LLM Chat Task

Task pool and task management tools for the llm-chat ecosystem. Lets an agent track work items, order them with dependencies, pick the next available task, group them into milestones, and record results. Each task carries a short title plus structured plan fields: description, acceptance criteria, priority, type, reference links, and plan arrays for steps, context, constraints, out-of-scope work, verification, and edge cases. Tasks read as small, executable specs.

Features

• store-backed task pool with dependency tracking, cycle detection, and optional file persistence
• add and remove task dependencies (dependency_id / remove_dependency_id on update_task), with readiness derived automatically
• structured tasks: title (required) + description, acceptance criteria, priority, type, links, and plan arrays
• optional milestone grouping label per task; read_task filters listings by exact milestone
• instructive tool descriptions that coach the LLM on how to fill each field
• read_task filters by status, priority, type, and milestone, searches all task fields with a case-insensitive regex, and annotates matches with the fields that matched
• shortened task ids (at least 8 characters) are accepted wherever an id is expected, for reading, updating, and declaring dependencies

Prerequisites

• Node.js >= 18

Installation

npm install @johannes.latzel/llm-chat-task

Documentation

Full documentation at johanneslatzel.github.io/llm-chat-task/

License

MIT. See LICENSE.

Contributing

Issues and PRs welcome at github.com/johanneslatzel/llm-chat-task.