Skip to content

v0.3.5

Choose a tag to compare

View all tags
@zk-xyz zk-xyz released this 06 Apr 05:36
v0.3.5
0e6c241

Session progression improvements based on real agent testing feedback.

Improved

  • Session 2+ pitched by runtime payoff, not checklist. After Session 1, agents are told what Session 2 adds to brand-runtime.json ("agents will reject off-brand layouts and enforce visual consistency") rather than just "run brand_deepen_identity."
  • Clarification gate removed. Clarification items no longer block Session 2. The conversation guide presents both in parallel — resolve clarifications AND consider visual rules — instead of gating one behind the other.
  • brand_write gap surfacing. When content is generated with only Session 1 data, warnings explain what the runtime is missing concretely ("agents would know not just the right colors but how to use them").
  • Report session timeline. Each session description in the HTML report now references the specific artifact it adds to brand-runtime.json.

The core insight from testing: agents stop when the task is done, not when the product wants them to continue. Sessions 2-4 need to feel like they unlock something concrete, not complete a checklist.

Assets 2
Loading