Continue runs source-controlled AI checks as agents on every pull request, enforceable in CI as GitHub status checks.
Continue runs agents on every pull request as GitHub status checks. Each agent is a markdown file in your repo at `.continue/checks/`, returning green if the code looks good or red with a suggested diff if not. AI checks are powered by the open-source Continue CLI (`cn`). There is also a VS Code extension available.
Writing source-controlled AI checks for a codebase that run as agents on every pull request as GitHub status checks, enforceable in CI.
Continue runs markdown-defined agents as GitHub status checks on pull requests. The described behavior is advisory: each check returns green if code looks good, or red with a suggested diff if not. The agent reviews code and surfaces recommendations/suggested diffs as CI status checks; it does not itself merge, commit, or apply changes — a human decides whether to act on the suggested diff. No autonomy signals were detected in the entrypoint code (no subprocess, shell=True, exec/eval on LLM output, or autonomous-loop language in the README). The example check (security review) emits findings rather than executing changes, placing this at Tier 2.
Side effects
• Posts GitHub status checks (pass/fail) on pull requests
• Surfaces review comments and suggested diffs on PRs