Definition
Plain language
A layer that hands an AI agent relevant how-to notes when it starts a new task.
As stated in the literature
In Life-Harness, the layer that retrieves procedural guidance — reusable workflows and operating notes distilled from training trajectories — into the agent's context at the start of a task; the fourth of the harness's four lifecycle layers.
Why it matters: It lets an agent reuse hard-won know-how from earlier tasks instead of rediscovering the same workflow from scratch every time.
For example, when an agent starts booking travel, this layer slips in a saved note reminding it to confirm the dates before paying, drawn from past runs.
Heard on the show
“And a Procedural Skill Layer that retrieves relevant notes distilled from training trajectories when the agent starts a new task.”Episode 071 — When the Model Is Fine and the Plumbing Is Broken: Fixing Agents at the Interface