Agent Skill
The browserlane agent skill — the token-light, shell-driven third surface alongside the CLI and MCP.
browserlane ships an agent skill — a single SKILL.md that teaches a coding agent (Claude Code and other skill-aware agents) to drive the bl CLI directly. It is the third way to use browserlane, alongside the human CLI and the MCP server — not another engine: the skill teaches token-efficient CLI workflows and implements no browser capability of its own. Install it with bl add-skill.
Three surfaces, one engine
| Surface | Who uses it | How it is consumed |
|---|---|---|
| CLI | People and shell scripts | You type bl … commands |
| MCP | AI agents | The agent calls tools over bl mcp (JSON-RPC) |
| Skill | AI agents | The agent reads SKILL.md, then runs bl … in its own shell |
Why a skill (token efficiency)
The MCP server advertises 127 tool schemas that an agent loads into its context up front. The skill is one Markdown file the agent reads on demand (progressive disclosure) and then drives the ordinary bl CLI through its normal shell — no per-tool JSON schemas resident in context. Same engine, far fewer tokens. It is the lightest way to give a shell-capable agent a browser.
Install
bl add-skill
# Installs to ~/.claude/skills/browserlane/SKILL.md
bl add-skill --stdout
# Print the skill to stdout without installingWhen to use which
- Skill — the agent has shell access (e.g. Claude Code) and you want the smallest context footprint. The recommended default for CLI-capable agents.
- MCP — the agent has no shell, or you want browser actions exposed as first-class, individually described tools.
- CLI — you are driving the browser yourself or scripting it.
See the full skill file for exactly what the agent reads, the bl add-skill command, One engine, three surfaces for how the surfaces relate, and the CLI ↔ MCP mapping.
Agent recipes
A playbook for driving browserlane from an AI agent — the map/act/re-map loop, ref hygiene, token-efficient reading, and verification.
The skill file
Browser automation for AI agents. Use when the user needs to navigate websites, read page content, fill forms, click elements, take screenshots, assert page state, or manage browser tabs.