Topic hub
We track coding agents as operating systems for software work: visible state, reviewable changes, bounded skills, and governance matter more than demo polish.
Continuity watch: Agent interfaces, skill packaging, usage accounting, and safe merge loops are the signals to watch.
Use this hub to follow the durable shift from chat assistants to inspectable agent work: coding loops, skill libraries, dashboards, policy gates, and reviewable automation.
Dataset context: State of Open Source AI 2026 tracks the public GitHub projects behind these weekly signals.
Developers reclaimed TVs, phones, GPUs, and local workflows while agent skills specialized and a PHP disclosure triggered exploit-heavy churn.
Agents moved below the interface into version control, reverse engineering, and creative tools, while governance and measurement lagged.
Agents moved onto devices while Rust and local-first tools challenged incumbent desktop software, with governance still trailing autonomy.
Prediction tracking for this topic is not yet available. As more weekly issues are published, accuracy data will appear here.
Developers reclaimed TVs, phones, GPUs, and local workflows while agent skills specialized and a PHP disclosure triggered exploit-heavy churn.
Agents moved below the interface into version control, reverse engineering, and creative tools, while governance and measurement lagged.
Agents moved onto devices while Rust and local-first tools challenged incumbent desktop software, with governance still trailing autonomy.
Agents spread into browsers, messaging, games, and devices, but permissions, verification, and trustworthy evaluation remain badly behind.
Agents moved beyond chat into deployment and creative production, while security and proof of adoption continued to lag.
Local typed-decision models promise cheaper, faster agents, but a coordinated Jev-shaped launch wave makes evidence more valuable than attention.
AI development shifted from generic agents toward proofs, kernels, production media workflows, and the data infrastructure needed for physical systems.
Agent tooling moved deeper into memory, skills, and orchestration while security churn and fork-heavy WARP repos distorted the signal.
Agent tooling shifted toward executable research, skills, and control surfaces while infrastructure and trust gaps sharpened.
Agent skills spread into design, security, and devices while trust, provenance, and review boundaries became the real bottleneck.
Agent work shifted from models to harnesses, skills, and control surfaces while safety and discovery trust lagged behind.
Agent skills spread into real workflows while security, reviewability, and discovery noise became the week’s hard constraints.
Agent tools moved deeper into work while security, local inference, and discovery noise exposed the missing trust layer.
Agent tooling shifted toward interfaces, skills, local control, and trust gaps while exploit and automation noise stayed high.
Agent tooling kept hardening into products while security, robotics, media skills, and coordinated discovery spam accelerated.
Agent tooling moved from experiments to packaged products while spam and abuse campaigns kept gaming GitHub discovery.
Week 28 turns agent work toward cost control, scientific workbenches, and offensive automation while spam keeps gaming discovery.
Week 27 shifts from agent packaging toward measurable inference, evaluation, and security while crypto and bypass bait keep rising.
Week 26 turns agent tooling into product infrastructure while fork-inflated crypto spam and free-Claude bait distort the leaderboard.
Week 25 marks the first real Fable ecosystem eruption on GitHub — a clustered developer response to Anthropic's Fable tier — while the deeper story is practitioners pairing model hype with cost discipline, a live AUR supply-chain attack driving genuine defensive tooling, and Apple Intelligence generating real access-circumvention repos for mainland China.
Week 24 deepens two W23 patterns — agent skills verticalization and local-sovereignty tooling — while a high-star hardware-crossover project (skylight) anchors the week's legitimate creativity and a heavier-than-usual noise floor of coordinated spam, activator repos, and crypto fraud tools demands editorial filtering.
Week 23 amplifies two W22 trends — agent memory infrastructure and skills verticalization — while a suspicious 56k-star self-hosted AI workspace, a coordinated Russian censorship-bypass wave, and the heaviest offensive-security agent activity of the year reshape what noise looks like at scale.
Week 22 delivers the clearest defensive-security signal of the year alongside a crystallising agent-skills economy — both nearly buried under the most concentrated coordinated star-farming campaign the crawl has caught.
W21 2026 is defined by two opposing forces: a maturing agent infrastructure stack — agent skills, MCP adoption, and efficient small models — and a coordinated wave of piracy, exploit, and SEO-farming repos that pollutes trending charts and makes signal extraction harder than it should be.