24 supported coding agents (and counting)

Run yolonot setup once to auto-detect installed coding agents and wire hooks. Every agent shares the same .yolonot repository policy, linked git worktree session memory, SHA-256 script cache, and judge pipeline.

1. Agents with native interactive Ask support (18)

These agents support three-state decisions (allow, interactive ask confirmation, and deny). When you approve a command or script once in any of these tools, its SHA-256 approval is remembered across your project session:

Coding Agent	Hook Lifecycle & Special Capabilities	Rule Import (`rules discover`)
Claude Code	`PreToolUse` + `PostToolUse` Optional `--mod` adds native in-TUI risk badges & replay bindings	✓`.claude/settings.json` + autoMode
OpenAI Codex CLI	`PreToolUse` + `PermissionRequest` Paired calls deduplicate via `.codexpre` so judges never run twice	✓`.codex` rules
OpenCode	V2 `permission.asked` / `permission.replied` + V1 plugin Full interactive ask on V2; binary guard on V1	✓`opencode.json`
Cursor	`beforeShellExecution` + `afterShellExecution` Protects both IDE agent tab and Cursor CLI	✓`.cursor/cli.json`
Google Antigravity	`PreToolUse` + `PostToolUse`	✓`rules discover` + `.yolonot`
GitHub Copilot CLI	`PreToolUse` + `PostToolUse`	✓Copilot settings
12 Additional Native-Ask Agents Amp, Pi, Qwen, Grok, DeepSeek, Kilo, Devin, ZCode, Muse, Hermes, OpenClaw, Mirror	Unified `baseHarness` `PreToolUse` + `PostToolUse`	✓Host config via `rules discover`

2. Binary hook agents without native Ask (6)

Six supported agents expose hook protocols that only accept binary allow/block exit codes with no interactive prompt channel. On these agents, yolonot automatically translates high/critical ask verdicts into deny (or all ask verdicts when "unresolved": "deny" is set) and never fabricates human approvals on PostToolUse:

Coding Agent	Protocol Behavior on Borderline / High Risk
6 Binary-Hook Agents Goose, Kiro, Cline, Crush, OpenHands, Kimi	✓Auto fail-closed on high/critical risks Returns structured block message to the agent so it can adjust its command without hanging

3. Day-2 multi-agent management commands

Installed a new coding agent today? Wire it in seconds or inspect your active hooks across all 24 harnesses:

$ yolonot setup --harness goose         # Wire a newly installed agent
$ yolonot setup --harness claude --mod  # Enable Claude Mod in-TUI badges
$ yolonot models && yolonot judges      # Bind Ollama, Jev, or cloud LLMs
$ yolonot status                        # Active profile, judges & agents
$ yolonot doctor                        # Check hook health & latency
$ yolonot rules discover                # Import IDE rules into .yolonot