Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add instructions/browser-use/terminal/agents-mdgit clone --depth 1 https://github.com/browser-use/terminalWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00739 | $0.00739 |
| Opus 5 | $0.00369 | $0.00369 |
| Sonnet 5 | $0.00148 | $0.00148 |
| Haiku 4.5 | $0.00074 | $0.00074 |
Grade A, and why
terminal AGENTS.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 68 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Agent Notes
Rust Rewrite Verification Loop
This branch is Rust-first. The old Textual app and src/llm_browser runtime are intentionally gone.
Use these checks after runtime, state, provider, worker, or TUI changes:
cargo fmt --check
cargo test
uv run --with pytest python -m pytest -q
Use deterministic Ratatui dumps for visual iteration:
cargo run -q -p browser-use-tui -- --state-dir /tmp/but-rust-empty --dump-screen
cargo run -q -p browser-use-tui -- --state-dir /tmp/but-rust-ready --seed-demo done --select-latest --dump-screen
cargo run -q -p browser-use-tui -- --state-dir /tmp/but-rust-running --seed-demo running --select-latest --dump-screen
cargo run -q -p browser-use-tui -- --state-dir /tmp/but-rust-browser --seed-demo done --select-latest --overlay browser --dump-screen
Keep any saved dump outputs under /tmp/but-design-loop/.
Terminal UI Testing Standard
For any change touching crates/browser-use-tui, Ratatui rendering, keyboard handling, overlays, terminal output, scrollback, paste, resize behavior, completed output, or terminal state, do not claim the work is complete from cargo test, cargo run, or Ratatui/TestBackend dumps alone.
Terminal UI testing means acting like a keyboard user:
- launch the app in a real terminal, PTY, or tmux session
- send real keys for the changed workflow
- capture the visible terminal after meaningful actions
- assert expected UI text appears
- assert broken terminal artifacts do not appear: duplicate app chrome, leaked escape sequences, stale intermediate redraws, bracketed paste markers, or ANSI escapes in plain output
- inspect deterministic dumps for the relevant states
For any terminal UI change, run the required full verification command before calling the work done:
scripts/verify-terminal-ui.sh
This command is the repo-owned definition of done for TUI work. It runs formatting, Rust tests, Python tests, deterministic Ratatui dumps, and the real tmux terminal smoke test.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- yesterday First seen · 68 lines · 739 tokens per session scan A fa2a3200faff
terminal AGENTS.md is an instructions file published in the GitHub repository browser-use/terminal (632 stars, last pushed 15d ago), licensed MIT. It adds 739 tokens to every session, about $0.0037 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other instructions, from other repositories
browser-harness CLAUDE.md
Instructions for browser-use/browser-harness, a project described as: Browser Harness | Self-healing harness that enables LLMs to complete any task.
core AGENTS.md
Instructions for TideSurf/core, covering tidesurf agent guide, project, architecture, commands and website design.
open-browser-use AGENTS.md
Instructions for iFurySt/open-browser-use, covering open-browser-use, 每轮开始先读, 代码改完前要读, 按任务需要选读 and 工作规则.
open-browser-use CLAUDE.md
Instructions for iFurySt/open-browser-use, a project described as: 🔮 Platform-neutral Browser Use for AI agents: real Chrome automation with a CLI + SDKs, no lock-in, dead simple.
core CLAUDE.md
Instructions for TideSurf/core, a project described as: Connect Chromium to LLM agents via token-efficient DOM compression. 50-200 tokens per page.
browser-use CLAUDE.md
Instructions for browser-use/browser-use, covering claude.md, high-level architecture, core components, event-driven browser management and cdp integration.