Topic
Source-linked workflows that name coding-agents. Part of the AI agent use cases on UsecaseBoard, or browse all use cases.
Anthropic’s Building Effective Agents appendix highlights two proven patterns—support agents that pull data and take actions like refunds, and coding agents that iterate against tests—with clear success metrics and human review.
Summer Engine (Oct 16) ships hosted MCP (https://mcp.summerengine.com/mcp), local `summer mcp`, and a full CLI so Claude Code/Codex/Cursor can generate assets, drive the engine, publish store pages, and pull analytics—`summer_search`/`ops search` map plain language to exact ops; CLI-only setup via skills with `--dry-run`/`--json` for safe agent scripting.
MCP-connected runtime QA maps critical flows and entry points, then re-tests the blast radius of each Claude Code/Cursor/Codex change against live Postgres/Kafka/APIs/MCP tools—contracts, authz, OWASP, errors—before merge, with self-heal when behavior changes on purpose.
Native macOS app orchestrates official Claude Code, Codex and Grok CLIs as a role-based crew (lead/builder/reviewer/docs/auditor/on-call) on your Mac or VPS with schedules, one-click risky-command approvals (Telegram too), and server screen/browser/terminals—no Bandito cloud for your code.
RAX (Robot Agent eXploration) lets a frozen LLM coding agent drive real robots via APIs, explore online, and compile successful perception–planning–control loops into reusable programs—no teleop demos or policy training; +37.4% avg vs baselines in sim; open skill libraries per embodiment.
SuperCompress v2 compresses context against the query (keeps answer-critical lines verbatim): ~64% mean cut on their B5 coding-agent suite with 24/24 evidence-retention passes; MCP/plugin for Cursor/Claude Code/Codex plus hosted API and open weights.
JetBrains Junie Local ships native support for NVIDIA RTX Spark (up to ~55 tok/s in early tests, auto-sized context, FP8 + speculative decoding) and moves Windows Ampere+ 24GB RTX from experimental to stable—free local coding agent with no token charges and code that never leaves the machine.
US lab Reflection open-sourced Beam (501B MoE / 23B active, 1M context, Apache 2.0) for coding and long-horizon agents, with day-0 vLLM/SGLang/Transformers support, Mirror CLI harness, and 40+ inference partners including OpenRouter and Ollama.
AWS Labs open-source (5k+★): one harness-neutral core turns Claude Code/Codex/Cursor/Copilot/Kiro/opencode into structured, verifiable delivery workflows via `aidlc config --harness` + doctor—install script, not a new agent.
PATH-fronted proxies for terraform/aws/kubectl: TOML policies allow plan/read but block destroy/terminate/recursive deletes before the real binary runs—no LLM judgment in the safety path.
SDK + stdio MCP server run agent()/parallel()/pipeline() scripts that fan out to Claude Code, Codex, OpenCode, or pi via ACP, with journaling, resume, and git-worktree isolation.
Open team chat where you @-mention Claude Code/Codex/Pi/Goose in one Room; agents open corners to plan/build/test and bring a single approve-merge decision—multiplayer harness instead of single-player agent sessions.
Local-first npm app: isometric office UI where each Claude Code session gets a desk; red windows flag waits for you, with tmux integration—watch many agents without scrolling terminals.
Open-source local procedural memory that learns only after tests pass, then hands Claude Code/Codex/Cursor a short verified skill note on similar tasks (reported token/tool cuts in published benchmarks).
Local gate: one policy set so Claude Code/Codex/Cursor check with Paveo before shell or file write; dangerous actions refused, every decision logged—no network required.
Open-source CLI+MCP (morluto/rea) lets Claude Code/Cursor inspect binaries, Electron/.NET/apps via Hopper/Ghidra/IDA, returning evidence-backed observations so agents can rebuild features from behavior without copying source.
macOS menu-bar app that finds npm/dev servers and daemons started by Claude Code, Cursor, Codex, OpenCode, or T3 Code—including ghosts after the agent exited—and kills the process tree in one click.
OSS (AGPL): index a folder of Markdown notes into local SQLite, then expose search/recall/remember tools over MCP so Claude Code/Codex reuse decisions across sessions; hosted option at renor.cloud.
Open pipeline turns a task.md into idea→experiments→ablations→sealed eval→paper, orchestrating coding agents (Docker/yolo by default) for automated scientific co-discovery.
