ReviewSensei

Spec sheet

What it does

ReviewSensei is an open-source engine for AI reviews of GitHub pull requests using repository-owned knowledge, configurable review stages, and explicit coverage.Read 11 Oct 2026 · reviewsensei.dev

Purpose

ReviewSensei is open-source alpha software for AI code review of pull requests using repository-owned knowledge and configurable review stages.Read 11 Oct 2026 · reviewsensei.dev

Integrations

The site offers a GitHub App, GitHub Action, and CLI, and its provider matrix lists Ollama, Ollama Cloud, OpenAI-compatible, and OpenRouter support with different workflow availability.Read 11 Oct 2026 · reviewsensei.dev

Alpha status

The project identifies itself as alpha software and says capabilities, defaults, and distribution channels evolve.Read 11 Oct 2026 · reviewsensei.dev

Controls

Separate switches govern analysis, comment publication, automatic approval or request changes, learning, conversations, and artifact retention.Read 11 Oct 2026 · reviewsensei.dev

Coverage

Opt-in large-change orchestration reports per-path outcomes such as reviewed, partially reviewed, excluded, unsupported, or budget exhausted.Read 11 Oct 2026 · reviewsensei.dev

Data handling

The core does not persist raw prompts or provider responses; published comments, artifacts, approved learnings, and provider-side copies have separate retention locations.Read 11 Oct 2026 · reviewsensei.dev

Data persistence

The core engine does not persist raw prompts or raw provider responses; review results, GitHub comments, artifacts, and approved learnings have separate retention locations.Read 11 Oct 2026 · reviewsensei.dev

Diagnostics

The doctor and plan commands preview configuration, stages, budgets, and skip reasons without provider calls or writes.Read 11 Oct 2026 · reviewsensei.dev

Evidence limitation

Schema and diff-location validation do not prove that a finding is correct or that review coverage is complete, and legacy evidence policy is the default.Read 11 Oct 2026 · reviewsensei.dev

Execution

Review logic runs in the user's GitHub Actions environment, self-hosted runner, or local CLI environment.Read 11 Oct 2026 · reviewsensei.dev

Finding lifecycle

Incremental reviews use stable fingerprints to label findings new, still present, fixed, outdated, or uncertain.Read 11 Oct 2026 · reviewsensei.dev

Finding tracking

Incremental reviews use stable fingerprints to track findings through states including new, still present, fixed, outdated, and uncertain.Read 11 Oct 2026 · reviewsensei.dev

Installation

The site describes three installation routes: App-assisted GitHub reviews, manual GitHub Actions setup, and local CLI.Read 11 Oct 2026 · reviewsensei.dev

Learning

Proposed repository learnings can become draft pull requests and require maintainer approval before future selection for matching paths.Read 11 Oct 2026 · reviewsensei.dev

Model providers

The getting-started page lists local Ollama, Ollama Cloud, OpenRouter, and an OpenAI-compatible CLI option.Read 11 Oct 2026 · reviewsensei.dev

Provider costs

Users provision and pay for model inference, GitHub Actions minutes, and self-hosted runner capacity.Read 11 Oct 2026 · reviewsensei.dev

Provider data limits

ReviewSensei says it does not guarantee provider-side retention, training, residency, or deletion practices.Read 11 Oct 2026 · reviewsensei.dev

Publication controls

Comment publication defaults to off, and separate settings control review policy, learning, conversations, and artifact retention.Read 11 Oct 2026 · reviewsensei.dev

Release status

The site labels ReviewSensei open source and alpha.Read 11 Oct 2026 · reviewsensei.dev

Repository learning

Proposed learnings can become draft learning pull requests and are selected for matching paths after maintainer approval.Read 11 Oct 2026 · reviewsensei.dev

Review coverage

Opt-in large-change orchestration reports per-path outcomes including partially reviewed, excluded by policy, unsupported, and budget exhausted.Read 11 Oct 2026 · reviewsensei.dev

Review stages

Its specialist reviewers cover correctness, architecture, security, and performance before synthesizing results.Read 11 Oct 2026 · reviewsensei.dev

Security limitation

ReviewSensei says it does not guarantee provider retention, training, residency, or deletion practices.Read 11 Oct 2026 · reviewsensei.dev

Where Specifiction places it

• Code review software#42 of 507.4