Fetching from the wire…
01
02
03
04
05
06
07
08
OSS2026-08-16 · source-backed
ShawnPana/phone-harness (created August 7, MIT, Python, 170 forks) lets an agent drive a real device (GitHub). Star velocity is roughly 200 per day, well above the 50/day threshold that usually precedes press coverage. It arrives alongside browser-use at 109,380 stars as the second surface agent harnesses are being pointed at, and it inherits the identical unresolved problem: no meaningful sandbox exists between an agent's tool call and a device holding your authenticated sessions.
Each link below shares sources, entities, or timing with this story.
A Chinese lab shipped a runtime that manages two American coding agents as subagents, and it went from repo creation to 145,439 stars in four days. deepseek-ai/deepseek-harness published dsh-v0.1.0-rc.7 at 12:01 UTC today, its first tagged release since the repo appeared on Au...
440 stars, created August 11, covering 12 sections from model APIs through structured output, RAG, evals, agent loops, LoRA versus fine-tuning, security, LLMOps and serving, plus three case studies and a capstone (GitHub). The stance is that you write the agent loop, RAG and e...
cbrock84/headcount appeared August 28 and reached 482 stars in two days. It structures an agent organization as a company: a chief executive over 16 departments holding 143 skills, where each department is an independently installable plugin so a project loads only what it nee...
omaekumiko2-create/kru, created August 24 and at 88 stars, exposes passwords, API keys, SSH identities and TOTP codes to an agent over MCP without the plaintext entering the transcript. It targets the specific leak path where secrets loaded for a tool call end up in transcript...
Three moves, two days, no coordination between them. August 10–11: GitHub shipped Ollama as a BYOK provider inside Copilot for JetBrains (GitHub Changelog). Unsloth released Unsloth Desktop with a command literally named unsloth start claude, which points Claude Code and Codex...
mikehasa/agentacct (547 stars, MIT, Python, pushed August 7) breaks each agent task into work steps: tools used, files changed, tests run, time and tokens spent, across Claude Code, Codex, OpenCode and others in a local-first dashboard. The no-telemetry positioning is the diff...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.