coding-agents

BoundaryBench is an open-source benchmark for coding agents running under hardened sandbox policies. It compares harnesses like Claude Code, Codex, Terminus 2, and Grok on Terminal-Bench tasks and includes quickstart, policy controls, and result export tools.

Open-source CLI for searching past coding agent sessions locally and tracing code changes back to the original agent transcript. It also offers a paid pro add-on for “git blame, but for agent sessions,” with cited results and local-only indexing.

compactdiff snapshots, diffs, replays, and branches AI coding-agent sessions locally. It reads Claude Code and Codex logs natively, redacts sensitive data by default, and exports shareable replay HTML or forked JSONL branches.

ReactBench is a benchmark for evaluating coding agents on realistic React work, with published scores, cost comparisons, and example tasks focused on production-grade frontend issues.

Mindwalk is a local Go tool that replays Claude Code and Codex sessions on a 3D map of a codebase, showing where an agent searched, read, and edited files over time.

Looma is a local-first Python CLI that turns coding-agent history into resumable project context. It reconstructs work items, decisions, blockers, linked commits, and next steps from Claude Code, Codex, and Cursor sessions, with optional local LLM extraction and no cloud or API keys.

A local-first tool that mines reusable skills from coding agent transcripts and turns approved patterns into SKILL.md files for future sessions.

A research prototype for a BitTorrent-style mesh where coding agents share idle capacity, delegate tasks, and settle credits across peers.

Matrix OS is a cloud computer for background AI agents, offering persistent hosted workspaces with terminals, repos, previews, and agent sessions that keep running after your laptop closes.

Deterministic, MCP-first browser automation runtime for coding agents. It exposes a public stdio MCP server with 50 browser tools, direct CDP control, and no API key or LLM fallback in the core path.

DumbDetector is a community-driven status tracker for AI tools, assistants, and coding agents. It shows report-based signals for services like Claude Code, Codex, Cursor, GitHub Copilot, Gemini, and more, plus short troubleshooting guides for common AI behavior issues.

Supercov is a coverage tool for coding agents that turns uncovered paths into the next test to write, then reruns the suite to verify progress.

Shed is a GitHub-hosted CLI for managing read-only repo catalogs and disposable writable workspaces for terminal coding agents. It syncs fresh checkouts, supports parallel agent sessions, and integrates with Claude Code, Cursor CLI, and opencode.

Pier gives each coding-agent session its own cloud VM on the user’s AWS account, with parking, resuming, and repo setup built in.

Forkbench is a native Mac control room for running multiple coding agents in parallel, each in its own git branch, from a single board. It adds shared coordination, notes, secrets handling, worktree controls, and remote oversight for supervising agent-heavy workflows.

Open-source memory layer for coding agents with fast search, MCP recall, auto-context, secret redaction, stats, and sync over SSH or shared folders.

Local browser-control tool that lets coding agents drive a signed-in Chromium browser through a CLI, daemon, and extension.

Ouijit is an open-source Git worktree-based task and terminal session manager for CLI coding agents, with integrated terminals, lifecycle hooks, session-aware commands, and basic VM sandboxing.

A Go-based local orchestration platform for AI coding agents, with live event streaming, session replay, and durable execution history.

Nimbalyst is a visual workspace for building with Codex, Claude Code, and other coding agents. It combines collaborative editors, session management, task tracking, git/worktree tools, and a mobile app for managing agent work on the go.

A source-linked public archive of documented incidents involving AI coding agents and chatbots, with summaries, lessons learned, and an incident index.

A macOS feedback layer for coding agents that captures comments on text, screenshots, web elements, and voice, then passes them to agents through MCP with attached context.

An open-source upload command for coding agents that stages screenshots on a branch and auto-attaches them to GitHub PRs as a tidy comment. It also supports tagging, finding uploads later, device frames, one-shot screenshot capture, and shareable galleries.

A small Linux CLI for running coding agents in a sandboxed environment. The repository shows installation options, TOML-based configuration, supported agents, and troubleshooting notes.