Introducing Tessl Verifiers: Enforce AI Coding Standards in Every Change
Summary
Tessl Verifiers introduce a mechanism to enforce complex, high-level coding standards and invariants in a codebase that traditional linters and unit tests cannot reliably cover. Verifiers allow developers to write rules in plain language, which are then automatically checked on every code change. This capability is designed to improve code quality consistency, especially when dealing with code generated by AI agents, by making previously subjective rules into deterministic, enforceable checks that can be run locally, pre-commit, or as a CI gate.
Key takeaways
-
Verifiers vs. Traditional Checks
Verifiers sit between open-ended code review and deterministic linting. They enforce rules that require a degree of judgment (e.g., 'every user-facing error must tell the user what to do next') but can be codified into a clear, repeatable rule set.
-
Addressing AI/Agent Code Drift
As AI agents write more code, Verifiers provide a way to codify specific conventions and invariants that human reviewers or agents might forget or fail to consistently apply, ensuring standards are met every time.
-
Enforcement Scope
Verifiers are suitable for narrow, critical rules (e.g., 'all user-facing text must be in US English') and can be run affordably on every single commit in a CI pipeline, allowing for tens of thousands of judgments.
Technical details
-
Verifier Functionality
0s
Verifiers are rules about codebase invariance that are checked using a model as a judge. They allow developers to write rules in plain language to enforce standards like: 1) Every user-facing error must guide the user on the next steps. 2) Every API route must use typed response helpers. 3) Handlers must not log sensitive data (request headers, tokens, passwords).
-
Implementation and Usage
0s
To get started, users can interact with the Tessle agent via the CLI, asking open-ended questions or utilizing built-in loops like 'review my PRs.' Verifiers can be integrated into the development workflow as local checks, pre-commit hooks, or mandatory CI gates.
Mentioned resources
Channel & topics
Watch on YouTube · Back to latest
This independent, AI-assisted summary is provided for commentary and informational purposes. It may contain errors or omit important context. Please watch the original video for the creator's complete presentation. Video, thumbnail, and related copyrights belong to their respective owners.