Deterministic automation for AI agents.

swamp is a CLI that gives AI agents deterministic, repeatable building blocks for automation — so the work they produce is reviewable, shareable, and accurate.

swamp gives AI agents structured building blocks for automation — typed models, declarative workflows, secure vaults, and versioned data — so they can operate on real infrastructure with confidence, not guesswork.

Everything lives in a .swamp/ directory inside your Git repo. Human-readable, version-controlled, auditable.

• Models — Typed representations of external systems (cloud resources, APIs, CLI tools). Define metadata, arguments, methods, and inputs with schemas.
• Workflows — Declarative YAML DAGs that orchestrate model methods across parallel jobs and steps, with dependency ordering and triggers.
• Vaults — Secure storage for secrets and credentials, referenced via CEL expressions.
• Data — Versioned, immutable artifacts produced by method runs. Searchable by tags.
• Extensions — Community-built models, vaults, drivers, datastores, and reports. Publish and share via the swamp registry.

curl -fsSL https://swamp-club.com/install.sh | sh

swamp works with your favourite AI coding agent — Claude Code, Cursor, OpenCode, and Codex. Point your agent at a repo with a .swamp/ directory and start automating.

Full documentation is available at swamp-club.com/manual.

Only employees of Elder Swamp Club, Inc. contribute code to swamp. We do not accept external pull requests.

We welcome your bug reports and feature requests — file them using the swamp CLI (swamp issue bug, swamp issue feature) which submits directly to our swamp club issue tracker. If we agree to fix or implement them, we will do so with engineers and AI agents under our direct control. We're happy to include you as a co-author on the resulting pull request at your request.

swamp is licensed under the GNU Affero General Public License v3.0 with the Swamp Extension and Definition Exception.

Copyright 2026 Elder Swamp Club, Inc.