Module agents 

Multi-agent support — agent identification, tool name normalization, and protocol implementations.

This module provides:

• AgentKind — enum identifying each supported agent
• Canonical tool alias table — maps agent-native tool names to internal names
• Permission mode resolution — normalizes agent-specific mode strings
• Per-agent HookProtocol implementations

Modules

amazonq

Amazon Q Developer CLI hook protocol implementation.

claude

Claude Code hook protocol implementation.

codex

OpenAI Codex CLI hook protocol implementation.

copilot

GitHub Copilot CLI hook protocol implementation.

gemini

Gemini CLI hook protocol implementation.

opencode

OpenCode hook protocol implementation.

Enums

AgentKind

Identifies which coding agent is calling.

Functions

canonical_to_internal

Given a canonical name (e.g. “shell”), return the internal name (e.g. “Bash”).

display_name

Return the best user-facing display name for a tool.

internal_to_agent

Given an internal name and target agent, return the agent’s native tool name.

internal_to_canonical

Given an internal name (e.g. “Bash”), return the canonical name (e.g. “shell”).

resolve_any_to_internal

Resolve any tool name to its internal form.

resolve_permission_mode

Given an agent’s native permission mode string, return the canonical mode.

resolve_tool_name

Given an agent’s native tool name, return the internal (Claude-style) name.