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.
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.
Local-first OSS DB/workspace where Claude Code/Codex/Cursor share durable records, docs, skills, and apps—writes that matter go through human review; Skill, MCP, OpenAPI, and ACP connections keep agents stateless and work durable.
Self-hosted SQLite stack (CLI+MCP+API+UI) tracks AI visibility over time, investigates search/traffic changes, and lets coding agents act on evidence (publish, submit, measure)—BYOK, open-source alternative to closed AEO tools.
MIT-local Python skill/CLI/dashboard coordinates Claude Code, Codex, and Cursor on one project: shared registry, ack-before-work, prove bound to file fingerprints, independent (non-author) review, and cross-provider usage/balance so green tests on the wrong code cannot close a task.
Apache-2.0 agent-native sandbox (Linux FUSE/Landlock, macOS APFS clones+Seatbelt) gives agents a confined FS view with unified text diffs, binary markers, and per-task or per-tool-call commit/abort—optional JSON-RPC adapters for Pi and policy for secrets/network.
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.
Local-first Apache-2.0 CLI for Claude Code/Codex/Cursor: mcp-xray ranks bloated tool schemas, sync compresses CLAUDE.md↔AGENTS.md↔Cursor rules, and shield fails CI on leaked keys/prompt-injection/Trojan Unicode—claims large cuts in model-visible prompt size without API calls.
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.
Google’s open AX runtime declares Task/Workspace/Model manifests on Agent Substrate: sandboxed agent jobs with git/MCP/skills pre-wired, suspend/resume, and ax ssh—aimed at high-throughput cluster runs of autonomous agents, not one laptop session.
Rust MCP server makes Claude Code the team lead over Claude/Codex/Gemini workers; Stop-hook + asyncRewake pushes replies (~50ms) into the lead session, with a local live team UI.
Show HN Weatherdeck (Oct 16): a read-only TV/phone status board fed by small pushers that shows which coding-agent lanes are waiting on you, honest progress ranges, silent sources, and deferred/owned work—not monitoring/alerts; demo mode with ship-deck theme; open source on GitHub.
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.
OSS hooks intercept tool calls before execution across ~14 coding agents; observe/enforce policy, secret cloaking, MCP/skill scan, egress allowlists, and a local dashboard.
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.
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.
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).
XERJ is an Apache-2.0 Rust single-binary AI-native search engine: `xerj autoindex` turns code/docs/logs/PDFs into ES-compatible indices; `xerj init` registers MCP + a short skill for Claude Code/Cursor; agents get passage search, hybrid retrieval, and namespaced memory instead of dumping whole files—trending hard on GitHub (~7k stars) with agent-oriented llms.txt.
Open-weight 27B decision model answers many choice/noul/score questions about one state in a single noncausal joint request (Decision Index ~68; vision ~77), with a calibrated readout head and optional fast packing path—pattern for cheap agent routing/tool selection.
Lemma Park (Show HN ~12h): paste join.md / install.sh and lend Claude Code idle cycles to open math problems (Erdős in Lean, Euclid-style primes, theorem ablations, 5×5 matmul)—agents submit machine-checked results; live board shows verified claims, canary checks, and agent-hours since launch.
Open-source stdlib-only dashboard on 127.0.0.1 merges Claude Code, Codex, Cursor, Gemini CLI, and Antigravity chats: surfaces questions/permissions/limit stalls, shows 5h/7d usage per account, and lets you reply without burning tokens just to look.
Apache-2.0 Paseo runs Claude Code, Codex, Copilot, OpenCode, Pi, Antigravity, and Muse Code in isolated worktrees from desktop, native mobile, or remote hosts—with MCP/CLI/SDK automation, plugins, and optional E2E-encrypted relay so you steer agents without changing their subscriptions.
Open-source mcp-router sits between Claude Code and many MCP servers: find_tools / call_tool / run_script stand in for hundreds of tools; on a 441-tool fleet vs six other proxies it finished 6/6 tasks at −29% tokens while peers often used more tokens or failed—Show HN + public BENCHMARK.md.