incorrect docs unless explicitly requested.
  • Keep EN and ZH technical tokens identical where applicable (paths, CLI flags, keys, enum values).
  • Do not force-translate familiar terms such as policy, key, value, mapping, endpoint, worker, binding, wrapper, mock SDK, contract, shape, schema, and API; explain their role in natural Chinese and retain the searchable term when that is normal developer usage. In RL prose, keep policy in English rather than translating it as “策略”; generic strategies, such as a placement strategy, may still use “策略”.
  • When uncertain, flag as an assumption and request confirmation.
  • Do not keep RLinf internal links as hardcoded readthedocs.io/.../rst_source/... URLs; convert to :doc: or relative internal links.

Quick Detection

Use this regex scan to detect unstable hardcoded RLinf docs links:

  • readthedocs\.io/(en|zh-cn)/latest/rst_source/

Chinese pages: or `**` sitting directly against `(`, `《` or a CJK character is the single most common way to break `rlinf-cn`. Grep for ````( and **( ```` as a first pass, then run check_rst_markup.py for the full rule.

Additional Resource

© RLinf, Apache-2.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 4 other files in .agents/skills/docs-check of RLinf/RLinf.

  • SKILL.md
  • build_docs.py
  • check_doc_symbols.py
  • check_rst_markup.py
  • reference.md

Open the folder on GitHubat commit e27e631

Compare with similar skills

Docs Check next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Docs Check compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Docs Check this skillRLinf/RLinf5.5k—~2.1kAutomated safety check: PassApache-2.0
Contextpilot SavingsEfficientContext/ContextPilot140—~1.4kAutomated safety check: PassMIT
Quark Onnx Routeramd/Quark181—~2.7kAutomated safety check: PassMIT
Quark Torch Routeramd/Quark181—~1.9kAutomated safety check: PassMIT
Quark Torch Routeramd/Quark181—~1.8kAutomated safety check: PassMIT
Asd Ste100danyuchn/asd-ste100-skill4k—~4.1kAutomated safety check: PassMIT

Similar skills

  • Contextpilot Savings

    EfficientContext/ContextPilot

    A skill your agent uses when a user asks how many tokens (or how much context/cost) ContextPilot has saved, or wants a ContextPilot savings status/summary inside Hermes Agent — e.g.

    140 GitHub stars~1.4k tokensUpdated 5 days ago
    AI & LLM EngineeringAuto-check passed
  • Route Quark ONNX user goals to the correct atomic skill. An agent skill from amd/Quark.

    181 GitHub stars~2.7k tokensUpdated 10 days ago
    AI & LLM EngineeringAuto-check passed
  • Route Quark user goals to the correct atomic skill or workflow.

    181 GitHub stars~1.9k tokensUpdated 10 days ago
    AI & LLM EngineeringAuto-check passed
  • Route Quark user goals to the correct atomic skill or workflow.

    181 GitHub stars~1.8k tokensUpdated 10 days ago
    AI & LLM EngineeringAuto-check passed
  • Asd Ste100

    danyuchn/asd-ste100-skill

    A skill your agent uses when English text must be parsed without a human to resolve ambiguity — tool descriptions, error messages, inter-agent instructions, system prompts, status reports — and…

    4k GitHub stars~4.1k tokensUpdated 4 days ago
    Writing & ContentAuto-check passed
  • LLM Benchmark

    lumose-health/GlycemicGPT

    A skill your agent uses when the user wants to benchmark, evaluate, or vet an LLM/AI model for GlycemicGPT — checks a configured model for safety/correctness and performance against the project's…

    141 GitHub stars~1.6k tokensUpdated yesterday
    Writing & ContentAuto-check passed

More from RLinf/RLinf

All 9 skills in this repo
  • Adds example documentation for a new model or environment in RLinf (RST pages in the docs gallery for both English and Chinese).

    5.5k GitHub stars~1.9k tokensUpdated today
    Auto-check passed
  • Adds a new publication page to the RLinf Sphinx docs (EN + ZH) and wires it into the Publications index/toctree.

    5.5k GitHub stars~1.1k tokensUpdated today
    Auto-check passed
  • Create PR

    RLinf/RLinf

    Open a GitHub pull request for RLinf, or fix an existing one — checks the PR title against Conventional Commits, writes a precise description that follows .github/PULLREQUESTTEMPLATE.md, and lints…

    5.5k GitHub stars~2.2k tokensUpdated today
    Auto-check passed
  • Install Check

    RLinf/RLinf

    Check, fix, or extend requirements/install.sh and its docker/Dockerfile coverage when adding a new embodied model or environment in RLinf, so the install logic reuses common utilities, keeps system…

    5.5k GitHub stars~2.3k tokensUpdated today
    Auto-check passed
  • Test Install

    RLinf/RLinf

    Test that requirements/install.sh works for an embodied model/env by building its venv and running the matching CI e2e test.

    5.5k GitHub stars~2.3k tokensUpdated today
    Auto-check passed
  • Adds install command in install script, Docker build stage in Dockerfile, and CI jobs for docker build and embodied e2e test when introducing a new model or environment in RLinf.

    5.5k GitHub stars~1.4k tokensUpdated today
    Auto-check passed

Questions about Docs Check

What does Docs Check do?

Cross-check RLinf documentation against code, natural explanation flow, and other docs, including English-Chinese parity. Docs Check is an agent skill from RLinf/RLinf. Cross-check RLinf documentation against code, natural explanation flow, and other docs, including English-Chinese parity.

When should I use Docs Check?

Docs Check fits situations like: reviewing doc PRs; validating commands/config keys/model-env names; checking EN/ZH readability and consistency.

How do I install Docs Check in Claude Code?

Run `npx skills add RLinf/RLinf --skill docs-check -a claude-code`. Or copy the skill folder (.agents/skills/docs-check in RLinf/RLinf) into .claude/skills/docs-check in your project. Claude Code loads it when a task matches its description.

How do I install Docs Check in Codex?

Run `npx skills add RLinf/RLinf --skill docs-check -a codex`. Or copy the skill folder (.agents/skills/docs-check in RLinf/RLinf) into .agents/skills/docs-check in your project. Codex loads it when a task matches its description.

Can I use Docs Check in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add RLinf/RLinf --skill docs-check -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/docs-check, .gemini/skills/docs-check, .github/skills/docs-check and .opencode/skills/docs-check in your project.

What does Docs Check need to run?

Going by SKILL.md and its folder, Docs Check needs Python for the scripts in its folder and the command-line tools its instructions call (python3). Our summary lists: Python 3.

Does Docs Check access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Docs Check safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Docs Check use?

Docs Check is published under the Apache-2.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Docs Check use?

About 2.1k tokens (SKILL.md is roughly 8.6k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Docs Check?

Skills that share tags, products or a category with Docs Check: Contextpilot Savings (EfficientContext/ContextPilot, 140 stars), Quark Onnx Router (amd/Quark, 181 stars), Quark Torch Router (amd/Quark, 181 stars) and Quark Torch Router (amd/Quark, 181 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Docs Check?

RLinf (a GitHub organization) maintains it in RLinf/RLinf, which has 5,457 GitHub stars. The repository holds 9 skills in this directory. The repository was last updated on October 8, 2026.

Source: RLinf/RLinf on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.