agent-graph

Long-horizon agent skill for Claude Code / Cursor / Codex / Grok Build — multi-task ledger loop, host-portable, clean-context supervisor, verified gates. Markdown library (loop-graph), not a framework.

Agentic coding scheduler, a Rust agent harness that runs background coding agents autonomously

Graph Engineering for Agent Skills: a specification and toolchain for dynamically discovering context and building observable, testable, and recoverable agent workflows.

Open, local-first, provider-agnostic multi-agent orchestrator: compose frontier models as a coordinated team with Thinker/Worker/Verifier roles, cost budgets, and an auditable agent graph. Zero dependencies, node-free core, TypeScript-first.

SRAO Stage 2-3: Agent capability graph construction. Decompose SRM into atomic task DAGs, map to 6 agent categories, output Agent Cards and capability graphs.

Observability and a transparent, auditable agent graph for Massive Intelligence (IM) swarms: OpenTelemetry-compatible span tracing, incremental drift detection, failure attribution, loop detection, thirteen alert channels, and SHA-256-chained, Ed25519-signed trace logs. Zero dependencies, node-free core, TypeScript-first.

『에이전트 엔지니어링 — AI 에이전트 설계의 결정 기준』 참조 구현. base/는 1권 『밑바닥부터 만드는 코딩 에이전트』 최종 코드(ch25)의 사본이고, plus/는 2권이 더하는 그래프 런타임과 평가 하네스다. 서로 의존하지 않는 독립 Gradle 빌드 둘(JDK 21만 필요).

Go 原生通用图执行引擎 — 服务编排、工作流引擎、ETL 管道的轻量级库。零外部依赖，单二进制部署，嵌入即用。

Graph-based identity resolution and runtime authorization for enterprise AI agents — models users, agents, tools, connectors, workloads, scopes, policies, and data boundaries to make explainable allow/deny decisions on agent tool calls (NetworkX + FastAPI).

Policy-governed autonomous security remediation: detects risk, scores impact, plans fixes, auto-executes low-risk actions, and escalates high-risk changes for human approval — with full audit trail. Demo with optional Claude-powered planning.

A distributed task runner with a group/project hierarchy, its own DSL for defining pipelines, cron scheduling, job chaining, and real-time execution tracking.