Python

Install an integration

A manifest is a file’s path, a dict, or None for the repository’s config.

import agent_harness_adapter as aha

mytool = {
    "version": 1,
    "name": "mytool",
    "instructions": "Run `mytool check` before you finish.\n",
    "skills": [{"name": "mytool", "description": "Check the project with mytool.",
                "body": "Run `mytool check`; fix what it reports.\n"}],
    "hooks": [{"event": "pre-tool", "tools": "shell", "command": "mytool hook {harness} {event}"}],
    "allow_commands": ["mytool"],
}

result = aha.install(mytool, ["claude", "codex", "gemini"])
for h in result["harnesses"]:
    print(h["harness"], [f'{p["state"]} {p["path"]}' for p in h["parts"]])

Keyword arguments take the options: scope ("project", "user" or "local"), root (the project directory), home, and for install force and without. uninstall and status take the same arguments. Results are the JSON of agent-harness-adapter install --json, as dicts.

Answer a hook

With command = "mytool hook {harness} {event}", run_hook reads the harness’s input as Claude Code’s and writes your answer in the harness’s form.

# mytool hook <harness> <event>: Claude Code's input in, its answer out,
# translated for whichever agent ran the hook.
import sys

import agent_harness_adapter as aha


def decide(hook):
    if "rm -rf" in str((hook.get("tool_input") or {}).get("command", "")):
        return {"decision": "block", "reason": "mytool: no rm -rf here"}
    return None  # allow


sys.exit(aha.run_hook(sys.argv[1], sys.argv[2], decide))

parse_hook and answer_hook do the two halves separately. Errors raise ValueError with the library’s message.