Acceptance plugin
Acceptance has the coding agent verify its work on the live product and publish a report with screenshots, recordings and logs.
- 97 ★ on the repository
- Last commit
Extended access
Why this level
- The agent runs the real product, drives a browser or windows and writes report files into the project.
- After signing in to LobeHub, screenshots and recordings are published at a link open to anyone who has it.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add plugins/acceptanceDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The agent runs the real product, drives a browser or windows and writes report files into the project.
- After signing in to LobeHub, screenshots and recordings are published at a link open to anyone who has it.
This is third-party code. Review the repository files before installing.
What it does
Acceptance is a skill and plugin for accepting work done by an agent. The agent turns a task into a list of checks, runs the real product and captures evidence: screenshots, recordings with chapters and logs. Every check gets its own attachments. The report is rendered into one `report.html` file in the `.acceptances/` folder. In it you can accept a check, send it back with a comment or circle a region on a screenshot. The command `lh acceptance feedback` reads those decisions, and the agent starts the next round from them. Supported surfaces are web, CLI, Electron, macOS and the iOS Simulator.
Who it is for. Teams that hand frontend and product tasks to an agent and want to see the result on screen and find a chat summary insufficient.
Good fit when
- The agent finished a screen and light and dark themes should be seen side by side in the same state.
- A button breaks on a narrow screen with long German labels and a recording has to show it.
- A feature was finished overnight and design and product need one report link in the morning.
- One check should go back to the agent while the rest stay accepted.
Not a fit when
- You need unit tests for logic: Acceptance looks at the product as a user would.
- The product cannot run on the agent machine and the surface is unavailable.
- No team review is planned and a few manual screenshots are enough.
Example request
Use Acceptance to verify the new sign-in form works in light and dark themes and on a 320 pixel wide screen.Limitations
The `lh` CLI needs Node.js 22.15 or newer; after installing the skill, reload skills or start a new agent session. Until you sign in to LobeHub, reports stay local. After `lh login` reports are published to LobeHub and use account storage, and anyone with the link can open a published report. Team comments and mentions work only with a LobeHub account. The agent must be able to run the product and drive the surface, otherwise a check cannot attach evidence.
How to disable. Delete the `.agents/skills/acceptance` folder in the project or remove the plugin with `/plugin` in Claude Code. Reports in `.acceptances/` stay on disk.
FAQ
How do I install Acceptance in Claude Code?
Add the marketplace with `/plugin marketplace add lobehub/acceptance` and install the plugin: `/plugin install acceptance@acceptance`. Then ask the agent to verify a feature with Acceptance.
Is an account required?
No. Without signing in to LobeHub the report is rendered locally as `report.html`. An account is needed for a shared link and team comments.
Which agents are supported?
Any agent that supports skills and can reach the surface you check. The README names LobeHub, OpenClaw, Claude Code, Codex, Amp, Gemini CLI, Cursor and OpenCode.
Which checks are supported?
Pages and flows, commands and APIs, Electron and macOS windows, gestures in the iOS Simulator. For each check the agent attaches screenshots, recordings and logs.
What stays on disk after a check?
The `.acceptances/` folder: check rounds, evidence and `report.html`. It opens without internet and without signing in to LobeHub.
Discussion
No questions or reviews yet.
Related
PluginsA self-improving QA agent: natural-language tests for web and mobile that self-heal and remember runs
A Claude Code plugin that makes an agent check the code around a change and prove the work with a test and independent review
Snapshot tests for AI agents: records tool calls today and tells you when they silently change after an edit
A skill that tests your agent's code in a real browser: reads your changes, builds a test plan and runs it through Playwright with your own cookies
QA agents, skills and an MCP server: test generation, coverage gaps, flaky tests, security scans and chaos testing
A local debugger for TypeScript agents: run step trees, CI trajectory checks and portable evidence bundles, with no account.
README in short
The idea is acceptance: the agent says it is done, and a person decides by looking at the evidence. Each round leaves criteria and examples of misses, and the agent checks them before its next handoff.