Augustus

Find, build, evaluate, and improve systems with decision models.

Augustus equips agents to find useful placements, build decision-driven systems, create evaluations, and hill-climb them against real outcomes. It combines decision theory, value of information, multi-criteria analysis, signal detection, search/control, and formal-methods boundaries. It applies to software, business, organizations, research, and everyday decisions.

TypeSafe Jev (Choice, Score, Noul) is the default hosted exemplar. The skill also covers classical classifiers, encoders, open decision heads, constrained readouts, rankers, and vision scorers. Choose the family by the task, then test it against the baseline. Sometimes the best result is a formula, a checklist, or no new model.

The working model is:

evidence → bounded judgment → explicit policy → checked action → observed outcome

Augustus is independent of TypeSafe. The official TypeSafe skill and current provider docs own API contracts; Augustus supplies composition, implementation, evaluation, and improvement methods. Named for Augustus De Morgan, mentor of William Stanley Jevons.

Watch the 2:38 film: Jevons' logic piano, De Morgan's boundary, and how Augustus places and tests a bounded judgment.

Try it

After installation, ask your agent:

Use Augustus to audit our refund-email workflow. Find the smallest useful classifier insertion, keep eligibility and payments in code, and propose an evaluation that could reject the change.

Use Augustus to compare ways our library could choose three programs under a fixed budget. Make the values, evidence gaps, and tradeoffs explicit.

Use Augustus to review this confidence threshold. Explain what the score means, when to abstain, and what we should measure on held-out cases.

Use Augustus to build a decision-model router and its evaluation harness. Keep our incumbent runnable, test complete episode outcomes, and set up a bounded improvement loop with untouched confirmation data and rollback.

Advice produces a concise design card and falsifier. Build requests produce working adapters, policy and evals; improvement requests produce a bounded incumbent–challenger loop. The agent reads only relevant references. Source popularity and proxy-score gains do not establish improvement.

Examples

Problem	Placement	Evaluate
Expensive generated-JSON email routing	Bounded intent classifier before existing handlers	Action errors, review coverage, total cost
Search results need ordering	Retrieve candidates, then rank relevance	Recall, nDCG, final task success
Many plausible projects under a budget	Explicit utility/MCDA with exact constraints	Sensitivity, feasibility, stakeholder outcomes
Agent claims it is finished	Judge evidence gaps; verify artifacts and effects	False completion and recovery on real tasks
Need a decision under uncertainty	Compare act, defer, and gather-more-evidence	Expected loss and value of information
Model appears to approve a risky action	Treat judgment as evidence inside host policy	Unauthorized effects, failure paths, drift

A typed response is not proof of truth. Ranking scores, probability, confidence, calibration, and action success have different meanings. See the working skill.

Install

For agents supporting the Skills CLI:

npx skills add 24601/Augustus --skill augustus

This follows the repository's current default branch, which may contain unreleased work. For a reproducible source checkout of the last release:

git clone --branch v0.7.0 --depth 1 https://github.com/24601/Augustus.git

The skill directory is .agents/skills/augustus/. Use your agent's local skill installation mechanism to install that directory. Keep its references and scripts together. Merely cloning a repository does not install it into every agent.

The skill needs no API key to provide design guidance. Calling Jev or another hosted provider is a separate, optional integration with its own credentials and costs. Review installed instructions before granting any agent access.

Claude Code marketplace:

claude plugin marketplace add 24601/Augustus
claude plugin install augustus@augustus

In Claude Code, invoke /augustus:augustus; in Codex, use $augustus. If it is not visible, reload your agent's skills/plugins and check its installed version. See worked examples for the kind of result to expect. Avoid installing the same skill by multiple methods in one agent.

Project and evidence

• Skill and reference index: runtime guidance.
• Research archive: primary sources, historical claims, revisits, and the current decision-model review.
• Contributing: content boundaries, tests, behavioral review.
• Changelog: product changes, separate from research observations.
• Website: examples and ecosystem orientation.
• Feedback: installation problems, mistaken activation, and sanitized real-world failures.

For local development, install requirements-dev.txt and run make check. The offline helpers evaluate labeled binary predictions and paired workflow outcomes; neither calls a model. The composition calculus and build/improvement workflow connect methods to implementation and evidence. Tests and structural lint do not establish model quality or deployment benefit.

Versioning

Published release: 0.7.0. See the release notes for changes and migration details. This branch's skill package is development 0.7.1-dev until a release is cut. That development version only sharpens uncertainty routing; it is not a published package.

Historical TypeSafe skill provenance: v0.5.7 (65a39f3). Read live provider docs before writing integration code; that pin is not a current API guarantee. Install from the v0.7.0 tag when you need the published source revision; default-branch installation may include the unreleased 0.7.1-dev work.

License

MIT. See LICENSE. Security reports: SECURITY.md. Contributions: CONTRIBUTING.md.