PerpetualSoftware Releases Pad v0.7.0 for Agent-Era Project Management

What changed

PerpetualSoftware has released version 0.7.0 of Pad, a project management tool engineered for the emerging era of AI agents. This release emphasizes a local-first architecture, meaning the core functionality and project data reside on the developer's machine, utilizing SQLite for data storage. Pad provides a unified experience through a command-line interface (CLI), a web user interface (UI), and a dedicated AI agent skill. The tool is designed to operate without requiring user accounts and can be installed via Homebrew with brew install PerpetualSoftware/tap/pad.

The pad init command serves as a central entry point for project setup, automatically detecting necessary configurations, authentication, and workspace settings, including the AI skill. This command is safe to re-run, skipping completed steps and providing a status summary. Once initialized, users can onboard an AI agent into their project by invoking /pad onboard within an agent session. This process adapts the workspace's collections, conventions, and playbooks to align with the project's structure.

Pad distinguishes itself from cloud-centric tools like Linear, Jira, and Notion by focusing on developers working on their local machines. It offers native AI agent integration, a first-class CLI experience, and full offline functionality, contrasting with the read-only cache or limited offline capabilities of cloud alternatives. Data ownership is maintained locally via SQLite, with an optional Pad Cloud service for synchronization and team collaboration.

Features for Developers:

• CLI: Enables task creation, item searching, and status checks directly from the terminal. Examples include pad item create task "Fix OAuth redirect" --priority high and pad item list tasks --status in-progress.
• Web UI: A dark-themed interface accessible at localhost:7777 offering board, list, and table views with drag-and-drop functionality. It features keyboard navigation, a rich text editor, wiki-links, real-time updates, and a project dashboard.

Features for AI Agents:

• AI Skill Integration: The /pad skill allows AI coding tools such as Claude Code, Cursor, Windsurf, Codex, OpenCode, GitHub Copilot, Amazon Q, and JetBrains Junie to interact with project items using natural language. Agents can read, create, and update tasks, ideas, and other project items.
• Conventions and Playbooks: These define project-specific rules and multi-step workflows. Conventions, triggered by events (e.g., "run tests before marking a task done"), and playbooks (e.g., "implementing a feature: read spec, create branch, write tests, implement") help standardize processes. Agents automatically load relevant conventions, and all agent actions are logged in an activity feed.
• Agent Naming: Pad supports naming AI agents to distinguish their contributions in activity feeds and audit logs. Names can be set per-workspace in .pad.toml, via the PAD_AGENT environment variable, or by the runtime environment. Sessions can also be explicitly registered with pad session register --agent <name>.

Why it matters for builders

Pad's local-first design and deep AI agent integration provide developers with unprecedented control and efficiency in managing their projects. By bringing project management directly to the developer's environment, it reduces reliance on external services and enables AI agents to become more integrated, proactive collaborators.

Practical impact

Developers can begin using Pad by installing it via Homebrew and initializing a project with pad init. Experiment with the CLI commands for task management and explore the web UI for a visual overview. For those working with AI agents, integrating the /pad skill allows for natural language interaction with project items, streamlining workflows and automating routine project management tasks.

Caveats and source limits

The provided source material details the features and functionality of Pad v0.7.0 but does not include specific benchmark results comparing its performance against other tools. Information regarding pricing for Pad Cloud services, if applicable, is also not detailed. The source focuses on the technical capabilities and developer-centric design of the tool.