Agentic pull-request review, shipped both as the Tessl Review GitHub App and as tessl code review in CI. It reviews the current PR, a named PR, or an explicit git range, and can publish its verdict back to the pull request or approve on behalf of a bot login.
This is a registry capability that genuinely performs an SDLC stage, and it is the clearest instance of the sideways growth the supply layer page warns about: a company that sells capability distribution also sells the review those capabilities are meant to improve.
Its distinctive mechanism is lenses — named, versioned review perspectives that replace the default set, configurable per repository through a .tessl-code-review.yml profile and pinnable by version. That is the persona-panel idea (pattern-parallel-persona-review) rebuilt as installable, versioned units rather than as prompts baked into a framework’s skill. gstack and ce-code-review hard-code their reviewer roster; pstack-interrogate deliberately replaces personas with model diversity; Tessl makes the roster a dependency you install and pin. Same pattern, distributed through the layer it belongs to.
The premise is stated as a scaling problem rather than a quality one: “Code review doesn’t scale to agent-authored PRs.” That is the honest framing for why a supply-layer product ends up here — the volume of agent-written code is what creates the market for both halves.
See Also
- tessl-change-risk · tessl-change-verify — the two other lifecycle capabilities, aimed at a diff rather than a PR.
- gstack-review · addy-code-review · mp-code-review · ce-code-review · pstack-interrogate · sp-requesting-code-review — the framework-side review cluster.
- stage-review — the canonical stage this implements.