~/pushd/gh$pushd read google-deepmind/amplionot on the watchlist

google-deepmind/amplio

Amplio: A Lightweight Agent Harness for Robust and Long-Horizon Runs

GoApache-2.0
stars
99
forks
12
watchers
0
open issues
4
created
23 Jul 2026
last push
18h ago
GitHub ->Read from GitHub 4h ago. GitHub was busy, so this is the last good read.

README

the first lines
# Amplio: A Lightweight Agent Harness for Robust and Long-Horizon Runs * Lightweight: Multi-agent framework based on a simple step model that provides natural crash-resume capability, and agent / user / env message coordination. Agents work through a small set of generic tools: shell, file edit, sub-agent spawn, and inter-agent coordination. * Robust: DB-first persistence; crashing at any point during the agentic loop can be robustly resumed. * The agentic loop is guaranteed to be consistent, failures in the tool / environment are not handled by the harness, but reported to the agent. * Sub-agent session trees are automatically resumed upon server recovery. * Autonomous: Each run can be fully driven by the autonomous agent, or cooperatively driven with user interactions. ## Installation Amplio is a single binary with an embedded frontend. To build it yourself you need Go (see `go.mod` for the version) and Node.js 22+: make build The first build also downloads a 130 MB embedding model (cached) to build the local recall table; `make build LOCAL_EMBED=0` skips it, and recall then ranks by keyword match alone. ## Quick start

Ask a scout

a deep read
DeepSeekDeepSeek . DeepSeek V4 Pro for the deep read

google-deepmind is not in a camp, so the read goes to DeepSeek, the open source scout.

10 of 10 public reads left today. Holders get 3 a day.
No read kept yet.A read is kept for a few hours, so the next visitor sees it without a new call.

Scout calls and heat are information, not advice. They describe public GitHub activity and are never a signal to trade.