What Videlic does
Videlic reads the full log of an AI coding session — every file the agent read, every edit it made, every command it ran — and turns it into a report you can act on at a glance. Instead of trusting the agent’s summary, you see what actually changed, how risky those changes are, whether the tests are real, and exactly how the session unfolded.
It runs automatically on the work your agent pushes, so you get a report without changing how you code.
Quickstart
Three steps to your first report.
- 1
Install the GitHub App
Connect Videlic to GitHub and choose the repositories you want it to watch. Videlic only ever sees the repos you select.
- 2
Add the agent hook
Open Setup in your dashboard and run the one-line command once. It installs a Claude Code Stop hook that sends each finished session — the raw JSONL transcript — to Videlic and detects the agent automatically. No SDK, no code changes.
$ curl -fsSL https://videlic.dev/start | sh -s <your-token> - 3
Open a pull request
Code as usual. When you open a PR, the verdict posts as a comment and a status check in about a minute. Open the full report from the comment or your dashboard.
What’s in a report
Every session runs three independent engines, plus a full timeline. Each looks at the work from a different angle, and any one of them can surface a problem the others miss. Every plan gets all of them — Free included.
Pulse
Compares what the agent said it did with what it actually did. It catches summaries that overstate the work, steps that were quietly skipped, and claims the changes don't support.
Graph
Maps how your files depend on one another and scores every edited file by how much downstream code relies on it. High-impact changes stand out before they ship.
Probe
Looks at the tests. It flags assertions that can never fail, tests that were deleted or skipped, and edge cases the new code leaves uncovered.
Session Timeline
A complete, chronological record of every file read, edit, and command — so you can see exactly where the agent changed course or got stuck. It's a separate view, not a fourth engine.
Example
An agent reports that it hardened the auth path — but Pulse finds the change actually removed the check, Graph flags 14 files that depend on it, and Probe finds no test covering it. Verdict: Review.
Reading your report
Each report opens with a clear verdict, a one-line headline, and a short reason — so you know where things stand before you read another word.
- Clear — nothing needs your attention. The work holds up. Merge with confidence.
- Review — worth a look before you merge. The report points to exactly what, and where.
- Blocked — a serious problem. Don’t merge until it’s resolved.
Every finding is backed by concrete evidence — the agent’s own words, the file and line it points to, and the change it relates to — so you can click straight to the source instead of taking the result on faith. Nothing is a black box.
Supported agents
Claude Code is supported today — auto-detected from the session format, with nothing to configure. Cursor, Cline, and Codex are on the way.
Integrations
Videlic posts the verdict where your team already reviews code — GitHub today, with GitLab, Slack, and email on the way.
Plans & Teams
You’re billed per review — one per pull request, no matter how many agent sessions go into it. Re-pushes and re-reviews on the same PR are free, so you’re never charged for iterating. Every plan runs all three engines plus the timeline; you pay for scale, not features.
Free is free forever — no credit card. Pro raises your monthly reviews and removes the watermark from shared reports. Team adds a shared review pool, a team dashboard, an audit log, and member roles — Owner, Admin, and Member. See the pricing page for current limits.
FAQ
Do I need to change my code or add an SDK?
No. You install the GitHub App and run a one-line hook once. After that, every session is analyzed automatically — nothing in your codebase changes.
What does Videlic read?
Only the session log of the agent run — files it touched, commands it ran, and its own messages — for the repositories you connect. Nothing else.
Does it slow down my workflow?
No. Analysis runs after the agent finishes and posts the verdict to your PR in about a minute. You keep working as usual.
What happens when I run out of reviews?
Analysis pauses until your next cycle, or you can add a pack or upgrade. Re-reviews on PRs you've already paid for stay free.
Privacy & Data
Your session data is used only to generate your reports. You choose which repositories Videlic can see, and you can disconnect at any time. For the full details on what we collect and how it is handled, read our Privacy Policy.
Ready to see what your agent actually did?
Connect a repo and get your first verdict in minutes. Free forever — no card.
Get started free