Hlid

Analysis basis

This report is based entirely on the self-reported project description provided by the caller. It contains no external verification, archived history, or third-party corroboration. All claims are treated as stated by the author and not proven facts.

The description states that Hlid:

• Runs on Windows
• Works directly with projects inside WSL
• Connects to an Obsidian vault on the Windows side
• Provides a shared workspace for Claude, Codex, and ACP agents
• Enables persistent conversations, reference of notes and project files, use of skills across providers, opening terminals, answering agent questions, inspecting tool calls, managing permissions at the individual tool level, and tracking usage
• Allows sessions to be available from desktop or mobile
• Integrates with Codex app server for actual session, tool, approval, usage, and response events
• Supports Computer Use workflow where WSL agents can delegate tasks to Windows-native Codex workers

Confidence Low. The product is described as a command center that brings together multiple AI agent harnesses into one workspace, but no concrete features or interfaces are shown or detailed beyond the author's narrative.

The description states:

• Hlid was originally built for Claude because other tools like OpenClaw or Hermes didn't fit the author's workflow
• The goal is to give agent harnesses a shared home while keeping their unique features intact
• It aims to feel more "in the seat" than black-box tools, allowing visibility into what agents are doing, approvals when needed, terminal access, and session continuity across devices
• Hlid has evolved from being Claude-focused to supporting multiple providers (Claude, Codex, ACP)
• The author used Codex and GPT-5.6 during Build Week to extend it

Inference The positioning appears to have shifted from a niche tool for one provider to a multi-agent workspace platform. However, this evolution is inferred from the narrative rather than evidenced.

The description states:

• The author built Hlid for their own use case
• It supports users who work on Windows with WSL projects and Obsidian vaults
• Users want to see what agents are doing, approve things when needed, jump into terminals, and continue sessions from mobile
• The tool is designed for people who want to play with new AI tooling without waiting for official integrations

Confidence Low. No explicit customer segments or personas are defined; the target is inferred from the author's personal workflow.

The description states:

• There is no mention of pricing
• No revenue model, monetization strategy, or business model is described
• The tool appears to be a personal project built by one person

Confidence Very low. No evidence of any commercial structure or pricing.

The description states:

• Built with Bun, React, SQLite, Tailscale, TanStack, TypeScript
• Runs on Windows, works with WSL projects, connects to Obsidian vaults
• Integrates Codex app server for session-level access instead of scraping output
• Supports Computer Use delegation from WSL to Windows
• Uses WebSockets and local storage (SQLite)
• Cross-device sync was a challenge
• Performance improvements included progressive loading, background work, bounding tool history rendering

Confidence Medium. Technical details are provided but not verified.

The description states:

• The author uses it daily
• It existed before Build Week and was extended during the OpenAI 2026 hackathon
• No revenue, customers, or adoption data is mentioned
• The author mentions dogfooding extensively

Confidence Very low. No evidence of traction, users, or market validation beyond personal use.

The description states:

• Hlid was built because existing tools like OpenClaw and Hermes didn't fit the author's workflow
• It aims to be more transparent than black-box tools
• It supports multiple AI agent providers (Claude, Codex, ACP)
• The author mentions wanting to try new tools and add them to Hlid

Confidence Low. No direct competitors or market positioning are named.

The description states:

• The tool is a solo project (1 member team)
• Cross-device sync was a major challenge
• Performance issues arose with long sessions and large context windows
• The author admits to extensive dogfooding, which may skew perception of usability
• No evidence of external validation or user feedback

Inference Risk of limited scalability due to solo development, potential performance bottlenecks, and lack of independent validation.

1. What is the actual usage pattern of Hlid beyond personal use?
2. How many users are there, if any?
3. Is there a plan for monetization or commercialization?
4. What are the technical limitations of scaling this solution?
5. Are there plans to support additional platforms beyond Windows and WSL?
6. How does Hlid handle data privacy and security in multi-provider environments?
7. What is the roadmap for expanding support for non-ACP harnesses?

Note

These questions are based on the lack of evidence around traction, commercialization, and scalability.