What problem does it solve?
Teams lose time rediscovering prior work and repeating solved problems because institutional knowledge is scattered across wikis, GitHub, and Discord. This Skill searches Marin's Echo knowledge base before starting work and captures reusable findings as durable wiki notes.
Core Features & Use Cases
- Federated Search: Query wiki notes, files, PRs, and issues with natural language via
infra/echo/cli.py search, with optional Discord and exact-string grep modes.
- Result Grading: Fetch full result details by source ID and grade evaluated results with feedback to improve future search quality.
- Knowledge Capture: Create or revise Open Knowledge Format wiki notes with actionable titles,
use_when triggers, and canonical URL citations.
- Use Case: Before diagnosing a stalled TPU collective, search Echo for prior incidents, cite the canonical evidence URLs, and write a wiki note documenting the reusable diagnostic pattern.
Quick Start
Ask the assistant to search Echo for prior work on your current Marin task and cite the relevant canonical URLs before writing any new documentation.