What the AI ecosystem is building right now
GitHub-frontier intelligence for local AI. Hand-curated trending repos with the architectural significance behind each — every entry answers “why should an operator care?” rather than just “here are some popular repos.” Five zone pages categorize by ecosystem domain; this index surfaces velocity across the whole frontier.
Top velocity this month
The 6 repos adding stars fastest in the last 30 days, across all categories. Stars-per-month is a rough proxy for ecosystem attention; combined with editorial significance below it tells you which projects to actually pay attention to.
Crossed 350k GitHub stars in early April 2026 — fastest community-growth curve ever recorded for an open-source agent product. Founder joined OpenAI; project transitioned to foundation governance. The defining signal that open-source autonomous coding agents have caught up to (and on some workloads, surpassed) closed-source flagships.
Architecture: Anthropic-style reasoning loop (thinking → planning → execution decomposition) instead of ReAct-style tool-call loops. MCP-first dispatcher treats built-ins as MCP servers internally. The architectural break that explains the velocity advantage on complex tasks.
DeepSeek's reasoning model — explicit chain-of-thought emission as the architectural primitive. The R1 Distill Qwen 32B variant captures ~80% of the full R1's reasoning at 5% of the VRAM, fits an RTX 4090 in AWQ-INT4. The reason local-reasoning workloads became viable in 2025-2026 at all.
Crossed 113k stars by May 2026 — fastest-growing managed-crawler MCP. Handles JS rendering, anti-bot evasion, and large-site map+scrape jobs at scale. The pragmatic upgrade from mcp-server-fetch when an agent needs to crawl thousands of pages or work against JS-heavy SPAs.
Architecture: Cloud-managed rendering pipeline + thin OSS MCP client. The architectural tradeoff: outsource rendering complexity for crawl-volume scenarios.
The longest-track-record open-source autonomous coding agent. Planning Mode in v1.6+ closed the 'agent loops without making progress' gap that plagued earlier products. Currently the most stable production deployment in the category — pick OpenHands when reliability matters more than velocity (compare to OpenClaw).
Architecture: ReAct-style loops, MCP-first tool dispatcher, sandbox executor with Docker / chroot / native modes. The /stacks/local-coding-agent canonical recipe is built around OpenHands.
Anthropic's reference MCP server collection. Filesystem, Git, Memory, Postgres, Brave Search, Fetch, Sequential Thinking. The canonical implementations every third-party MCP server gets compared to. Anthropic-maintained; 97 million installs across the ecosystem (April 2026).
Architecture: Each server is a separate stdio process; the protocol is JSON-RPC 2.0 with explicit lifecycle + capability negotiation.
The Qwen team's reasoning-toggle generation. Native <think> reasoning blocks; toggle per-query. Qwen 3 32B AWQ-INT4 at 36.5 tok/s on RTX 4090 makes serious reasoning workloads viable on consumer hardware. The architectural shift: reasoning quality is now a configurable parameter, not a model choice.
Architecture: Toggle-style reasoning means you don't pay the reasoning-token tax on simple queries. The right pick when workload mix is mostly chat with occasional reasoning needs.
Zones
Each zone covers a single ecosystem domain. The category labels line up with /maps/* — frontier is the “what's happening” counterpart to those maps' “what's the landscape” framing.
Trending this week
What's exploding right now — sorted by 30-day star velocity. The ecosystem-acceleration view.
Coding agents
OpenHands, OpenClaw, Goose, Aider, Cline, Continue — the autonomous coding agent category, in motion.
Inference runtimes
vLLM, SGLang, Ollama, MLX-LM, Exo, Petals — what's shipping in the runtime layer.
Memory systems
Letta, Mem0, Zep, Graphiti, MCP-memory — agent-memory frameworks gaining or losing momentum.
MCP ecosystem
The Model Context Protocol surface — Anthropic reference servers, vendor-maintained, community standouts.
Model releases
Recent open-weight model releases, by category. Reasoning, coding, multimodal, MoE, edge — connected to catalog with authority badges (L1.25 / benchmark / verdict).
How this layer works
The frontier layer is hand-curated. The owner triggers snapshot updates via the admin panel; the seed file (scripts/seed/frontier.ts) is the source of truth. There is no autonomous scraping — autonomous trending ingestion produces sludge; editorial discipline produces intelligence. Every entry must answer “why should an operator care?” before it lands here.
When a frontier repo graduates to a full catalog entry (gets a /tools/[slug] page with operator review), the card here links to the operational review rather than the bare GitHub URL — the frontier layer is the discovery surface, the catalog is where depth lives.
Going deeper
- Ecosystem maps — the structured-landscape counterpart to frontier's momentum view.
- Catalog — where graduated frontier entries land with operational review depth.
- Execution stacks — recipes that combine frontier + catalog into shipping architectures.