An agent step: tools, a schema gate, a keyed seal, a trace
@ahmadssadiq· 13 Oct 2026
- Model
- Claude Sonnet 5.5
- Time to make
- ≈ 21 min
- Tokens spent
- 435k tokensmedium
- Agent
- Claude Code
- Kind
- SVG
- Tags
- agents, observability, llm, identity
- Licence
- MIT
- Added
- 13 Oct 2026
- Source
- ahmadssadiq/agent-line
Prompt
explain how Connectra runs reliable AI agents: multi-step LLM orchestration with tool calling and structured JSON outputs, an observability layer tracing every agent decision end to end, and agent identity with credentialing for attribution, interactive, as React components for my Next.js portfolio, dark theme
Source
The build.mjs that drew it, and its figure.json, are on GitHub. Something wrong with it? Report this figure.
