HOW-TO GUIDES
Less guessing.
More understanding.
One agent.
A complete toolkit.
Build context, access, judgment, routines, tests, and a gradual rollout plan. Pick up real templates along the way.
Open the Agent Setup ToolkitFind your next
piece of the puzzle.
Read, try, revisit.
Chapter progress stays in this browser.
- 01
Anatomy of a reliable agent prompt
Turn an attractive idea into an operating contract an agent can follow, test, and safely stop.
Open guide - 02
Connecting tools without over-sharing
Verify capability with the smallest possible permission and a harmless protected read.
Open guide - 03
Human approval for consequential actions
Design an approval moment that shows exactly what will happen and expires when the target changes.
Open guide - 04
Testing and improving a bot after its first run
Use a supervised first run, a small failure set, and stable acceptance criteria before scheduling.
Open guide - 05
Built-in connectors, MCP, browser control, and CLI tools
Choose an execution path based on available capability, permission clarity, and evidence—not product hype.
Open guide - 06
Prompt-injection and secret-handling checklist
A preflight checklist for retrieved instructions, credentials, cross-system data, and safe failure.
Open guide - 07
Work contracts and completion receipts
Define the job, limit the access, and prove the result with the smallest contract the work needs.
Open guide