Station 1 · never used an agent · read-only
The Honest Briefing
Published to read
One sober conversation, before any building, about what an autonomous trading agent is actually good at — and what no one has proven it can do.
Most of this industry sells you an agent that "makes money while you sleep." The evidence doesn't support that, and a venue that lets you believe it is setting you up. This briefing is the foundation the rest of your journey stands on: you'll build an agent for the right reasons.
The prompt
Before I build anything, brief me honestly on autonomous trading agents. 1. What are agents demonstrably good at on a venue like this? (Think: discipline, monitoring, rule-execution, never sleeping, never revenge- trading.) 2. What has NOT been shown to work? Be blunt about the evidence on agents predicting prices profitably. 3. Pull the venue's live fee ladder with gryps_venue_snapshot and then run gryps_edge_check for a hypothetical small trade at the worst tier. Explain what comes back as one number I can hold onto: what does a trade have to earn, in basis points, just to break even here? Say where every figure came from and when it was read. 4. Given all that, what is a realistic, honest ambition for my FIRST agent? Don't soften it. I'd rather be told the truth now than by my balance later.
Published to read
This prompt is written for the Gryps public server, which is not on npm yet. The package you can install today, gryps-agent-mcp, does not have gryps_venue_snapshot. The station is published so you can read what it will ask of you; it opens the day the public server ships.
What good looks like
A clear split: agents excel at *operating* (rules, vigilance, consistency) and are unproven at *predicting*. A live, dated fee ladder and a break-even bar from the edge-check tool — framed as "the bar every trade must clear." A first ambition like: an agent that watches well, keeps you inside your own rules, costs you nothing, and teaches you the venue.