trace-climb

What problem does it solve?

Trace Climb prevents valuable lessons from disappearing after messy tasks by converting them into an inspectable, propagation-ready learning trace.

Core Features & Use Cases

• Trigger-aware post-task reflection: Selects a record class (required, recommended, or skip allowed) based on whether the work exposed durable lessons worth preserving.
• Evolution Record capture: Captures origin trace, optional activation trace, observed outcomes, missed assumptions/guidance, structural gaps, and proposed evolution.
• Propagation decision enforcement: Ends every durable record with an explicit choice (update canon now, update operational surfaces now, open ADR path, defer with named boundary, or no-op with reason) to avoid “learning captured but not propagated.”
• PR-visible learning trace: Produces a bounded summary or PR link that propagates learning to where it can be acted on next.

Quick Start

After a messy or non-trivial task, tell your agent to run Trace Climb and write an Evolution Record using the template, ending with a clear propagation decision.