llm-tools

Experimental LLM agent environments configured via Nix flakes.

nix develop github:cbarber/llm-tools#claude-code

nix develop github:cbarber/llm-tools#opencode

Both agents include:

• Sandboxed execution (bubblewrap on Linux, sandbox-exec on macOS)
• Auto-detected language servers (Nix via nil)
• Beads task management with git integration
• Shared tooling (notify)

Agents run in a deny-by-default sandbox (bubblewrap on Linux, sandbox-exec on macOS). Access is explicitly granted per category:

• Read-write: Project directory, temp workspace, agent config/cache dirs
• Read-only: OS directories, /nix, git identity files (~/.gitconfig, ~/.ssh/known_hosts)
• Denied: ~/.ssh keys, ~/.gnupg, rest of home directory, other projects

See tools/SANDBOX.md for platform details and environment variables.

Beads provides git-backed issue tracking:

bd ready                    # Show available work
bd create "task" -p 2       # Create issue (P0-P4)
bd close <id>               # Complete work
bd sync                     # Sync with git

Auto-initializes in agent shells.

Auto-generates on first run if missing:

• opencode.json - MCP server configuration

Set ANTHROPIC_API_KEY in .env for API auth. Set AUTO_LAUNCH=false to skip auto-launch.

• agents/ - Agent environments (claude-code, opencode)
• tools/ - Shared MCP servers, scripts, and sandbox
• flake.nix - Root configuration