MIT Tauri app runs Claude Code/Codex/OpenCode/Antigravity/Grok Build side-by-side with approvals and session resume; keep agents on the PC, then reconnect to the same live terminals from phone via your own relay/LAN/VPN.
CLI reads Claude Code transcripts to show per-turn context cost—what ships before the first keystroke, unused payload, and skills that never reached the agent—so you can trim defaults instead of guessing.
Beta plugin sits between the model and your app: write hard-deny/soft-deny/allow rules in prose; Armor evaluates every tool call (powered by a decision model) and blocks policy violations before bash/git/exfil runs—Claude Code auto mode works via Ori on any OpenRouter model.
Open method: Blast/Mid/Inner write at system/subsystem/local altitude, then fixed T1 structural → T2 bug/ops → T3 adversarial gates with handoff packets—agents get AGENTS.md + prompt files, not a vague “please be careful.”
Anthropic’s PR Code Review left research preview: auto-reviews on new GitHub PRs in ~6–10 minutes, with Quick/Standard/Deep intensity per repo for Team and Enterprise.
Anthropic took Opus 5.5 fast mode out of research preview: set `speed: "fast"` (no beta header) for up to 2.5× OTPS at $8/$40 per MTok; separate rate limits; Claude Code `/fast` for interactive loops.
Oct 15 Every deep dive (Shipper + six OpenAI interviews): Cloud flop → Rust local agent → no-editor desktop app → ChatGPT merge; learnable arc on form-factor bets, separate model lines, and agent-native UX.
Anthropic’s Claude Code team documents a layered review stack—pre-PR verify, AI review + security block, risk-tiered human approvals, merge queue/flags, and fast rollback—so most agent PRs merge with ≤1 human while revert rate fell >40% as volume doubled.
Oct 15 open weights: multimodal MoE (816B total / 42B active, 1M context) trained for long-horizon coding agents—plan, tool-call, self-correct—with Muse Code demos shipping real iOS changes on-device; HF + API hosts.
Empirical study: 34% of popular active repos ship agent instruction files (AGENTS.md dominant); half exceed ~100 lines; ~3/5 rules could be script-checked but <1/20 name the check—actionable lint: short map, name enforcers, one rulebook.
AISLE October rollout (Oct 14): MCP server puts exploit-proven findings into Claude Code/Cursor so agents fix in-editor; plus an AI package firewall that intercepts npm/PyPI/Cargo installs from humans and coding agents, with cooldown and monitoring mode before enforcement.
inclusionAI/Ant Ling open-sourced Ling-3.1-flash (560B total / ~25B active, 1M context, MIT) for agents, office, search, and SWE—day-0 SGLang/vLLM recipes, tool/reasoning parsers, and strong Terminal-Bench / SWE-bench Pro / SkillsBench numbers; works in Claude Code, Hermes, OpenClaw, and related harnesses.
HN (Ted Haley): Pensieve locally indexes code, docs, and past agent sessions (watches repos/worktrees/uncommitted edits), exposes MCP search so Claude Code/Codex/Cursor get the few relevant passages with file:line instead of multi-round grep; Qwen3-Embedding-0.6B + SQLite/trigram + turbovec; macOS-focused OSS.
Anomaly OpenCode 2.0 (Oct 12; changelog still shipping 2.0.x Oct 16): auto-started local background service shared by TUI/web/desktop; rewritten API + @opencode/client; @opencode/codemode turns MCP/OpenAPI into scriptable functions (all MCP servers default to Code Mode); background subagents + async shell; 80+ hot-reload plugins; @opencode/sdk for embed. Upgrade: `opencode upgrade --major`.
Launch HN: Agent37 hosts Hermes/OpenClaw/Claude Code/Codex/OpenCode/Pi (or your Docker) in per-tenant gVisor sandboxes; prepaid compute meters by the minute (2 vCPU/4 GB ≈ $4.76 for 730 awake hours), disk-only when asleep (~$0.36/mo for 4 GB), $5 starter credit, free personal tier, managed MCPs + LLM gateway; avg customer spend cited at $0.37/sandbox/mo.
Almas Akchabayev built Deck to keep his Claude Code and Codex sessions in persistent VS Code tabs grouped by Git worktree.