Agentic OS workflow
A workspace with AGENTS.md and 49 skills for product managers: backlog, goals, priorities, PRDs and verified execution
- 119 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- The setup.sh script and integrations work with your files and external service keys
Install
- Works in
Manual install
git clone https://github.com/itseffi/personal-os.gitThen enter the folder and run ./setup.sh. For Claude Code add a bridge to .agents/skills.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The setup.sh script and integrations work with your files and external service keys
This is third-party code. Review the repository files before installing.
What it does
Agentic OS, the personal-os repository, is a ready workspace for an agent: Claude Code, Codex, Pi, OpenClaw and similar runtimes. It holds an `AGENTS.md` of rules, goal and backlog files, folders for tasks, knowledge and resources, workflows and 49 skills in `.agents/skills`. The agent turns a raw backlog into prioritized tasks tied to goals and requires commands and verification evidence for each. The skills cover a product manager's work: assumption mapping, opportunity-solution trees, JTBD, hypothesis testing, PRD and spec writing, competitor analysis, root-cause analysis, a decision journal, a power map and preparing difficult conversations. Integrations check against Jira, Linear, Slack, Google Calendar and Granola meeting notes. First run: clone the repository, run `setup.sh` and ask the agent to process the backlog from BACKLOG.md into Tasks under the AGENTS.md rules.
Who it is for. Product managers and founders who keep tasks, goals and decisions in files and want an agent to triage, prioritize and draft documents.
Good fit when
- A backlog has piled up in `BACKLOG.md` and must be sorted into tasks tied to goals
- Before a sprint you need the day's top three tasks with commands that verify the result
- An idea becomes a PRD and first needs an assumption map and a hypothesis test plan
- You are preparing a difficult conversation or an executive update and need a message structure
Not a fit when
- Team work with shared trackers and no file structure: the set is built for a personal repository
- You need one separate skill such as `tdd`: you would copy it out of the catalog by hand
- Notes are not kept in files, or data cannot be handed to a model
Example request
Process my backlog from BACKLOG.md into Tasks under the AGENTS.md rules and show unblocked P0 and P1 tasks aligned to GOALS.md.Limitations
This is a workspace template: you clone it and adapt it, and pulling upstream updates is a manual job. The Jira, Linear, Slack, Google Calendar and Granola integrations need your own keys and MCP setup. Claude Code, Pi and OpenClaw need symlink bridges to the skills directory. Subagent mode works only if the runtime can delegate tasks. The last commit is from 7 October 2026.
How to disable. Delete the cloned repository folder. If you created the .claude/skills and skills symlinks, remove them too, and revoke integration access in Jira, Linear, Slack and Google.
FAQ
How do I start Agentic OS?
Run `git clone https://github.com/itseffi/personal-os.git`, then `chmod +x setup.sh` and `./setup.sh` in the folder. Open the repository in your agent and give it the backlog task.
How do I connect it to Claude Code?
Create the bridge with `mkdir -p .claude` and `ln -sfn ../.agents/skills .claude/skills` from the repository root. For OpenClaw add a `skills` symlink to `.agents/skills`.
Why does each skill have an openai.yaml?
It holds routing metadata for Codex and OpenAI-style runtimes. Claude, Pi and OpenClaw read `SKILL.md` and do not need that file.
Can it run on a local model?
Yes. The README has a Pi section for a local or offline backend such as `llama.cpp`, or a hosted endpoint. Setup goes through `~/.pi/agent/models.json`.
Which integrations are included?
Checks against Jira, Linear and Slack, a Google Calendar schedule check and Granola meeting sync. Each lives in `System/integrations` with its own skill and setup note.
Discussion
No questions or reviews yet.
Related
WorkflowsPersonal assistant kit for Claude Code, Codex CLI and Grok CLI: file-based memory, 23 skills and a guided first-run setup
A scaffold for Claude Code and Codex: tasks in files, developer and QA roles, manual approval and shared state.
cafe3310's personal skill collection: spec-driven project work, reviews, reports, browser, 3D and video in one repository
A plugin for Claude Code and Codex: department folders with rules and 35 skills to run a one-person company through AI agents
A CLAUDE.md and AGENTS.md file with seven principles: the agent frames the problem and scope before it writes code
Set up Claude Code and Codex in one conversation: global instructions, rules, plugins, skills, a status line and correction memory
README in short
The README opens with a three-step quick start and a diagram: user prompt, agent runtime, instructions, skills, workflows and state. It then covers compatibility with Claude Code, Codex, Pi and OpenClaw, skill bridges, Pi setup with local models and the workspace file layout.