o8-code
Skill24 starsChanged 42 days agoMITView on GitHub
Inspect or change application code, API behavior, routing, model calls, persistence, or architecture contracts.
In this repository11 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- o8-checksSkill
- o8-contributeSkill
- o8-delegateSkill
- o8-releaseSkill
- o8-uiSkill
- shipSkill
- o8-devSkill
- reportsSkill
- shipSkill
The file
Files
Code and architecture
- o8 is a Next.js 16 and React 19 app in a Tauri v2 shell. Desktop and mobile are separate surfaces.
Orchestrator backends in
src/lib/lane/orchestrator-backends/and worker adapters insrc/lib/runtimes/are separate systems. Inspect the current registries before changing routing. - Runtime code is the source of truth for ports, routes, commands, and registrations. Never hardcode API or WebSocket ports, user home paths, or stale inventory counts. Use the existing resolvers and registries.
- API middleware is default-deny for
/api/*. Public routes require deliberate narrow allowlisting; externally reachable self-authenticating routes must verify their own credentials. Return structured route errors rather than throwing to the framework. - TypeScript is strict. Use the existing two-space, single-quote, semicolon style and
@/imports. The default file ceiling is 800 lines; existing waivers do not authorize new ones. - Ad-hoc model calls use the existing proxy and routing layer. The sanctioned AI SDK import boundary
is
src/lib/chat/gateway-client.ts. Strict-mode MCP schemas keep a plain top-level object; validate conditional relationships in the handler. - Architecture and extension contracts:
docs/internals/system-architecture.md,docs/internals/runtime-adapter-contract.md, and the implementation named above. Historical inventories in docs are orientation, not runtime truth.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

