What problem does it solve?
Long or multi-agent engineering sessions often generate repeated sequences of prompts, handoffs, and outputs that are never formalized, causing wasted manual repetition and inconsistent results. Dev Learner identifies those recurring patterns and converts them into discoverable, reusable skills so teams can automate repeatable developer workflows.
Core Features & Use Cases
- Pattern Detection: Scans conversation history for repeated workflows, prompts, outputs, or agent handoffs that appear three or more times.
- Structured Proposals: Generates a named skill proposal with triggers, step-by-step workflow, involved agents, and inputs/outputs for each detected pattern.
- Creation Hand-off: When approved, pairs with a skill-creator to automatically generate the new skill file or suggests creating a new agent for broader workflows.
- Use Case: During a sprint retrospective, extract a frequently repeated PR review flow into a single
dev-pr-review skill to standardize reviews and save time.
Quick Start
Run dev-learner on the active conversation to detect patterns that occurred three or more times and propose skill extractions.