research-prompt-triage-context

hook UserPromptSubmit multi-harness mode: context

Activate research hook state and inject research-method reminders.

Registry Entry

Field	Value
id	research-prompt-triage-context
mode	context
logical_event	UserPromptSubmit
command	python3 {repo_root}/hooks/wagents-hook.py research-prompt-triage-context --harness {harness}
timeout	5
harnesses	["codex", "claude-code", "github-copilot", "gemini-cli"]

Note

Sets research active state (env + file state). Policy _policy_prompt_triage. Also loads skills/research/scripts/research_hook.py for some behaviors.

Full hook config + command reference

config/hook-registry.json (entry)

{
  "id": "research-prompt-triage-context",
  "description": "Activate research hook state and inject research-method reminders.",
  "mode": "context",
  "logical_event": "UserPromptSubmit",
  "command": "python3 {repo_root}/hooks/wagents-hook.py research-prompt-triage-context --harness {harness}",
  "timeout": 5,
  "harnesses": [
    "codex",
    "claude-code",
    "github-copilot",
    "gemini-cli"
  ]
}

hooks/wagents-hook.py (policy)

def _policy_prompt_triage(payload: NormalizedPayload) -> int:
    if not _is_research_prompt(payload.prompt):
        return 0
    # activate state + inject reminders
    _write_state(payload)
    return _additional_context(payload, "Research mode active. Use evidence-ledger, obey readonly guards, verify before stop.", "research-prompt-triage-context")

Dispatcher also falls back to research_hook.py for some surfaces.

Resources

Hooks Hub All repo-managed lifecycle hooks.

wagents hooks Inspect and validate hooks via CLI.

research skill The research skill that pairs with these guards.

---

View source on GitHub