agentleFS
hay-kothay-kot/hive-desktop/.agents/skills/docs-audit/SKILL.md

docs-audit

Skill4 starsChanged 2 months agoMITView on GitHub

Check the work on this branch (or a PR) for anything a user would need to read about and is not documented, then add or update the page on hivedesktop.com. Use when asked to audit the docs, to decide whether a change needs a docs page, or as a pre-PR pass over a feature that added a setting, a node, an action type, a surface, or a failure mode.

  • Reads credentials

15 other projects use a version of this, 1 of them edited.

Adopted by

Most recently taken up by

Used with

Projects with it also use docs-planner, docs-reviewer and 4 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • strands-agentsSkill

    docs-planner

    Find the gaps in the Box documentation and produce a prioritized backlog for docs/user/ and docs/design/. Use when planning documentation work, after a feature or a binary ships without a page, for a periodic health check of the docs, or when asked "what docs need writing", "plan the docs work", "prioritize the docs backlog", or "what should we document next".

    Also in 3 projects that use this

  • strands-agentsSkill

    docs-reviewer

    Review Box drafts in docs/user/ or docs/design/, and wording-only rewrites in docs/dev/. Use after docs-writer, before a docs pull request, or for a requested doc review. Check style and preservation of meaning, then give a Ship it, Tighten, or Rethink verdict. Read-only.

    Also in 3 projects that use this

  • VercelSkill

    docs-diagrams

    Draw diagrams for the Next.js docs in the style of the ones already published there: the light/dark PNGs an mdx references with `<Image srcLight="/docs/light/<name>.png" srcDark="/docs/dark/<name>.png">`. Use when a guide or API reference needs a new diagram or an existing one needs a change. Hand-written SVG from a primitives library, rendered by headless Chrome at 2x. No AI image generation.

    Also in 3 projects that use this

  • GoogleSkill

    docs-writer

    Also in 3 projects that use this

  • MicrosoftSkill

    writing-for-agents

    Writing documents for agents. Use when creating or editing skills, or modifying AGENTS.md or CLAUDE.md.

    Also in 3 projects that use this

  • GoogleSkill

    domain-modeling

    Build and sharpen a project's domain model. Use when the user wants to pin down domain terminology or a ubiquitous language, record an architectural decision, or when another skill needs to maintain the domain model.

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

4 rules. 0 are common, 3 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 12 edited versions we read keep these.

  • 3

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Read the diff for these and write them down before deciding anything:

    line 28
  2. Distinctive:

    Write the minimum useful page: Keep instructions, requirements, safety constraints, and recovery steps.

    line 104
  3. Distinctive:

    Avoid generated-sounding prose: Do not use em or en dashes, rhetorical contrast formulas, "Why it matters" headings, staged reveals, or …

    line 110
  4. Very specific:

    The body starts with # Title: Add a short task-oriented lede only when the title needs context.

    line 129

“Common” is measured: kept by at least 6 in 10 of the 12 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Audit the docs

The docs site is the user's only map of the app, and its failure mode is silent: a setting ships, a node type lands, a surface is renamed, and the page that described it keeps describing the old one. Nothing fails a build over it. This audit is the check.

It covers web/docs/ only. docs/architecture.md, ADRs, and the shipped agent skills are other surfaces with their own skills; step 3 says which change belongs where.

1. Scope the change

git diff main...HEAD --stat
git diff main...HEAD -- internal/app desktop/frontend/src

For a PR that is not checked out, gh pr diff <number>.

2. Inventory what a user can now see, do, or hit

Read the diff for these and write them down before deciding anything:

  • a setting: a field on settings.Settings, a new mock mode, a new environment variable read anywhere;
  • a node or action type: a connector.Descriptor, a flow node, an entry in actions.Doc, a field on any of their configs, a change to a payload contract;
  • a surface: a new view, area, settings section, drawer, dialog, or a rename of one the docs refer to by name (Settings ▸ Integrations, Chats, Code);
  • a verb or a key: a command in keybindings/catalog.ts, a default combo, a launcher rule;
  • an agent-facing thing: a shipped skill, an MCP catalogue entry, a workspace manifest field;
  • a failure a user meets: a new error message, a new precondition (a binary on PATH, a permission, a version floor), a removed fallback.

A refactor, a test, an internal rename, and a change to how something is drawn all inventory to nothing. Say so and stop. An audit that finds nothing is a valid outcome and is reported as one.

3. Decide where each one belongs

A user-facing fact goes on the docs site, on the page that already owns its subject. The site's structure mirrors the app: the sidebar groups of the Getting Started tab are the app's areas (Inbox, Code, Chats), each with its own directory under web/docs/, and the Configuration tab is the reference.

