I built FactoryLine for my own AI-assisted workflow after losing too much time to tests that passed in CI but failed in real use. The first question is simple: could this test ever say “no”?
With FactoryLine 0.8.20 in JetBrains IDEs:
• Run factory first-proof --root . to catch hollow tests before review.
• Keep an evidence receipt beside the diff instead of trusting an agent’s summary.
• Use Graph Ops to see what is evidenced, blocked, or next.
• Use the local MCP/CLI workflow with human approval still in control.
For teams, FactoryLine adds an evidence supply line around agent changes: the exact file delta, independent checks, and hash-bound receipts. For individual developers, it is a quick sanity check before a review. The local proof workflow is free, and the JetBrains plugin is available here:
I would value feedback from JetBrains plugin developers and teams using AI coding assistants: what would make this proof flow fit your review process better?
A small update with a practical goal: catch the missing evidence before your iOS app enters another review queue.
I built this after the frustration of an app looking finished while the submission still had gaps. A passing build doesn’t tell you whether restore works, the iPad journey is complete, screenshots match the app, or the reviewer can actually sign in.
The 0.8.21 candidate adds AppForge Mission Control inside FactoryLine. Refresh local status to see the available design receipts, mission captures, App Review gates, quality audits, and submission dossiers. Open Graph Ops to inspect the wider storyboard and SaaS evidence. Missing evidence stays visible; it isn’t turned into a green approval.
For a solo developer, that means a clearer next step before submitting. For a team, it means a review handoff tied to the actual candidate, rather than another reassuring agent summary. Catching a preventable gap earlier could spare another round of rework and days of waiting—but we don’t claim guaranteed approval or a measured rejection-reduction percentage.
Junie or Copilot can remain your coding assistant; Qodana or SonarQube can remain your analyzer. FactoryLine binds their returned changes and supplied analysis to independent local proof. It doesn’t run those products for you or imply their endorsement.
The plugin does not upload builds, access credentials, start TestFlight, or submit to Apple. Final authority stays with you.
The verified 0.8.21 archive has now been submitted to JetBrains Marketplace and is awaiting their review. Submission is not approval.