Skip to main content

Automated Jobs (CI & Muti)

fibe.gg goes beyond simple environments by introducing Automated AI Repair Jobs. Using your Stored Agents and the Job Mode infrastructure, you can configure Playspecs to automatically spawn environments that detect, analyze, and fix code issues.

There are two primary automated job modes available: Playspec CI-Job and Playspec Muti-Job.

1. Playspec CI-Job (Continuous Integration Repair)​

When a CI pipeline or a test suite fails, it usually falls on a developer to investigate the logs and push a fix. A Playspec CI-Job automates this by spinning up a Job Playground, running the tests, and bringing in an AI Agent to immediately propose and implement fixes.

How it works:​

  • Trigger: The job is triggered (e.g., via a webhook or API when a branch is pushed).
  • Execution: A Job Playground is created from the Playspec.
  • Agent Injection: The configured AI agent executes the test suite.
  • Analysis & Repair: If failures occur, the agent receives the test output, analyzes the root cause, and uses its MCP tools to modify the source code inline.
  • Verification: The agent re-runs the tests to verify the fix and, depending on configuration, can push the changes back to the repository.

2. Playspec Muti-Job (Mutation Testing Repair)​

Mutation testing introduces deliberate bugs (mutations) into your code to verify that your test suite catches them. A "surviving mutation" means your tests are incomplete. Playspec Muti-Job automates the process of writing new tests to kill surviving mutations.

How it works:​

  • Trigger: Triggered periodically or on demand.
  • Execution: The environment generates a list of mutations using tools like Mutmut or specific Muti frameworks.
  • Agent Injection: The AI Agent reviews the surviving mutations diffs.
  • Analysis & Repair: The agent writes new, comprehensive test cases specifically designed to catch the altered behavior (the "mutant").
  • Verification: The test suite is re-run against the mutant. If the mutant is killed, the new test is pushed.

Configuration Overview​

Automated Jobs are configured directly within a Playspec. You specify:

  • Whether the feature is enabled (trigger_config["enabled"]).
  • Which Stored Agent ID should handle the job.
  • The repository Playzone ID and the target Branch.
  • A custom Prompt Template to guide the agent (e.g., "Fix the CI errors:\n{{logs}}", or "Fix this surviving mutation:\n{{diff}}").
  • The maximum number of retry attempts.

Once configured, these jobs drastically reduce the manual overhead of fixing broken builds and improving test coverage, leveraging the true power of your autonomous AI agents.