SuperAPI AI Directory
Partner with us中文AI newsFavorites

AI agents

2,941 results · Page 38 of 82

Browser Cash

AI agents

Browser Cash is a scalable browser automation platform for AI agents, also used for web scraping and internet intelligence gathering. By having users install a browser extension, it connects ordinary browsers to a distributed network, turning them into nodes where AI agents execute web tasks. AI can access websites, perform click actions, and collect public information within isolated sandbox sessions. After users join the network, they earn points that reflect online duration, number of completed tasks, and participation quality, and can be redeemed for access to the AI system, with future conversion to ecosystem tokens. The platform emphasizes security and privacy by design; sensitive operations such as private tabs, passwords, and banking access are blocked by default, and users can view, pause, or shut down nodes at any time.

Read more

browser-extension

AI agents

Taxy AI is an open-source browser extension that uses large language models such as GPT-4 to control the browser and perform repetitive operations on behalf of the user. The user only needs to enter an ad hoc instruction, and the extension can automatically complete interactions such as clicking and typing. It is currently in a research preview stage, and many workflows may fail or confuse the agent. The extension is fully open source, does not send page content or instructions to its own servers, and stores API keys securely locally in the browser. It supports use through a popup window or a developer tools panel, and must be built and installed by the user.

Read more

BrowserGym

AI agents

BrowserGym is a Gym environment framework for web task automation open-sourced by ServiceNow, aimed at accelerating web agent research. It provides an open, easy-to-use, and extensible framework, with built-in benchmarks such as WebArenaVerified and WebLINX, and supports customizing new benchmarks by inheriting from AbstractBrowserTask. BrowserGym is not a consumer-facing product and should be used with caution. The framework works with AgentLab to implement, test, and evaluate web agents across multiple benchmarks, and provides a demo agent based on GPT-4V.

Read more

bunkhouse

AI agents

Bunkhouse is an open-source AI employee platform for small and medium-sized main-street businesses. It does not treat AI as prompts in a chat window, but instead manages agents as real employees: assigning them a real email address under the company domain, job responsibilities, governed processes, a monthly token budget (salary), readable and editable memory, and a position in the org chart. Coworkers do not need to learn new software; they can simply email the agent to collaborate, and email threads become the primary audit anchor. The platform provides chat, voice calls, and an observable workbench sharing the same workspace, where you can view the agent's desktop, browser, terminal, files, and persistent history. Only Docker is required to launch PostgreSQL, Redis, the web app, Worker, the remote computer gateway, and other components with a single command. The project is licensed under AGPL-3.0, with role packs under MIT, and is currently in alpha, without an independent security audit yet.

Read more

BurnRate

AI agents

BurnRate is a monitoring and analytics tool for AI usage costs, helping users view spending across model providers such as OpenAI and Anthropic in one place, set budgets, and catch budget overrun risks. Its cloud dashboard can be started for free, supports connecting provider reporting accounts or importing aggregated usage, and offers an optional local CLI (command-line tool) to obtain detailed data at the session, project, and agent levels. The official emphasis is privacy first: cloud reporting does not request prompts, responses, or source code, and local CLI data remains on the user's machine. It supports analysis for coding tools such as Claude Code, Cursor, OpenAI Codex, GitHub Copilot, Windsurf, Cline, and Aider.

Read more

caesar-agent

AI agents

Caesar is an open-source autonomous AI research agent positioned as an open-source alternative to ChatGPT Deep Research, Perplexity, and Gemini Deep Research. It performs topological traversal through graph-based deep web exploration and combines adversarial answer synthesis (multiple rounds of drafting, self-challenge, merging) to generate novel and grounded answers. The project is maintained by jasonzliang/caesar-agent and is licensed under the Apache License 2.0. According to the materials, blind-test LLM reviews in the paper (Liang, Meyerson, Miikkulainen, 2026) show that it leads in creativity dimensions and maintains an advantage in controlled experiments with a budget of about $5; a 23-person human evaluation also independently corroborates its preference. It is suitable for open-ended, interdisciplinary deep research, such as hypothesis generation, literature review, arXiv citation graph mapping, competitive intelligence, etc.; it is not suitable for quick fact queries or latency-sensitive scenarios.

Read more