oo-agent
the execution layer for agents that work with your expertise. each agent is a working likeness of how you think and decide. built-in tools plus any tool via mcp, real-time execution, and persistent memory that stays yours.
what your agents can do
the agent layer that turns expertise into action
any model
gpt, claude, gemini, mistral. use the best model for each task. switch mid-conversation if needed.
28 builtin tools
file management, web search, code execution, browser automation, memory queries. everything your agents need to do real work.
real-time streaming
watch your agents think and work in real time. see tool results as they happen. no waiting for a final answer.
tool categories
28 builtin tools organized by what they do
files & code
- • read, write, edit, search files
- • run python and javascript in a sandbox
- • shell access for git, build tools, and clis
- • code analysis, refactoring, debugging
web & research
- • web search and content extraction
- • visit websites and fetch remote media
- • deep research with specialized sub-agents
- • image search and analysis
memory & creative
- • recall past conversations and decisions
- • store knowledge with automatic organization
- • generate images (dall-e, flux)
- • find relevant context across all memory
how your agents work
smart execution backed by persistent memory
memory-backed decisions
your agents check past projects and decisions before responding. no need to re-explain context every time.
gets smarter the more you use it
specialized sub-agents
complex tasks get broken into parts. a researcher gathers info while a coder writes code. each knows who did what.
parallel execution for faster results
secure execution
code runs in isolated sandboxes. trusted operations execute directly for speed. you choose the balance.
speed when safe, isolation when needed
workspace isolation
each agent is scoped to its workspace. client a's agents cannot access client b's data. ever.
enforced at every layer
pricing
transparent, usage-based