document skill
Writes or updates technical documentation by following project conventions and profile-specific rubrics after feature implementation or bug fixes.
- GitHub stars
- 155
- Files
- 16
- Last commit
- Nov 10, 2026
- License
- —
- Allowed tools
- —
What does the document skill do?
document writes or updates technical documentation by following project conventions and profile-specific rubrics after feature implementation or bug fixes.
Summary generated from the skill's SKILL.md.
About the document skill
Updating technical documentation after completing a feature or bug fix is the primary function of this skill. It is designed for developers who need to ensure that new code changes are accurately reflected in project files while adhering to specific organizational standards. The process begins by inspecting the feature directory and detecting active profiles through a specific detection procedure. Once profiles are identified, the skill loads relevant rubrics from index.md files located within profile-specific document subdirectories.
An agent should load this skill immediately following the implementation phase to maintain documentation integrity. It produces updated README files, architecture guides, API documentation, or Architecture Decision Records (ADRs) in the appropriate project directories. The skill relies on shared protocols like shared-capy-knowledge-protocol.md and shared-profile-detection.md to frame its writing process. Unlike standard writing tools, it requires explicit handling of required topics by either writing the content, stating N/A with a reason, or citing inherited sources to avoid silent omissions.
What an agent can do with it
- 1Documenting new feature changesWrite or update technical documentation for a recently implemented feature following project conventions.
- 2Recording architectural decisionsCreate an Architecture Decision Record if the code change involved choosing between several alternatives.
- 3Applying profile-specific rubricsEnsure documentation covers all required topics defined by an active profile's index.md file.
- 4Maintaining documentation completenessAddress every required topic by providing content, an N/A reason, or an inherited source citation.
Written from the skill's SKILL.md. Every use case is checked against a quote from it.
How to install the document skill
Each command copies the skill folder into the client's skills directory. Restart the client or start a new session to load it.Install the document skill in Plugin
/plugin marketplace add serpro69/claude-toolbox /plugin install kk@claude-toolbox
Install the document skill in Claude Code
~/.claude/skills/document/SKILL.md
$ npx degit serpro69/claude-toolbox/klaude-plugin/skills/document ~/.claude/skills/documentInstall the document skill in Claude Code (project)
.claude/skills/document/SKILL.md
$ npx degit serpro69/claude-toolbox/klaude-plugin/skills/document .claude/skills/documentInstall the document skill in Codex
~/.agents/skills/document/SKILL.md
$ npx degit serpro69/claude-toolbox/klaude-plugin/skills/document ~/.agents/skills/documentInstall the document skill in Gemini CLI
~/.gemini/skills/document/SKILL.md
$ npx degit serpro69/claude-toolbox/klaude-plugin/skills/document ~/.gemini/skills/documentInstall the document skill in Claude.ai
npx degit serpro69/claude-toolbox/klaude-plugin/skills/document document && zip -r document.zip document # claude.ai → Customize → Skills → + → Create skill → Upload a skill → document.zip
SKILL.md
View on GitHubDocumentation Process
Conventions
- Read capy knowledge base conventions at shared-capy-knowledge-protocol.md.
- Read profile detection at shared-profile-detection.md. When an active profile contributes a
document/subdirectory (e.g.,${TOOLBOX_PLUGIN_ROOT}/profiles/k8s/document/), itsindex.mdlists a doc rubric — required topics the documentation for that artifact type must cover. See the Workflow below for the load order.
Workflow
Mandatory order — instructions before action. The flow below is strictly sequential. Do not read feature-tree content, write, or edit documentation files until the shared protocols and all resolved profile content are in context. Bounded signal inspection for profile detection is the only content-read exception.
Read the shared protocols in Conventions before the steps below, even when the invocation ultimately needs no edits.
- Minimal-scope listing. List the feature directory (filenames and metadata only — no file-content reads). This is the input profile detection needs, and nothing more; content-level reading happens after profile content is loaded.
- Detect active profiles. Run the
shared-profile-detection.mdprocedure against the filename list from Step 1. - Load profile content. For each active profile that contributes a
document/subdirectory, load${TOOLBOX_PLUGIN_ROOT}/profiles/<name>/document/index.mdand read its always-load + any matching conditional content. The rubric named there specifies topics the documentation must cover for that profile's artifacts. - Read the feature-tree content the documentation will cover. This is the first step that touches subject-matter content; the profile rubric is now loaded and frames what to look for.
- Apply the doc guidelines below. Write or update documentation applying the rubric's required topics where applicable. Explain the purpose and applicable current or planned behavior before reference detail, retaining evidence limits and unresolved decisions.
- Suggest optional clarification. After creating or materially revising documentation, suggest a
/kk:clarify-docsinvocation naming those output paths. If no documents were created or materially revised, skip the suggestion. Clarification is a separate user-selected editing workflow; this skill neither loads its procedure nor runs it automatically. Drafting still owns clear explanations and every applicable profile-rubric topic, including explicit N/A reasons and inherited-source citations. The suggestion does not replace any project-prescribed review.
Guidelines
- Discover the project's documentation structure. List top-level doc directories and doc-related files at the repo root (e.g.,
docs/,README.md,ARCHITECTURE.md,CONTRIBUTING.md). Scan for architecture guides, testing guides, API docs, user guides, and contributing docs — common locations includedocs/contributing/architecture.md,docs/contributing/testing.md, but every project organizes differently. Update whichever docs are relevant to the change — don't limit yourself to a fixed set of paths. - If the code change included prior decision-making out of several alternatives, document an ADR at
/docs/adrfor any non-trivial/non-obvious decisions that should be preserved. - Profile-aware rubric. For each active profile, apply the doc rubric its
document/index.mdspecifies (loaded in Step 3 of the Workflow). Each required topic must be addressed in one of three ways: (a) write the topic if the feature touches it, (b) stateN/A — <reason>in a single line if the feature does not touch the topic, or (c) cite the inherited source explicitly if the feature assumes the topic but inherits it from elsewhere (e.g., NetworkPolicy defined in a platform repo). Silent omission is the failure mode — an explicitN/Acommunicates consideration; an absent heading communicates nothing.
Capy search: Before writing docs, search kk:arch-decisions and kk:project-conventions for decisions that should be reflected in documentation — decisions not obvious from code alone.
Files in this skill
16document skill questions
How do I install the document skill in Claude Code?
Run npx degit serpro69/claude-toolbox/klaude-plugin/skills/document ~/.claude/skills/document. Claude Code loads skills from ~/.claude/skills (all projects) or .claude/skills (one project) and uses a skill when its description matches the task.
Which clients can use the document skill?
Any client that reads the Agent Skills format: Claude Code, Codex, Gemini CLI and claude.ai (upload as a ZIP). Scripts in the skill run only where the client allows code execution.
What should I do if a required documentation topic is not applicable to the feature?
You must explicitly state 'N/A — <reason>' in a single line rather than omitting the heading entirely.
When should I create an Architecture Decision Record (ADR)?
You should document an ADR at /docs/adr if the code change involved non-trivial or non-obvious decision-making among several alternatives.
What is the mandatory order of operations for this skill?
The workflow must follow a strict sequential order: minimal-scope listing, profile detection, loading profile content, reading feature-tree content, applying guidelines, and suggesting optional clarification.