A reliability layer for AI-assisted software development.

AI coding agents are good at generating code, but long-running projects also need persistent context, structured process, independent review, and coordination across tools and agents.

Siltpoke provides that missing layer.

It currently includes:

• Siltpoke Review — independently reviews code changes and surfaces evidence-backed warnings without blocking the developer.
• Project Life Cycle — preserves project intent, decisions, progress, handoffs, and workflow across long-running coding sessions.

The core idea is simple:

Project state belongs to the project, not the agent.

AI suggests. Humans decide.

Siltpoke is ultimately trying to answer:

How can humans reliably build and maintain real software with coding agents over time?