Atlassian + OpenAI: Bring teamwork context to ChatGPT and Codex
Atlassian and OpenAI connect teamwork context across ChatGPT, Codex, Jira and Confluence through four integrations: the Atlassian MCP Server in ChatGPT, the Atlassian MCP Server in Codex, AI coding tool deeplinks to Codex, and OpenAI model selection in Rovo Agents.
Bringing ChatGPT into Atlassian embeds the model directly into an enterprise-scale context layer. Powered by the Teamwork Graph, every workflow, summary, and action completed by ChatGPT flows back into the system—creating a continuous feedback loop that keeps Rovo and your custom agents sharp and aligned.
The end-to-end experience
Plan: Define goals, requirements, and decisions in Jira and Confluence.
Connect context: Authorize ChatGPT or Codex through the Atlassian MCP Server.
Build: Research across work in ChatGPT, or use Atlassian context while coding in Codex.
Review: Review proposed Jira, Confluence, Bitbucket or Compass actions prepared by ChatGPT or the implementation plan Codex prepared.
Stay aligned: Track agent progress and activity without switching screens.
One connected loop: Atlassian provides the shared work context → ChatGPT or Codex reasons and makes suggestions → people review proposed actions → Atlassian remains the system of record.
Explore the ways to connect OpenAI and Atlassian
Atlassian MCP Server: Connect ChatGPT to Jira, Confluence and more
Best for: teams that want to work with their permitted Atlassian content (including Jira, Confluence, Bitbucket, and Loom) directly inside ChatGPT.
Add the Atlassian MCP Server as a connector in ChatGPT, then authorize access through Atlassian OAuth. After selecting the connector in a conversation, ChatGPT can call supported Atlassian tools using the user’s existing Atlassian permissions.
The same connector works across ChatGPT’s modes: use it in a standard conversation to summarize and search work, in Deep Research to synthesize evidence across many Jira, Confluence, and Bitbucket sources at once, or in Agent mode to run multi-step tasks and prepare supported actions while you stay in control of any changes.
Ground ChatGPT’s reasoning capabilities in Atlassian context to track, update, and move work forward.
Value moment: ChatGPT can work with current, permission-aware Atlassian context without users pasting tickets and pages into the prompt.
Requirements: ChatGPT Team or Enterprise with developer mode, access to the relevant Atlassian Cloud products, and any administrator approval required by organizational policy.
Atlassian MCP Server: Build in Codex with Atlassian context
Best for: developers who want Jira requirements and Confluence decisions available while planning, implementing, and reviewing code in Codex.
Codex supports Atlassian MCP Server and allows developers to authenticate to their Atlassian site. Codex can then use supported Jira, Confluence, and other Atlassian apps from its command-line, desktop, or editor workflow without requiring developers to reconstruct the work context manually.
Alternatively, developers can install the Teamwork Graph CLI (TWG CLI)—a terminal-native command-line interface that sets up agent skills so Codex can query and act across Atlassian products and the wider Teamwork Graph, using your existing permissions.
Not sure whether to use the CLI or MCP for Codex? See Atlassian’s Teamwork Graph CLI and Atlassian MCP decision guide.
Set up Codex with Atlassian MCP
Add Atlassian MCP Server to Codex and complete OAuth authentication.
Open the target repository in Codex.
Ask: “Read PAY-214 and its linked architecture decision, then propose an implementation plan for this repository.”
Review the retrieved context and plan before approving code changes.
Value moment: Codex begins with the team’s agreed requirements and decisions—not only the local codebase.
Requirements: Codex on a paid ChatGPT plan (Plus, Pro, Business, or Enterprise), using the Codex CLI or IDE extension with MCP support; access to the relevant Atlassian Cloud products; repository access; and the product permissions required for each Atlassian tool call.
Model selection: Choose an OpenAI model for a Rovo Agent
Best for: Rovo agent builders who want to match the model family to the reasoning and workflow needs of an agent.
Build Rovo Agents on ChatGPT models. Match the model to what the task demands - deeper reasoning for complex workflows or faster responses for routine ones. The agent stays anchored in your Atlassian context no matter which model you choose.
Choose OpenAI models to build Rovo Agents
Value moment: organizations get model choice without giving up the Atlassian context and governance that make the agent useful at work.
Requirements: an eligible Rovo plan and permission to create or configure Rovo Agents. Available model options can vary by plan, region, and product configuration.
AI coding tool deeplinks: Open Jira work directly in Codex
Best for: developers who want to move from a well-defined Jira work item into their coding agent without rebuilding the prompt.
Start from the Jira work item that already contains the goal, requirements, and acceptance criteria, then open it directly in Codex. The deeplink shortens the handoff from planning to implementation while keeping Jira as the source of intent.
Example:
Open a ready-to-build Jira work item.
Select the option to open the work in Codex
Watch Codex receiving the Jira context and preparing an implementation plan.
Review the plan before allowing code changes.
Requirements: Jira Cloud, Codex installed and authenticated, and access to the relevant repository and Jira work.
How the experiences work together
These experiences share a foundation but solve different jobs. The ChatGPT connector brings Atlassian context into conversations, Deep Research synthesizes evidence, Codex supports software development, and Rovo model selection brings OpenAI into Atlassian-built agents.
If you want to… | Start with… |
Access Atlassian context from Jira, Confluence, and more in a ChatGPT conversation | Atlassian MCP connector in ChatGPT |
Use Atlassian requirements and decisions while coding | Codex with Atlassian MCP Server |
Use an OpenAI model behind an Atlassian agent | Model selection in Rovo Agents |
Synthesize evidence across Atlassian work | ChatGPT Deep Research |
FAQs
Do ChatGPT and Codex respect my Atlassian permissions? Yes — every MCP request runs under your existing Atlassian permissions, and high-impact actions can be reviewed before they run. Additional admin controls are available.
What do I need to connect ChatGPT? ChatGPT Team or Enterprise with developer mode, access to the relevant Atlassian Cloud products, and any admin approval your policy requires.
Should I use the Teamwork Graph CLI or the Atlassian MCP server with Codex? They’re complementary. Use the Atlassian MCP server for IDE, web, or sandboxed hosts and PAT access; use the Teamwork Graph CLI for terminal or CI/CD workflows, scriptable `twg` commands, and OAuth. See our decision guide.
Can I open a Jira work item in Codex? Yes — an AI coding-tool deeplink opens a Jira work item directly in Codex, which picks up the context and drafts an implementation plan you review before any changes. Needs Jira Cloud, Codex authenticated, and repository access.
Can I use Atlassian context inside Codex? Yes — Codex supports Atlassian MCP Server; add the Atlassian MCP Server and authenticate to your site.
Is my Atlassian data sent to OpenAI? When Atlassian content is sent to ChatGPT or Codex, your organization’s OpenAI plan and terms govern OpenAI-side processing. This differs from OpenAI models used inside Atlassian-managed Rovo experiences.
Getting started
Choose the surface. Start in ChatGPT for research and knowledge work, Codex for software development, or Rovo Studio for an Atlassian agent.
Confirm access and terms. Review Atlassian permissions, OpenAI entitlements, data handling, and administrator policy.
Begin read-first. Test search and summarization before enabling write or agent actions.
Keep humans in control. Review proposed updates and code through existing organizational governance.