Moli is a Rust-based headless browser that gives AI agents a complete browser runtime for retrieving pages, extracting content, searching the web, and automating browser tasks. It is for agents and developers who need JavaScript, DOM, CSS, networking, storage, rendering, screenshots, or browser automation through CLI, CDP, or WebDriver protocols. The catalogue contains skills and instructions for using Moli.
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 skills add lexmount/moli --skill moli-websearchgit clone --depth 1 https://github.com/lexmount/moliWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/lexmount/moli/moli-websearch)<a href="https://agentmods.dev/skills/lexmount/moli/moli-websearch"><img src="https://agentmods.dev/badge/skills/lexmount/moli/moli-websearch/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/lexmount/moli/moli-websearch"><img src="https://agentmods.dev/badge/skills/lexmount/moli/moli-websearch.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What 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 5.1 | $0.00049 | $0.01437 |
| Opus 5.5 | $0.00020 | $0.00575 |
| Sonnet 5.5 | $0.00010 | $0.00287 |
| Haiku 4.5 | $0.00005 | $0.00144 |
Grade A, and why
moli-websearch scanned grade A with 1 finding 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 11d ago.
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.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
curl --proto '=https' --tlsv1.2 -fsSL \ How it starts
The opening of the file, as written. The whole thing — 121 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Search the Web with Moli
Search by keyword or image and verify source pages with Moli throughout.
Parallel Execution
- Batch first. Launch 4–6 independent searches together, covering at least 3 suitable engines when available. For broad requests, split the work into independent subquestions and keep 8–12 searches and source fetches in flight. Honor explicit source/budget limits and available work. Regional aliases and DuckDuckGo HTML/Lite count as one engine.
- Dispatch before waiting. Use parallel tool calls or a subprocess pool; refill slots as results arrive. Deduplicate promising URLs and fetch source pages concurrently, without waiting for the slowest search. Keep errors and outputs separate per job so one failure cannot abort the batch. Limit each engine/source host to 2 concurrent requests; reduce load on rate/resource limits.
- Recover across engines with Moli. On CAPTCHA, access errors, timeouts, empty shells, or irrelevant results, queue another suitable engine or route while other jobs continue. Use the full engine tables when needed. Stop expanding once the requested coverage has been verified or the task budget is reached; report remaining gaps.
- Keep tool changes explicit. A single blocked engine is not a reason to switch to built-in search. Switch only if the user requests it, Moli cannot run after setup, or two diversified batches for an unresolved question yield no usable results. Disclose the attempts and tool change; honor Moli-only requests by reporting blockers instead. Track actual Moli search calls separately from source fetches so execution counts can be reported accurately.
Fast Path
These routes returned relevant results locally. Choose several for the first
batch by language and input; row order is not priority. Availability varies
with network and session. Encode the query as Q or the public image URL as
ENC once with encodeURIComponent(...).
Text search
| Engine | Search URL |
|---|---|
| Brave | https://search.brave.com/search?q=Q |
| DuckDuckGo HTML / Lite | https://html.duckduckgo.com/html/?q=Q / https://lite.duckduckgo.com/lite/?q=Q |
| Yahoo | https://search.yahoo.com/search?p=Q |
| Baidu | https://www.baidu.com/s?wd=Q |
| Shenma | https://m.sm.cn/s?q=Q |
| Toutiao | https://so.toutiao.com/search?keyword=Q |
| Sogou Weixin | https://weixin.sogou.com/weixin?type=2&query=Q |
| Naver | https://search.naver.com/search.naver?query=Q |
What ships with it
6 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 11d ago Changed · -1 lines 9a7fb3145676
- 1mo ago First seen · 122 lines · 49 tokens per session scan A e059a5c6c563
moli-websearch is a skill published in the GitHub repository lexmount/moli (18,671 stars, last pushed today), licensed Apache-2.0. It adds 49 tokens to every session and 1,437 once invoked, about $0.0002 per session on Opus 5.5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-10.
Other skills, from other repositories
playwright-rs-usage
Procedural reference for using playwright-rs in Rust browser-automation code — object model (Browser/Context/Page/Locator), the locator!() macro, builder pattern for options, auto-wait semantics, adding the crate and installing its browsers, and how to capture / inspect traces for failure diagnosis. Use when writing…
browse
Use the browse CLI for Browserbase browser automation, Browserbase cloud APIs, Browserbase Functions, templates, web fetch/search, diagnostics, and Browse.sh skill discovery/installation. Use when the user asks to navigate pages, inspect browser state, run local or remote browser sessions, manage Browserbase…
stagehand-facade
Browse, navigate, inspect, screenshot, and automate websites with the exact fx tools mcpstagehandrun, mcpstagehandsnapshot, and mcpstagehandscreenshot.
webcmd-browser
Use when a live browser task requires Playwright interaction, authenticated handoff, visible UI verification, or ad-hoc page inspection.
actionbook-web-test
Run browser-based web tests against websites using Actionbook CLI. Activate when the user wants to test a website workflow, run smoke tests, verify a user flow, check if a web application works, run regression tests, or validate browser-based interactions. Supports test definition, execution, assertion, reporting, and…
active-research
Deep research and analysis tool. Generates comprehensive HTML reports on any topic, domain, paper, or technology. Enhanced with advanced browser automation — SPA handling, network idle wait, batch operations, stealth browsing, and intelligent page analysis. Use when user asks to research, analyze, investigate…