autoresearch-goal

Coordinates professor-critic research missions with Codex goal-mode state reconciliation.

Updated Feb 1, 2024
One-click install
npx skills add https://github.com/ogiboy/portfolio-app --skill autoresearch-goal-ogiboy
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: autoresearch-goal
Source: https://github.com/ogiboy/portfolio-app/tree/main/.codex/skills/autoresearch-goal
Command: npx skills add https://github.com/ogiboy/portfolio-app --skill autoresearch-goal-ogiboy

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Long-running research missions need durable state tracking and objective completion criteria, but Codex goal-mode state and external orchestration tools can drift apart. This Skill binds a research mission to Codex goal-mode focus while keeping OMX as the durable state owner, with professor-critic validation gating completion. ## Core Features & Use Cases - Mission and Rubric Creation: Creates a mission with a professor-critic rubric via omx autoresearch-goal create, storing mission.json, rubric.md, ledger.jsonl, and completion.json under .omx/goals/autoresearch/<slug>/. - Goal Lifecycle Reconciliation: Emits a model-facing handoff, records critic verdicts (pass/fail/blocked) with evidence, and reconciles the final Codex get_goal snapshot against durable OMX state. - Strict Completion Gate: Blocks completion until a passing professor-critic verdict and a matching complete Codex goal snapshot exist; assistant prose or partial tests are insufficient. - Use Case: A researcher runs a multi-session literature investigation, records each critic verdict in the ledger, and only marks the Codex goal complete after the rubric audit passes, then clears the thread goal before starting another. ## Quick Start Ask the assistant to create an autoresearch goal mission with a topic and critic rubric, then iterate research while recording verdicts until the mission passes and the goal is reconciled.

Frequently Asked Questions about autoresearch-goal

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run a professor-critic research workflow with Codex goal mode?

Create a mission with `omx autoresearch-goal create --topic ... --rubric ... --critic-command ...`, emit the handoff, then create or reuse a Codex goal in the active thread. Record each critic verdict with the verdict command and complete only after a pass verdict and matching goal snapshot.

What artifacts does the autoresearch-goal workflow write?

It writes mission.json, rubric.md, ledger.jsonl, and completion.json under `.omx/goals/autoresearch/<slug>/`. These files provide durable OMX-owned state for the mission, its validation rubric, the verdict history, and the final reconciliation record.

Can shell commands or hooks update Codex goal state directly?

No. Shell commands and hooks must not mutate hidden Codex goal state; only `get_goal`, `create_goal`, and `update_goal` called through the active Codex thread change it. The OMX completion command performs read-only reconciliation plus durable OMX state updates.

When should I not use the autoresearch-goal workflow?

Do not use it for ordinary pre-planning or best-practice lookup, and never revive the deprecated `omx autoresearch` direct launch surface. It is intended only for research missions needing Codex goal-mode management plus professor-critic validation.

Why is my research mission completion blocked?

Completion is blocked until the professor-critic validation records `verdict=pass` and a matching complete Codex `get_goal` snapshot exists. Assistant prose, partial tests, or a failed/blocked verdict are not sufficient to pass the completion gate.