Context.devOne API to scrape, enrich, and extract the internet
Promoted
OpenTrade
Open-source trading harness for Claude Code / Codex.
230 followers
Open-source trading harness for Claude Code / Codex.
230 followers
OpenTrade is an open-source harness for Claude Code / Codex agents that provide the tools and guardrails to trade effectively via Robinhood's official MCP. Out of the box, agents can setup cron schedules, custom scripts to notify themselves, and persistent background sessions – all on your machine.
Congrats on the launch! Running the agents locally while keeping per-agent order approvals feels like the right approach for something as sensitive as trading.
Curious how you handle conflicts when multiple agents want to act on the same position at the same time?
OpenTrade
@alpertayfurr Thanks Alper!
Aside from Robinhood's native order execution (i.e. which would guarantee only one order being filled from conflicting orders), we don't do any sort of position locking. Nothing stops two agents from placing individually valid orders on the same position at the same time.
That said, we've been exploring ways to divvy up the portfolio and enforce that at the OpenTrade level (position-wise or otherwise). Do you see a use case for it?
@pranavnair Yes, especially when agents run different strategies or time horizons. I’d probably give each agent a capital/exposure budget and enforce portfolio-level limits before execution, so two individually valid orders can’t accidentally stack into an oversized position. That seems safer than hard position locking while still letting agents operate independently.
OpenTrade
@alpertayfurr That sounds really interesting. Would you mind opening an issue on our Github? We can continue the discussion there.
The idea is compelling, but for me trust will come down to the guardrails and audit trail. If I can clearly see what the agent tried to do and why, I’d be much more willing to test it.
OpenTrade
@elsie_ramsey The guardrails and order history features are live. Given the chat history is persisted for all agents, do you see value in a dedicated audit trail vs reading the agent's direct output?