Bloblings

Product Introduction

1. Definition: Bloblings is a native macOS desktop application that functions as a real-time visual agent management dashboard for Anthropic's Claude Code and other AI coding assistants. It is a system-level utility that installs a local file-watching hook to monitor AI agent activity without requiring cloud connectivity.
2. Core Value Proposition: It exists to solve the cognitive overload and context-switching friction developers face when managing multiple concurrent Claude Code, Codex, or terminal-based AI agent sessions. Its core value is providing at-a-glance session awareness, intelligent thread prioritization, and seamless inter-agent communication directly from the macOS menu bar, ensuring no thread is forgotten or lost.

Main Features

1. Visual Agent Dashboard (The Drawer): A persistent, hideable panel that provides a live visual status for every active AI coding session. Each session is represented by a unique, animated "Blobling" character. The dashboard uses a local file-watching system (likely fsevents or similar) to detect tool usage (edits, commands, reads) from Claude Code, translating each action into a specific, non-intrusive animation (e.g., typing, hammering, reading). It renders natively using macOS frameworks for low system impact.
2. Cross-Thread Communication & Team Orchestration: Facilitates structured communication between isolated AI agent threads via user-mediated "notes." Notes are plaintext files passed locally between session hooks, allowing threads to share context (up to 5,000 characters) without exposing full conversation histories. This enables multi-agent workflow setups where specialized threads (e.g., "frontend," "backend," "Codex debugger") can collaborate. Features include automated note passing (up to 40/hour), manual delivery control, and one-click team templating.
3. Subagent Activity Monitoring: Provides unique visibility into Claude Code's subagent (helper) system. When a main session delegates tasks, the primary Blobling visually indicates delegation, and a detailed, fold-out "Subtasks" panel lists each helper agent with its own micro-pet, current task, model type, and effort level. This allows for real-time oversight of parallelized AI work without switching contexts.
4. Integrated Time Tracking & Productivity Desk: A built-in work timer that correlates billable time directly with active Claude Code threads. It automatically attributes work segments to specific client/project codes and can split a timer's duration across the multiple AI threads that were active during the period. It generates invoice-ready summaries and logs with time-stamped notes, leveraging its unique position as a session monitor.

Problems Solved

1. Pain Point: Lost Context & Thread Proliferation. Developers using Claude Code for complex projects often spawn dozens of threads across different folders and windows. It becomes impossible to track which session is active, which is blocked awaiting input, and which has completed a task, leading to wasted time searching and duplicated effort.
2. Target Audience: Professional Developers & Technical Teams who extensively use AI coding assistants (Claude Code, Codex) in their daily workflow. This includes full-stack engineers, DevOps specialists, and tech leads managing complex, multi-repository projects. It also serves freelancers and agencies who need to track time per-client across multiple AI-assisted tasks.
3. Use Cases: Monitoring a Multi-Agent Development Sprint where one thread writes API logic, another builds the UI, and a Codex agent troubleshoots errors, with the developer overseeing all. Debugging a Complex Issue with a dedicated "investigation" thread that runs tests while the main thread continues feature work. Client Work with Strict Time Budgeting, using the integrated desk to track time spent by AI agents on specific project modules.

Unique Advantages

1. Strengths & Limitations (Pros & Cons):

   • Pros:
     • Unparalleled Session Awareness: Solves a fundamental UX gap in AI developer tools by making concurrent session states visually intuitive.
     • Native & Private: A fully native macOS app. All data (notes, status, hooks) remains on-device, ensuring privacy and compliance for sensitive codebases.
     • Deep Claude Code Integration: Offers features like subagent monitoring and effort-mode visuals that are not available in any other third-party tool.
     • Reduces Context Switching: The ⌥⌘P ("Next up") hotkey and one-click thread jumping drastically reduce mental overhead.
   • Cons:
     • Platform Lock-in: Exclusively for macOS, excluding Windows and Linux developers.
     • Agent Dependency: Primary features are optimized for Anthropic's Claude Code. Support for other agents (Codex, Hermes) requires manual setup and may have limited feature parity.
     • Learning Curve: The density of features (themes, note systems, team templates, the desk) requires initial configuration to tailor to a user's workflow.

2. Key Alternatives & Differentiation:

   • Alternative 1: Manual Window/Tab Management. The standard method of using native app windows or terminal tabs. Bloblings differentiates by aggregating all sessions into a single, prioritized visual interface with status intelligence, which manual management completely lacks.
   • Alternative 2: IDE/Editor Plugins for AI. Plugins like Claude for Cursor or VS Code integrate the AI chat within the editor. Bloblings is complementary, not competitive; it manages sessions across all these tools (Claude app, Codex, terminal), providing a unified control layer that editor-specific plugins cannot.
   • Alternative 3: Generic Time Trackers (e.g., Toggl Track). While these track time, they cannot automatically attribute work to the specific AI agent threads that performed it. Bloblings' desk feature is uniquely capable of providing a "thread-by-thread" breakdown of AI-assisted work sessions.

Frequently Asked Questions (FAQ)

1. Is Bloblings secure for proprietary code? Yes. Bloblings is a local-first macOS application. It operates by reading very small, anonymized activity logs generated by Claude Code's local hook (e.g., "tool used: edit, file: utils.ts"). No code, conversation data, or prompts ever leave your Mac.
2. How does Bloblings work with Codex or other AI agents? For agents like Codex that can execute local commands, you paste a setup prompt. The agent then safely modifies its own configuration to add Bloblings' local reporting hook. This establishes a connection similar to Claude Code, enabling visual pets and status for those chats.
3. Does Bloblings slow down Claude Code or my Mac? No. The monitoring hook is designed to add minimal overhead (claimed ~10ms per activity). The app itself is a lightweight native utility. It does not intercept or process any of the AI model's computations.
4. Can I use Bloblings to manage teams of AI agents? Yes. Bloblings includes team templates to quickly spin up coordinated groups of AI threads (e.g., "Researcher," "Architect," "Coder"). You can facilitate "Huddles" for cross-thread reporting and use the note-passing system to orchestrate workflows between specialized agents.
5. What happens if I have dozens of active sessions? The panel supports a condensed view, shrinking each card. The sorting algorithm prioritizes sessions that need attention (blocked on user input) at the top. You can also configure sorting to be static or use the "Next up" (⌥⌘P) hotkey to cycle through waiting sessions efficiently.