​

Missions and agents

Missions are repeatable Qodex jobs for work that should happen more than once. In the web app they appear as agents on the Autopilot page. Use an agent when you want Qodex to keep a product area healthy over time, such as writing and replaying API tests for a collection, running API security checks, exploring for bugs, or following up on new bugs.

​

Find your agents

In the sidebar, open Autopilot > Agents. The page lists:

• Switched on: agents that run on their trigger.
• Switched off: agents you added but paused.
• Add an agent: the library of agents you can add to the project.

What agents produce, such as proposals, questions, and updates, lands as cards under Autopilot > Tasks. See Autopilot for the board. Agents only run while Autopilot is on for the project. If it is off, an agent’s page says Autopilot is off for this project until an admin turns it on.

​

Add an agent from the library

Each library card shows what the agent does, what it runs against (a collection, a flow, or an endpoint), and its default trigger. Click Add on a card. You can edit the default instructions and choose which connectors the agent may use before you add it.

​

Configure an agent

Click an agent to open its page. From there you can:

• Switch it on or off with Enabled.
• Click Run now to start it immediately. An agent that works on one bug asks you to pick the bug first.
• Read Run history for every run and open the chat behind it.
• Set the Trigger: A bug is opened, A bug is reproduced on staging, On a schedule, or Only when I press Run now. Bug triggers can be narrowed with Severity includes and a daily run cap. Schedules offer presets or a Custom schedule.
• Choose what it Runs against, such as a collection, flow, or endpoint.
• Edit its Instructions: project-specific rules, constraints, and preferences.
• Pick its Connectors: the observability tools, Environment (or The project default), and Code access it may use during a run.

Click Save to publish trigger, target, and connector changes. Instructions save from their own editor.

​

Run now

Click Run now when you want an agent to run immediately, without waiting for its trigger. This is useful after importing a new collection, merging an API change, changing auth settings, or finishing a product area that needs fresh coverage.

​

The Auto-fix agent

Auto-fix bugs pushes a branch and opens a pull request on your repository, so only a project admin can add it, change it, switch it on, or run it. Every member can read its page. Once it is added and on, project admins also see Fix in code on open findings, which sends that bug to the agent. See Triage workflow.

​

Collection sync

Keeping an API collection in line with your code is set per repository rather than as an agent. Open Code > PR review, go to Settings, choose the repository, and open the Automation tab. Under Collection sync, turn on Sync on merge and pick the collection. When a pull request merges into the default branch, Qodex works out what changed and brings the collection back in line. When sync finds new or changed endpoints, review the catalog rows, open important requests in the API Playground, and promote stable requests into API scenarios.

​

Review before promotion

Agents are designed to do the repetitive work while keeping you in control of what becomes active. When an agent proposes scenarios, catalog changes, or follow-up work, review the card on Autopilot > Tasks before relying on the result in schedules, CI, or merge gates. Keep drafts as drafts until the request shape, auth profile, assertions, and environment are correct.

​

Where agents fit

Agents do not replace scenarios or test runs. They help create and maintain them.

• Use scenarios for reusable test logic.
• Use test runs to execute scenarios and review results.
• Use endpoint catalog to understand API coverage.
• Use agents when Qodex should repeatedly keep those assets up to date.

​

Related