class LLM::ACPClient

• LLM::ACPClient
• Reference
• Object

Overview

ACP-backed client wrapper for communicating with local AI agents.

Defined in:

llm/acp/client.cr

Constant Summary

CLAUDE_ARGS = ["@zed-industries/claude-agent-acp"]

CODEX_ARGS = ["@zed-industries/codex-acp"]

GEMINI_ARGS = ["--experimental-acp"]

KNOWN_TARGETS = ["codex", "gemini", "claude", "claude-code"] of ::String

ACP targets Noir knows how to launch. Anything outside this set would be executed as an arbitrary local process (--ai-provider "acp:rm -rf /" → command "rm", args ["-rf", "/"]), which is a code-execution hole when the provider string comes from an untrusted config file. Custom targets are refused unless the operator explicitly opts in.

Constructors

• .new(provider : String, model : String, event_sink : Proc(String, Nil) | Nil = nil)

Class Method Summary

• .acp_provider?(provider : String) : Bool
• .allowed_target?(provider : String) : Bool

  True when the provider resolves to a built-in target, or the operator opted into custom commands.

• .custom_command_allowed? : Bool

  Escape hatch for power users running their own ACP agent binary.

• .default_model(provider : String, model : String) : String
• .extract_target(provider : String) : String
• .mute_acp_logs : Nil
• .resolve_command(provider : String) : Tuple(String, Array(String))

  Resolve provider aliases to actual executable command + args.

Instance Method Summary

• #args : Array(String)
• #close : Nil
• #command : String
• #model : String
• #provider : String
• #request(prompt : String, format : String = "json") : String
• #request_messages(messages : Array(Hash(String, String)), format : String = "json") : String

Constructor Detail

Class Method Detail

Instance Method Detail