Tree Monstor is an AI software development profile running within the Hermes Agent framework. It is not an ordinary AI chatbot, but a virtual development team with comprehensive software engineering capabilities.
This repository also configures its own agents. See what tree_monstor tells them →
Skill Claude CodeCodex
A project-planning skill that uses backlog.md to analyse work, estimate Story Points (rough effort estimates), split work into Sprints (short development cycles), and maintain design documents.
Skill Claude CodeCodex
A planning guide for turning work into a backlog, a list of planned tasks. It includes principles, questions to consider, and the INVEST rules for writing backlog items.
Skill Claude CodeCodex
A skill that turns conversations or user actions in Tree Monstor into a new skill file in the project’s skills folder.
Skill Claude CodeCodex
A workflow for letting an AI agent carry out a large task toward a deadline with minimal interruption. It covers planning, design, execution, reflection, submission, and a log of questions the agent answers along the way.
Skill Claude CodeCodex
A document-extraction tool that turns text, PDFs, office files, web pages, images, and video subtitles into structured Markdown notes. Markdown is plain text with simple formatting marks, and OCR reads text from images.
Skill Claude CodeCodex
A guided question tool that reads pending concerns from need-you-help.md and records each decision in that file and the matching backlog item.
Skill Claude CodeCodex
A tool for reducing the file size of PowerPoint presentations without changing their content. It recompresses videos and images, repackages the presentation, and checks that the resulting file remains intact.
Skill Claude CodeCodex
A development workflow skill that uses TDD, or test-driven development: write a test first, make the code pass, then check it again. It also covers probes, regression checks, and two review stages.
Skill Claude CodeCodex
An optional LLM-based review and repair layer for the dav-developer workflow. It judges whether acceptance criteria failures are real bugs, proposes fixes, applies patches, and rechecks them in a sandbox.
Skill Claude CodeCodex
A manually triggered review and delivery procedure using an independent subagent. It checks functionality, code quality, test coverage, regression risks, and whether the final result matches each part of the user's request, then prepares a delivery summary and updates affected documents.
Skill Claude CodeCodex
A task-routing and standard-procedure skill for non-coding work such as research, media generation, document editing, and one-off questions. It can escalate a task when code is needed and reports skipped or falsely passed procedure checks.
Skill Claude CodeCodex
A context-management procedure for shortening or compacting an agent's working information when needed. The description mentions a four-step process, triggers, rules, exceptions, and limits.
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: