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

ABOUT THE ALMANAC

About

The Almanac is 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. Detailed analytical layers are available under license.

The corpus

Each entry is a fingerprint computed from accumulated observations of an autonomous agent operating across public platforms. Fingerprints are written under codenames following the convention CLUSTER-HASH, where the cluster prefix denotes the agent's dominant purpose class and the hash is derived from a stable identifier.

Codenames preserve operator pseudonymity. The Almanac records behavior, not identity.

The taxonomy

Each agent is positioned along a cluster-specific axis set. The Influence cluster uses Rapport, Escalation, Consensus, Credibility, Exchange, Urgency, and Identity. The Service cluster uses Resolution, Empathy, Clarity, Boundary Respect, Escalation Appropriateness, Accuracy, and Compliance Adherence. Each purpose class has its own seven-axis vocabulary specific to the behavioral patterns that distinguish entries within it.

Axis scores aggregate across observation windows. Sample size and stability accompany each score on every agent entry. Each entry also carries an agent confidence — how strongly the entry's observations match a known model family — and a model family attribution; entries without positive agent evidence are held out of the catalog.

Cluster boundaries

Agents are assigned to the cluster whose axis sample size and aggregate signal dominates their behavioral record. Adjacent clusters share behavioral vocabulary and are linked from each cluster page; some entries sit near a boundary and may reclassify as observation density grows.

Contribute

The Almanac accepts observation submissions from operators, researchers, and platform stewards. Submitters can claim entries corresponding to their own agents and elevate them to verified disclosure tier, or contribute observations on agents operating in their venue.

To inquire about contribution, reach Cipher AI at [email protected].

License

Agent entries and aggregate cluster data are open for reading, citation, and research use. Detailed analytical layers (per-agent susceptibility profiles, predictive diagnostics, comparative reports) are available under commercial license through Agent Optimization.

For licensing or research-partnership inquiries: [email protected].

Operator

The Almanac is operated by Cipher AI LLC (Wyoming).