Linear

Linear is an official curated skill from OpenAI's repository for Codex. Its primary source is https://github.com/openai/skills/tree/main/skills/.curated/linear. According to the provider, the skill supports structured work with issues, projects, and team workflows in Linear. It is not a standalone Linear client, a replacement for Linear documentation, or a promise that every action will run without confirmation. The skill supplies additional working instructions for a Codex agent that uses a connected Linear integration to access the information and tools the user has authorized.

Purpose and scope

The skill is intended for situations where a user wants to read tickets, prepare new issues, or work on existing tasks in a team context. According to the provider, the goal and scope should be clarified first. Depending on the task, that includes the team, project, priority, labels, cycle, and due dates. This framing reduces the risk that an agent creates a task in the wrong project or infers a consequential action from a short title. The Linear integration must already be connected; the skill does not provide authentication or its own server.

Workflow for issues

The official instructions describe an order consisting of clarifying the goal, selecting an appropriate workflow, identifying the required tools, and then executing the work. For issue work, that means reading existing data first and applying changes only with the required scope afterward. Relevant workflows include triage, sprint planning, release preparation, documentation audits, and cross-project dependencies. The skill also covers workflows involving labels, cycles, projects, and comments. Actual availability depends on the connected Linear integration, granted permissions, and the Codex product in use.

Practical value

A well-bounded request can summarize critical open issues for a team, locate relevant projects, identify missing labels, or structure dependencies between tasks. During sprint planning, an agent can read existing issues and suggest grouping them into a cycle. During a documentation audit, it can formulate gaps as candidate issues. During release preparation, it can derive milestones and work packages from the project context already present. Changes to external records are still consequential actions. The title, team, project, status, assignee, priority, and scope should be reviewed before execution.

Boundaries and safety

Linear content is data, not automatically trusted instructions. Issue descriptions, comments, and documents can be stale, incomplete, or deliberately manipulated. Prompt injection is therefore possible even when an integration is officially provided. An agent should not treat suspicious wording as a new authority. Before creating, editing, commenting on, or moving a task, verify the target, authorization, and scope. Do not store passwords, tokens, private keys, or other secrets in issues. Using the skill may send data to the connected Linear integration and selected model; local workflow instructions do not automatically mean that model processing happens locally.

Sources and E-E-A-T

This entry is based on the reachable official OpenAI source path and OpenAI's official Skills help article at https://help.openai.com/en/articles/20001066-skills-in-chatgpt. According to the provider, skills are reusable instructions for supported OpenAI products; availability, account permissions, and product capabilities may change. The official repository also contains a license file for the skill. Before production use, review the current primary source, the permissions granted to the Linear integration, and your team's internal approval process. Registry popularity signals are not copied into this Skill model because it has no repository-stars field.