ata

Create and scaffold plugin directories for Codex with a required `.codex-plugin/plugin.json`, optional plugin folders/files, valid manifest defaults, and personal-marketplace entries by default. Use when Codex needs to create a new personal plugin, add optional plugin structure, generate or update marketplace entries for plugin ordering and availability metadata, or update an existing local plugin during development with the CLI-driven cachebuster and reinstall flow.

Use when the user asks how to build with OpenAI products or APIs and needs up-to-date official documentation with citations, help choosing the latest model for a use case, or model upgrade and prompt-upgrade guidance; prioritize OpenAI docs MCP tools, use bundled references only as helper context, and restrict any fallback browsing to official OpenAI domains.

Babysit a GitHub pull request after creation by continuously polling review comments, CI checks/workflow runs, and mergeability state until the PR is merged/closed or user help is required. Diagnose failures, retry likely flaky failures up to 3 times, auto-fix/push branch-related issues when appropriate, and keep watching open PRs so fresh review feedback is surfaced promptly. Use when the user asks Codex to monitor a PR, watch CI, handle review comments, or keep an eye on failures and feedback on an open PR.

Run a final code review on a pull request

Update the title and body of one or more pull requests.

How to run tests using remote executor.

INTERNAL SUBAGENT SKILL — never invoke directly. Spawned by $paper-discovery to search, deduplicate, cluster, and rank papers for a research topic.

Discover and rank papers for a research topic. Use when a user asks a research question, wants to find papers, or asks how something is done in the literature.

Adapt code to work with a GPU node's installed environment. Use when an experiment fails due to dependency version mismatches, import errors, missing packages, or API incompatibilities between the code's expected environment and what's actually installed on the GPU node. Triggered by errors like 'ModuleNotFoundError', 'ImportError', version conflicts, or when the user says 'adapt this code to the environment'.

Generate a focused report from the current conversation and persist insights. Use when a user has been exploring papers, asking questions, and wants to capture what was discussed. Triggers -- "make a report from this", "save what we discussed", "write up what we covered", "turn this into a document".

Generate integrated cross-paper explanations from KB cards. Use when a user asks to explain, compare, synthesize, understand, or deep-dive into one or more knowledge base cards, or asks how cards relate or work together.

INTERNAL SUBAGENT SKILL — never invoke directly. This is called automatically by $hn-synthesis to discover relevant HN threads. If a user asks about HN discussions, use $hn-synthesis instead.

Search Hacker News for practitioner discussions on a topic, tool, library, or paper, and synthesize findings into structured KB cards. Use when a user asks about community sentiment, real-world experience, practitioner opinions, or HN discussions about a technology, project, or research paper.

INTERNAL SUBAGENT SKILL — never invoke directly. This is called automatically by $hn-synthesis when it spawns a thread-analysis subagent. If a user asks about HN discussions, use $hn-synthesis instead.

Knowledge Base operations for card-based research knowledge management. Use when reading, writing, searching, updating, resetting, or managing KB cards and files. Also handles "reset KB", "clear KB", and "wipe KB" — immediately resets to empty state without asking questions. Other research skills reference this skill via $kb for KB operations.

REQUIRED for any paper explanation or synthesis. Do NOT call attach_url_files and summarize inline — always use this skill. Use when given an arXiv URL, DOI, paper title, or Zotero reference.

Subagent skill for paper synthesis. Spawned by $paper-synthesis to fetch and extract information from a single paper.

Zotero library management through the `ata zotero ...` CLI namespace. Use when the user mentions Zotero, their library, a collection, group libraries, or "my papers", and you need to search, inspect, organize, import, or update Zotero records without exposing the large native Zotero tool family.

Multi-repo workspace management: clone repos, create execution runs, track papers/datasets/artifacts, manage snapshots, and review audit logs. Use when the user wants to organize multi-repo work, run experiments in sandboxes, or track research resources.

Generate a concise research briefing from KB cards or a topic. Use when a user wants a quick overview, orientation, summary of approaches, or wants to understand what their options are before diving deep. Use instead of cross-paper-report when the user wants brevity over depth. Examples -- "give me a quick overview of VLA approaches", "what are my options for sim-to-real", "summarize what I have on diffusion policies", "brief me on action tokenization methods".

Install Codex skills into $CODEX_HOME/skills from a curated list or a GitHub repo path. Use when a user asks to list installable skills, install a curated skill, or install a skill from another repo (including private repos).