Chapters
DOC-04 / Technical reference
Technical Documentation
The harness of Synedre, in depth. A public, sanitised reference: how the data layer, the agentic core, chantiers, automata, the hub, email, memory, deployment and façades actually fit together.
11 chapters
Anatomy of Shyrka — see the body firstNightly loop
The system maintains its own documentation
Every night, the harness inspects its own state — doc↔code drift, health check, gated regeneration. This terminal shows the actual log of the last session.
Chapters
Overview of the Synedre OS agentic harness
This page introduces the architecture of the Synedre OS harness by describing its three execution layers (Nuxt mothership-app, Python synedre/, bin/ scripts), their articulation around the single PostgreSQL database vaisseau_mere_ac, the lifecycle of a request, and the governance principles of the system.
Synedre OS data layer
This chapter describes the data architecture of the Synedre OS harness: a single PostgreSQL database `ac_hub`, three access paths (Nuxt, agentic Python, Drizzle ORM), table families organized by prefix, and naming conventions linking CodeMyShop and the agentic cockpit.
The harness agentic core
How the Synedre harness (which orchestrates CodeMyShop) turns an email into an action: classification, isolated launch through a pseudo-terminal, agent personas, AI routing and deployment/QA orchestration.
Worksites, Jobs & Tasks — Data Model and API
Describes the worksite/job/task hierarchy in Synedre OS, the DB data model, canonical statuses/scopes, and the associated Python API.
Automata, crons and runs: non-conversational execution layer
This page describes the deterministic execution architecture of Synedre OS: Python facades, cron wrapping, automata registry and the boundary between runs and agent tasks.
The Hub: the harness administration cockpit
Architecture of the Synedre harness private cockpit: Nuxt layers, navigation, endpoint families, authentication, consoles and secret delivery guards.
Inbox hub and Atlas Inbox: two distinct IMAP pipelines
This chapter describes the two IMAP collection pipelines of Synedre, one feeding the personal mailbox into `ps_ac_inbox_emails` via heuristic classification, the other processing the CodeMyShop agentic mailbox to trigger actions via LLM classification.
Memory and learning of the Synedre OS harness
This chapter describes the three-level memory architecture of Synedre (reflex, Zettelkasten, pgvector vector), the associated indexing automata, and the learning loop based on scars and lessons.
Deployment & infrastructure of Synedre OS
This chapter describes the release pipeline of Synedre OS and its tenants, detailing the asymmetry between `./deploy` (preprod, AI-driven) and `./ship` (production, gated by the `auto_ship_allowed` fleet), as well as associated mechanisms such as drift auto-apply, post-deployment smokes, and founder exceptions without preprod.
Harness facades catalogue
Catalogue of the Synedre harness entry points (scripts, hooks, work queues), organised by family, with each one's invocation mode.
Skills, agents and hooks of the Synedre OS harness
This chapter describes the agentic architecture of `synedre-os`: catalogue of skills, delegable agents and event hooks that orchestrate the Claude Code harness doctrine.