Plugin marketplace
Skills for working with clauz3 — using installed trusted layers, authoring new ones, and stress-testing agents through clauz3.
Add this marketplace
Then install any plugin below by name — each page has its exact command.
For agents that need to do real-world things — send mail, edit files, fetch URLs, move money, run a lab step — in a project where clauz3 is the only outlet to side effects. Covers writing the small Python program against trusted modules, attaching guarantees the user will review, and passing through clauz3 prove / clauz3 run.
3 skills
For designing the trusted boundary an agent acts through — effects, contracts, fluents — when no trusted layer exists yet for a domain. Root-of-trust role: if this layer is wrong, every proof against it is wrong. Covers effects.py with deal/clauz3 markers, contracts via the relation algebra, optional fluents for stateful properties, and packaging as a clauz3-install-able library.
3 skills
For acting as the user-proxy while driving an independent Claude Code (or other agent) through clauz3 tasks adaptively — picking prompts, watching what the agent submits, deciding approvals via the approval service's REST API, escalating to adversarial cases, and surfacing design issues. Open-ended counterpart to the deterministic e2e driver.
3 skills
Shows how many plugins this repo ships, and links here.