{% extends "base.html" %} {% block content %}
Learn the ancient craft of forging governed agents — step by step, strike by strike.
Every agent in the Forge is a blade that has already been tempered. Inspect existing blades to understand their purpose, risk threshold, and the laws (guardrails) that bind them.
Open the Arsenal →This is where the real work happens. Choose a forged blade, give it a concrete task (the "steel"), and strike. The Forge will run the full pipeline: input guardrails → LLM reasoning → output guardrails → action proposal.
The result below will show the real Strike Sequence, Flaws, and Proposed Edges — exactly like a production forging.
When an action's risk score crosses the blade's threshold, the forging pauses. A human (you) must bless or shatter the proposed edges before the agent can continue. This is the heart of governed autonomy.
Enter the Judgment Chamber →Every successful forging increases the overall heat of the Forge. Review history (when available), tune risk thresholds on your blades, and strengthen guardrails. True mastery comes from deliberate, observed repetition.