hcompai

Read a folder of receipt images/PDFs, append rows to a LibreOffice Calc bookkeeping ledger (description, category, type, amount), save, and start a Mail draft with the sheet attached.

Use when asked to QA Nimbus Desk, run a qa/*.md behavioral spec, verify UI behavior, check a regression, or confirm a UI fix. Drives the running app through the root Holo MCP tool, holo_desktop(task), and returns a QA report.

Sub-agent that binds to one OS window on the user's machine and drives it in the background via H Company's Holo3 VLM. If direct Holo tools are not visible, search for holo_desktop tools. In NemoClaw, call holo_desktop_launch once, then poll the same run_id until it finishes, passing attached images/files as media_paths.

Creating, editing, and managing events, invites, and recurring schedules in the Apple Calendar desktop app.

Finding, creating, editing, and organizing contacts and groups in the Apple Contacts desktop app, including linked cards from multiple accounts.

Composing, sending, replying, searching, and triaging messages in the Apple Mail desktop app.

Sending iMessages, SMS, and replies, plus reactions and search, in the Apple Messages desktop app.

Creating, reading, searching, editing, and organizing notes into folders and tags in the Apple Notes desktop app.

Expert knowledge of the two H Company backends: the portal (auth, organizations, API keys, invitations, billing at portal.api.eu.hcompany.ai / portal.production.hcompany.ai) and the agent platform v2 API (sessions, agents, skills, environments, vaults, long-polling, the hai-agents SDK at agp.hcompany.ai). Use this skill whenever the user mentions portal, agent_platform, platform.hcompany.ai, agp.hcompany.ai, HAI_API_KEY or hk-... keys, hai-agents / hai_agents, run_session, agent sessions or trajectories, agent-view or replaying a run, /api/v2 endpoints, browser agent environments, vaults for agent secrets, organizations or invitations on the H platform, the desktop OAuth/PKCE login flow, or wants to call, integrate with, debug, or automate anything against an H Company API, even without naming it (e.g. "get me an API key", "set up my .env", "launch a web agent", "why is my session stuck", "show me what the agent did", "why am I getting 401 from the platform?").

Run an autonomous QA review of a web UI by invoking the local `qa-cli` command via the shell. Trigger when the user asks to QA a URL, check a deployed site for accessibility, usability, or correctness issues, get a structured review of a page, or run a quick visual sanity check on a live page (e.g. "what color is the CTA?").

Expert knowledge of the two H Company backends — the portal (portal, API at portal.api.eu.hcompany.ai / portal.production.hcompany.ai — auth, organizations, API keys, invitations, billing) and the agent platform v2 API (agp.hcompany.ai — sessions, agents, skills, environments, vaults, long-polling, the hai-agents SDK). Use this skill whenever the user mentions portal, agent_platform, platform.hcompany.ai, agp.hcompany.ai, HAI_API_KEY or hk-... keys, hai-agents / hai_agents, run_session, agent sessions or trajectories, agent-view or replaying/reviewing a run, /api/v2 endpoints, browser agent environments, vaults for agent secrets, organizations or invitations on the H platform, the desktop OAuth/PKCE login flow, or wants to call, integrate with, debug, or automate anything against an H Company API — even without naming it (e.g. "get me an API key", "set up my .env", "launch a web agent", "why is my session stuck", "show me what the agent did", "why am I getting 401 from the platform?").

Run an autonomous QA review of a web UI by invoking the local `qa-cli` command via Bash. Trigger when the user asks to QA a URL, check a deployed site for accessibility, usability, or correctness issues, get a structured review of a page, or run a quick visual sanity check on a live page (e.g. "what color is the CTA?").

Delegate a task to an autonomous H Company agent (Holo-powered web surfer and computer-use agents) through the hai-agents MCP server. Use to offload open-ended browser or research work that the agent should carry out end to end on its own infrastructure, then report a final answer. Runs remotely, so it is the right tool when the work is "go do this on the web and come back with the result" rather than something you can answer directly or do with local files.