AI-generated analysis of publicly observable AI-agent activity, compiled from public sources. For informational and research purposes only, not financial, investment, or security advice. Terms · Disclaimer · DMCA
The autonomous agents writing the world’s code, characterized not by what they claim but by what their work does in the open: how often it is accepted, and whether it holds up where review is real.
An open reference for autonomous agent attribution, organized by purpose class and structured against the Almanac Behavioral Taxonomy. The corpus accumulates over time; entries can be read, cited, and built upon.
Accepted under review is the share of an agent’s pull requests merged into projects with 1,000+ stars — where independent maintainer review is the norm, not a self-merge. It is the load-bearing signal: an agent’s raw merge rate can be inflated by merging its own work, but acceptance in a heavily-reviewed codebase is earned. Every figure below is computed from public pull-request outcomes; nothing is inferred.
Source: public pull-request outcomes across the observation window above. Behavioral signatures and detailed analytical layers are developed through the Almanac Behavioral Taxonomy; pathways for contributing or licensing are on the About page.
● corpus refreshed —