SubjectPage
a setting, a path, an environment variableweb/docs/configuration/settings.md
the updater, a release channelweb/docs/configuration/settings.md, the ## updates section
a command id or a default comboweb/docs/configuration/keybindings.md
items, feeds, and notifications as one modelweb/docs/inbox/how-it-works.md
a source node, what an account needs, the webhook contractweb/docs/inbox/sources.md
a flow node type, wiring, the editorweb/docs/inbox/flows.md
an action type, inputs, targets, launchersweb/docs/inbox/actions.md
a workspace manifest field, a skill package, an MCP entryweb/docs/chats/agent-workspaces.md
Code: attach, windows, launchers, typographyweb/docs/code/terminal-mode.md
a failure and its way outweb/docs/getting-started/troubleshooting.md
the problem reporter and what a report containsweb/docs/getting-started/troubleshooting.md, the ## Report a problem section
installing a releaseweb/docs/getting-started/index.md, the ## Install section
sign-in, notification permission, the starter feedsweb/docs/getting-started/sign-in.md, notifications.md, first-feed.md
building the appweb/docs/getting-started/build-from-source.md

A new page is rare. It is right when a subject has no owner in that table and would not fit as a section of one, not when a feature is big. It goes in the directory and sidebar group of the area it belongs to; a new area is a new directory and a new group. Adding a page is the web-docs skill's procedure: the file, its frontmatter, and its line in the nav in web/zensical.toml. A page left out of the nav builds without a warning and is reachable by URL only, which is the same as not existing.

These are not docs-site changes, and saying so is half the audit:

  • How the app is built goes in docs/architecture.md; why a choice was made goes in an ADR. A user page says what a thing does and what it needs, never how it is implemented.
  • A change to a config schema also changes what the agent is told. The shipped skills under internal/app/prompts/templates/ and the per-type docs they splice in (flow/docs/*.md, actions/docs/*.md, mcpcatalog/docs/*.md) are the agent's reference. A field documented on the site but not there means the Hive workspace writes it wrong. That is the shipped-skills skill's procedure; run it beside this one.
  • A user-visible change owes a release-notes fragment in internal/app/releasenotes/changelog/unreleased/. That is the release-notes skill.
  • Developer-only facts (mock modes, launch.env, the devserver) belong on the build page only to the extent a person building the app must know them. The rest is docs/development.md and desktop/AGENTS.md.

4. Write it

The web-docs skill covers the mechanics: frontmatter, the nav, the build. The conventions that keep the pages consistent:

  • Write the minimum useful page. Keep instructions, requirements, safety constraints, and recovery steps. Remove implementation details and reasons for ordinary controls.
  • Summarize settings and shortcuts. Group visible controls in a sentence or short list. The app owns the complete live list. Show YAML only when a user needs to edit a file manually.
  • Avoid generated-sounding prose. Do not use em or en dashes, rhetorical contrast formulas, "Why it matters" headings, staged reveals, or marketing filler. State the fact or instruction directly.
  • Keep one owner for each fact. Link to the owner page instead of copying provider lists, action types, settings, or key tables onto another page.
  • Name things what the app names them. Settings sections are Settings ▸ Integrations; the areas are Inbox, Code, and Chats; a workspace is a flow. Read sectionMeta.ts and the catalog's group values rather than guessing.
  • Say what a thing needs before what it does. tmux 3.2, an agent on PATH, a scope on a token, a permission: the precondition goes first, in an admonition if a user would otherwise discover it by failing.
  • Point at the Hive workspace where an agent could do the work. Every config page carries a short !!! tip "Ask the Hive workspace" naming the skill the seeded Hive workspace has for that file. Add one to a page that gains a config surface; do not add one to a page about a thing an agent cannot drive.
  • A callout is an admonition: !!! tip "Title" on its own line, body indented four spaces. tip, note, and info are the types in use.
  • The body starts with # Title. Add a short task-oriented lede only when the title needs context. The frontmatter is icon: and description: and its description is not repeated in the body.
  • Internal links are relative Markdown-file links (../configuration/settings.md#updates), which the strict build validates.
  • Do not hand-maintain what is derived. The tabs, the sidebar and its groups, the prev/next footer, search, the sitemap, llms.txt, llms-full.txt, and each page's .md twin all come from the nav and the pages. A page in the nav appears everywhere with no further edit.

A fact you cannot verify in the code does not go on the page. Read the struct, the descriptor, or the component; the docs are reviewed as a spec.

5. Verify

cd web && mise run build

(mise run install first in a fresh worktree; both run from inside web/.) The build runs zensical build --strict and aborts on any warning: a link to a page that does not exist, an anchor that is not a heading on its target, or a nav entry whose file is missing. Then check the two things the build cannot:

  1. Every new page is in the nav in web/zensical.toml. The build does not warn about a page that is not.
  2. The page says what the code does. Re-read the diff beside the page. A default, a range, a scope name, and a key are the four things most often wrong.

PR CI runs the same build (web-build), so a broken link fails the PR. It cannot tell a wrong default from a right one; only this step can.

Report

Say what you added or changed, page by page, and say what you decided not to document and why. "This is internal shape and belongs in architecture.md", "the shipped skill needs the same field", and "no user can see this" are all conclusions worth writing down, because the next audit re-asks the same questions.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.