FAQ

Pricing

How much does TriRev cost?

Billing goes through GitHub Marketplace. We do not see your card details and we do not handle invoicing - GitHub does.

The Free plan is permanent and does not require a credit card. The cap resets at 00:00 UTC on the 1st of each calendar month and is counted per installation, by unique (repository, pull request) pair.

What counts as a review?

One pull request equals one review for the entire calendar month, regardless of how many commits you push to it. Force-pushes, rebases, and follow-up commits to the same PR do not multiply the count. Closing and reopening the same PR does count as a new review.

Will the Free plan ever change?

We will give 30 days of notice on this docs page and via a comment on every active install's welcome issue before changing the Free plan limits.

Data handling

What data does TriRev send to Anthropic?

For every PR review, TriRev sends to the Anthropic Claude API:

• The PR diff (added and removed lines).
• File context: full file content for changed files under 500 lines.
• PR metadata: title, description, branch names, author username.

That is the full list. TriRev does NOT clone your repository, does NOT read files outside the PR scope, and does NOT send any data from repositories where it is not installed.

Is my code stored?

No. Code is processed in memory during the review and discarded after the review comment is posted on GitHub. We do not retain diffs, file contents, or any code data beyond the duration of the review.

We do retain operational metadata: GitHub organization and repository identifiers, install identifiers, review counts and timestamps, and plan information. None of this includes code.

Is my code used to train AI models?

No. As of 3 May 2026, under the Anthropic Commercial Terms of Service, submitted data is not used to train Anthropic's foundation models. Anthropic may retain inputs and outputs for up to 30 days for trust and safety review under those Commercial Terms; we have not enabled the Anthropic Zero-Retention option at MVP. We monitor changes to those terms and will update this page if their commitments change.

Where does TriRev run?

TriRev's backend runs on Railway (a cloud infrastructure provider). Connections to GitHub and Anthropic use TLS. Webhook payloads from GitHub are verified using HMAC-SHA256 signatures.

For full details including GDPR rights and data retention, see our Privacy Policy.

Model and architecture

What AI model does TriRev use?

TriRev runs three specialized review agents on every PR:

• Correctness agent - logic errors, null safety, off-by-one, race conditions.
• Security agent - injection, auth bypass, secret leaks, common CWE patterns.
• Style agent - naming, structure, dead code, language idioms.

Each agent calls the Anthropic Claude API independently. Results are merged into a single review comment posted directly on your PR.

We may route different agents to different Claude model tiers (Sonnet vs Haiku) to balance review depth and cost. The user-visible behaviour does not change with this routing - you always get a unified review.

Is TriRev open source?

The TriRev application code is currently in a private repository while the product stabilizes. The public-facing documentation you are reading lives at trirev.dev. We will revisit the licensing question once the product is past M7.

Languages and frameworks

Which languages does TriRev support?

TriRev's agents are LLM-based, so they can produce useful feedback on most mainstream languages. We have specifically tested and tuned for:

• TypeScript and JavaScript (primary - this is what we use ourselves).
• Python.
• Go.

For other languages (Rust, Java, C#, Ruby, Kotlin, Swift, etc.) the agents still run, but finding accuracy is highest for the four primary targets above. We rely on user feedback to prioritize tuning for additional languages - tell us at the Contact link if your stack is not covered.

Does TriRev work with monorepos?

Yes. TriRev reviews each PR independently, regardless of repository structure. Use ignore_paths in .trirev.yml to exclude generated code, vendor directories, or untouched parts of the monorepo:

ignore_paths:
  - "**/dist/**"
  - "**/vendor/**"
  - "**/*.generated.*"

Custom rules

Can I write my own review rules?

Not as code. TriRev's behaviour is configured via .trirev.yml, not via custom plugins or scripts. The configuration covers:

• Which agents run (agents.correctness, agents.security, agents.style).
• Severity threshold (severity_threshold: low / medium / high / critical).
• Confidence threshold (confidence_threshold: 0.0 to 1.0).
• Path inclusion and exclusion (ignore_paths).
• Project-specific context for the agents (custom_rules - free-form text the agents read on every PR).

The custom_rules field is the closest thing to "custom rules". You write a short text block describing your project's conventions ("we always use named exports", "controllers must call assertNoNulls on inputs", "this is a Next.js app router project") and the agents take it into account when reviewing.

The full schema lives in the Configuration reference.

Can I disable a single category of finding?

Yes. Set the agent producing it to false:

agents:
  style: false       # disables ALL style findings
  correctness: true  # keeps correctness on
  security: true     # keeps security on

For finer control, raise severity_threshold to filter out lower-severity findings without disabling the agent entirely.

Can I have TriRev skip a specific PR?

Yes. Add a label named trirev:skip to the PR. TriRev will not review PRs carrying that label.

Team and enterprise

Do you have a Team plan?

Yes - $25 per developer per month, 5-seat minimum. Team adds shared org-level configuration, admin controls, and priority support on top of Pro. Buy via the Marketplace listing.

Do you offer self-hosted or on-premise?

Not at v1. Self-hosted is on the roadmap (M10) but is not committed. If on-premise is a hard requirement for you, tell us at the Contact link - it helps us prioritize.

Do you offer enterprise terms (DPA, SOC 2, custom contract)?

Not at v1. Once we have repeatable enterprise demand, we will publish a separate enterprise plan. For DPA/SOC 2 questions in the interim, contact us at the link below and we will share what we currently have.

Offboarding and uninstall

How do I uninstall TriRev?

1. Go to your GitHub Settings → Applications → Installed GitHub Apps.
2. Click Configure next to TriRev.
3. Scroll to the bottom and click Uninstall.
4. Confirm.

The uninstall is immediate. TriRev stops processing webhooks for your account or organization within seconds.

What happens to my data when I uninstall?

• Code data: nothing to delete - TriRev never stored any.
• Operational metadata (review counts, install identifier, billing history): retained while the install is active and for up to 90 days after uninstall, then purged. Earlier deletion on request via support@trirev.dev; we comply with GDPR right-to-erasure within 30 days of receipt.
• Existing review comments on your PRs: they stay on GitHub, owned by your repository. TriRev does not touch them on uninstall.

To request deletion of operational metadata, contact us at the link below with your GitHub organization name.

What happens if I cancel a paid plan?

You keep Pro (or Team) features until the end of your current billing period. At the next monthly rollover, your install reverts to the Free plan (30 PRs/month) automatically. There is no proration - GitHub Marketplace handles refunds and credit per their policy.

You can resubscribe at any time via the same Marketplace listing.

Contact