Task flow / execution graph plugin for DeepSeek Harness (DSH) — visualize a single task's full run: agents, tools, skills, subtasks, retries, live status. DSH 单任务执行流程图谱插件。
Governed spec-driven development for AI coding agents. Keeps requirements, memory, planning, execution, and verification under control with executable guardrails and human approval, making agentic software development more traceable, verifiable, and predictable.
OpenTelemetry for AI agents — every token traced, outcomes scored, rival strategies auto-compete
A net that werks: plan a project as a graph of verifiable tasks, let a Claude Code loop build it, and watch it live — why it stopped, start and stop it. Ships a harness kit and a skill.
Agent-native Software Development Lifecycle Management for DeepSeek Harness
Spec-driven development harness for AI coding agents with human-in-the-loop approval gates. Turns one-sentence ideas into approved specs, dependency-aware task graphs, and verified run logs for Claude Code, Codex, and Gemini CLI.
Self-measuring agent orchestration core: a task graph that cannot be lied to (done-only-via-verified-push), an operator decision queue, append-only worklog telemetry. Zero dependencies. Bun or Node >= 18.
To associate your repository with the task-graph topic, visit your repo's landing page and select "manage topics."