# Constal Foundations documentation

> Foundations concepts, workflows, operations, and reference for humans and agents.

- [The Constal model](https://constal.ai/docs/foundations.md): Understand the small Agent language and strict runtime model that make long-running Agent systems durable, governable, and inspectable.
- [The seven Agent primitives](https://constal.ai/docs/foundations/seven-primitives.md): Learn the complete compositional language for Agent judgment, durable history, time, delegation, and volume.
- [From primitives to production](https://constal.ai/docs/foundations/composition.md): See how conversational, tool-using, human-supervised, multi-Agent, long-horizon, and data-intensive Agents compose from seven primitives.
- [Attempts and durable truth](https://constal.ai/docs/foundations/execution.md): Understand interrupted execution, script and durable modes, yields, safe points, replay, and explicit failure semantics.
- [Ledgers, hashes, and append-only history](https://constal.ai/docs/foundations/history.md): Distinguish replay journals, durable facts, logical branches, content-addressed values, controls, and downstream analytics.
- [Resources, effects, and recovery](https://constal.ai/docs/foundations/resources-effects.md): Understand why every model, Tool, API, database, Memory, browser, sandbox, and storage operation crosses one governed boundary.
- [Identity, Policy, and authority](https://constal.ai/docs/foundations/authority.md): Learn how authenticated identity, exact Resource names, pinned Policy, scoped bindings, and safe-point controls constrain Agent execution.
- [Scale with map and reduce](https://constal.ai/docs/foundations/volume.md): Use the two volume primitives for partitioned work, deterministic aggregation, durable stages, and content-addressed datasets.