Developer preview. Pre-admission. Not for sensitive, proprietary, regulated, or production code. Expect breaking changes and data loss.
Read the limits.
# Bridlewise
> Bridlewise is an experimental, local-first coding cockpit. A run is a causal
> record: model turns, governed tool calls, file effects, checks, findings, and
> typed refusals. The product is a developer preview, pre-admission
> (`harnessReady: false`), not production-ready, and not for sensitive,
> proprietary, regulated, or production code.
This file is a curated map for language models and agents. It is not a
substitute for the repository or the security policy. The source and the
tests are the authority.
## The one-page orientation
- [Landing page](index.md): the concept,
the three doors, and the warning.
- [Run record](https://llm-4-people.github.io/bridlewise-testing/run/): an
illustrative, synthetic causal record. It is not a live model run.
- [Policy lab](https://llm-4-people.github.io/bridlewise-testing/policy-lab/): an
illustrative governance console. Toggling a switch does not admit a real
capability.
- [Architecture](architecture.md):
cockpit, loopback server, engine, guardian, timeline, and memory.
- [Security and limits](security.md):
the safety floor, the confinement model, and what is not defended.
- [Glossary](glossary.md): the
vocabulary the product uses precisely.
- [Evidence](evidence.md): what can
be verified and what cannot.
- [Known issues](known-issues.md):
current gaps, without marketing.
## Repository facts
- Product: https://github.com/LLM-4-People/bridlewise-testing
- Development floor: https://github.com/LLM-4-People/bridlewise-testing
- Release: https://github.com/LLM-4-People/bridlewise-testing/releases/tag/v0.0.3-preview
- License: AGPL-3.0-only.
- Runtime: Node.js >=24.18.0 <25.0.0. Linux-first; Landlock confinement is
Linux-only. The cockpit server binds loopback by default.
- The public smoke suite is reproducible from a fresh clone; the full
corpus-backed development-floor suite is not, because the 4.1 GB corpus is
not published.
## What not to infer
- A green CI badge is not a security audit.
- A recorded run is not a proof that every effect was benign.
- A successful check is not a guarantee that the check was sufficient.
- An illustrative UI is not a connected engine.
- A public preview is not a production support commitment.
## Contact
- Security: private advisory through
https://github.com/LLM-4-People/bridlewise-testing/security/advisories/new
- Questions: https://github.com/LLM-4-People/bridlewise-testing/discussions