Pixellint
Validates impression pixels, click trackers, and conversion postbacks against vendor-documented rulepacks.
A pixel with a misspelled event name or a raw unhashed email returns 200 and looks healthy. Attribution quietly degrades for weeks before anyone notices.
What it does about it
Check the URL against what the vendor actually documents, including PII handling, before it ships to production.
Findings are structured, not prose. Every one carries a stable rule id you can suppress, track, or cite in a ticket, a severity, the exact location in the payload, and guidance on the fix. The same verdict comes back whether you called the CLI, the library, the browser build, or the MCP server.
Coverage
- Rulepacks. Core, plus Meta, Google Analytics, Floodlight, LinkedIn, and TikTok
- Declarative. Rulepacks are JSON, so a team can add its own vendor without touching Rust
- Evidence tiers. Every finding carries a severity, an evidence level, and a link to the document it came from
- PII checks. Flags raw email and identifier values where the vendor requires SHA-256 hashing
Calling it from an agent
Pixellint exposes 2 tools over the Model Context Protocol, served over stdio from the local binary. See the MCP overview for wiring it into Claude Code, Cursor, or your own agent loop.
Use it without installing anything
Pixellint’s own site covers what it checks and why. The browser tooling lands as the project leaves early access.
Where it fits
Pixellint grades one crossing in the delivery chain. The other two cover the crossings on either side of it: RTBlint and VASTlint. Same output contract, same license, so a team can adopt one and add the others without relearning anything.