feature-done
Skill17 starsChanged 12 days agoMITView on GitHub
Use before planning, briefing or reporting any user-facing change in hivemind, hivemind-plugins or HiveHub (a feature, a new command, a plugin kind, a data or format change). The definition of done for this open-source project, surface by surface, so nothing ships with the CLI, docs, READMEs, app screens, migrations or HiveHub left behind.
2 other projects use a version of this.
Adopted by
Most recently taken up by
- ApurvBazari/claude-pluginstheir own version, first seen Sep 29, 2026
- midFang/ai-agent-skills-workflowtheir own version, first seen Jun 25, 2026
Other versions2 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 2
- ApurvBazari/claude-plugins★ 3same name
- midFang/ai-agent-skills-workflow★ 2same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository11 other files it gives its agents.
- AGENTS.mdAGENTS.md
- guide/AGENTS.mdAGENTS.md
- spec/agents.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- adding-an-agent-providerSkill
- ascii-character-animationSkill
- hive-browserSkill
- hivemind-custom-viewSkill
- hivemind-live-wallpaperSkill
- maintaining-docsSkill
- test-auditSkill
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 17
When several agents work at once, give each one the files and contracts it owns (a manifest field, a schema, a command) and its own …
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill 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
Feature done
Hivemind is an open-source product that people and their agents drive. A change is done when every surface below that it touches is done, not when the code compiles. The maintainer should never have to name a missing surface.
Before you start
Walk the table and write down which rows the change touches. A row that does not apply gets one word why ("no data change"). Brief every subagent with the rows it owns.
When several agents work at once, give each one the files and contracts it owns (a manifest field, a schema, a command) and its own worktree and branch. No two agents add the same field, and no agent moves or merges into another agent's branch. Ready work lands as one combined PR per round, not one PR per change.
The surfaces
| # | Surface | Done means |
| 1 | Agent-native CLI | An agent can do it end to end from hive alone: create, check, install, switch on, update, undo, inspect (--json, stable exit codes, a fix line on refusal). Nothing needs a hand-written file or a click. |
| 2 | App UI | The same action exists in the app, in the place users already look, using components/ui. Checked in the running app (headless xvfb, isolated HOME) with screenshots of every state, looked at before reporting. |
| 3 | Phone and browser page | Says whether the change reaches the phone/web page; if it should, it does. |
| 4 | Guide | A page in site/src/content/guide/ explains it for a user, with real commands. The CLI reference (guide/(reference)/cli.mdx) and any API reference list new commands, flags, routes and shapes. One source per fact: link, never repeat. |
| 5 | Agent briefs | PROMPT.md starters, hive guide, and the hivemind skill text in packages/hive-core/src/templates.ts say how an agent does it. They say how Hivemind does it and never name or ban an agent's own tools. |
| 6 | READMEs and package metadata | Every README it touches names what it is and links its guide page by full URL; no stale command. homepage/repository in published packages point at the guide. |
| 7 | HiveHub | If plugins, ids, kinds or the index change: HiveHub validates, lists and installs it, its site shows it, and docs/operations.md in the hub says how to deploy it. Say the deploy order when the app and the hub must change together. |
| 8 | hivemind-plugins | Manifests, READMEs and dist match the new rules; its CI passes against main and the release. Merging there publishes, so say when it may merge. |
| 9 | Local data | If what is on disk changes, a numbered migration ships in the same release (packages/hive-core/src/migrations/), tested on old-shaped data. Users never lose installs, settings or state. |
| 10 | Contracts and versions | A changed protocol, manifest, hive ctl --json shape or CLI flag bumps its own version and gets a Breaking: CHANGELOG line. No shims for older app versions. |
| 11 | Security | New input from plugins, peers, the phone or the network is checked at the boundary; secrets by name only; sandboxes hold. Address every security-review finding before reporting. |
| 12 | Tests | Behaviour tests at the real boundary pass .claude/skills/test-audit/SKILL.md, each shown failing first. Locally run only the test files for what you touched, plus typecheck; CI (about 5 min, on GitHub) runs every suite on the PR, so do not repeat it on the laptop. One focused e2e for a new user flow; the full e2e runs once per round, after merge, and before a release. |
| 13 | CHANGELOG and release | One line under [Unreleased] in the user's words. Release and installer steps still work (scripts/install-*-test.sh when assets or install paths change). |
| 14 | Cleanup | Every worktree, shell, monitor, dev server and container you started is closed or listed. Old text, flags and code the change replaces are removed in the same change. |
Report
End with the table: each row done (with where: PR, file, screenshot path), n/a (why), or open (what is left and who has it). An open row is said first, not buried.
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.

