A browser automation CLI with agent skills and an MCP interface.
On this page
Overview
Agent Browser's instructions cover the CLI, MCP tools, documentation site, and skills. They name the source that owns detailed skill content, show how to run real-browser tests, and connect release notes to publication.
Keep discovery stubs separate from working instructions
The documentation checklist routes overview changes to the core skill and detailed content to its references. The installable stub remains a redirect, preventing two diverging versions of the feature instructions.
Source excerpt starting at line 24.
243. `skill-data/core/SKILL.md` (and its `references/`) — so AI agents know about the feature when they load the core skill. Edit `skill-data/core/SKILL.md` for overview/workflow changes; edit `skill-data/core/references/*.md` for detailed reference content. Do **not** put feature content in `skills/agent-browser/SKILL.md` — that file is an intentionally thin discovery stub for `npx skills add` and exists only to redirect agents to `agent-browser skills get core`.
02 / Pointing at the source of truth
Require CLI and MCP behavior to move together
The parity section covers commands, flags, output, environment variables, and parsing. It recommends reusing the canonical CLI parser and demands tests for agreement between the two interfaces.
Source excerpt starting at line 32.
32When adding or changing any CLI command, flag, behavior, output, environment variable, or parser semantics, update the MCP server in `cli/src/mcp.rs` in the same change. MCP tools should stay in sync with canonical CLI behavior by delegating through the normal CLI parser where possible. If a CLI command has no dedicated MCP tool, add one or document why it is intentionally omitted. Add or update tests that prove the CLI and MCP surfaces remain aligned.
03 / Verification by change type
Make real-browser test execution distinct
The test section separates fast unit tests from ignored tests that launch Chrome. It states both the serial-execution requirement and the contention that requirement avoids.
Source excerpt starting at line 119.
119- Must run serially (`--test-threads=1`) to avoid Chrome instance contention
04 / Architecture as narrative
Describe the release-note extraction boundary
Release preparation uses markers around only the current changelog entry. The guide explains that CI extracts the GitHub release body from those markers, so moving them is part of preparing the release.
Source excerpt starting at line 54.
54When the PR merges, CI compares `package.json` version to what's on npm. If it differs, it builds all 7 platform binaries, publishes to npm, and creates the GitHub release automatically. The GitHub release body is extracted from the content between the `<!-- release:start -->` and `<!-- release:end -->` markers in `CHANGELOG.md`.
Ideas for your repo
01Distinguish skill discovery documents from the detailed instructions they locate.
02Define parity obligations across CLI and protocol interfaces.
03Document test isolation requirements and their causes.
04Explain the markers consumed by release automation.
Sponsored by
Give your agents the whole story.
These instructions explain how to work in Agent Browser. Modem shows your agents what customers said, who is affected, and what changed.