Computer-Use Agents API MCP server
io.github.hcompai/hai-agentshcompaistreamable-httpv1.0.0Scanned
Run and manage H Company's Computer-Use Agents from any MCP client.
9 tools: 1 can delete or cancel things, 1 do something that can't be undone, 3 change things and 4 only read.
- Reads private data · 2
- Reads text others write · 1
- Can send data out · 1
- No findings
Tools listed without sign-in. Remote posture
With Anlyon
1 tool here can delete or cancel things and 1 does something that can't be undone. Govern them.
- Anlyon lists the tools and adopts none.
- Each tool you adopt gets approvals and receipts.
- Your agent never holds the key.
A tool your agent calls directly is not routed through Anlyon and is not governed by it. Anlyon Radar only lists and labels, and these labels are the same either way.
What the Computer-Use Agents API MCP server can do
The Computer-Use Agents API MCP server is published by the GitHub account hcompai. It has 9 tools. It can delete or cancel things (1 tool: cancel_session), do things that cannot be undone, such as sending, paying or publishing (send_message) and create or change data (3 tools: run_agent, share_session and write_file). 4 tools only read.
2 tools read private data (list_files and read_file); 1 tool brings in text other people can write (list_files); 1 tool can send data out of your system (send_message). On its own it has all three parts of a prompt-injection exfiltration path: private data, text others write, and a way out.
None of the fixed checks matched anything in its tool definitions. That is not a safety guarantee.
Its remote endpoint lists its tools without sign-in, and its tools changed once since Anlyon Radar first saw them.
Last scanned Nov 19, 2026, version 1.0.0.
Tools
Grouped by what each tool does, most consequential first. Open one for its reasons and full description; descriptions are the publisher's own words.
9 tools
Deletes or cancels 1Claude Code: deny
cancel_sessionCancel a session; a no-op if it does not exist or already finished.No findings
Why it is in this group
- The name "cancel_session" says "cancel": a word for removing or ending something.
Description, as the server wrote it Cancel a session; a no-op if it does not exist or already finished.
Can't be undone 1Claude Code: ask
send_messageSend a follow-up message to a running session.Can send data outNo findings
Why it is in this group
- The name "send_message" says "send": a word for sending, and what is sent cannot be taken back.
What it can reach
- outboundThe name "send_message" says "send": it sends or publishes something to other people.
Description, as the server wrote it Send a follow-up message to a running session.
Changes things 3Claude Code: ask
run_agentStart an agent on a task; return the answer or a session handle to wait on.No findings
Why it is in this group
- The name "run_agent" says "run": a word for a change.
Description, as the server wrote it Start an agent on a task; return the answer or a session handle to wait on. ``agent`` is a name from list_agents, which runs its latest revision, or ``name:N`` to run revision N.
write_fileWrite base64-encoded content to a file on a session's machine.No findings
Why it is in this group
- The name "write_file" says "write": a word for a change.
Description, as the server wrote it Write base64-encoded content to a file on a session's machine. ``session_id`` is the id ``run_agent`` returns, or an env-manager session id.
Only reads 4Show 4 read-only tools
list_agentsList agents the caller can run: their org's agents plus the public ``h/`` ones.No findings
Why it is in this group
- The name "list_agents" says "list": a word for reading.
Description, as the server wrote it List agents the caller can run: their org's agents plus the public ``h/`` ones.
list_filesList files on a session's machine, newest first; by default its environment's own, such as a browser's downloads.Reads private dataReads text others writeNo findings
Why it is in this group
- The name "list_files" says "list": a word for reading.
What it can reach
- private-dataThe name "list_files" says "list" and "file": it reads files.
- private-dataThe description's first sentence says "list" and "file": it reads files.
- untrusted-inputThe description's first sentence says "list" and "browser": it reads web pages, which anyone can write.
Description, as the server wrote it List files on a session's machine, newest first; by default its environment's own, such as a browser's downloads. ``session_id`` is the id ``run_agent`` returns, or an env-manager session id. When a listing is truncated, page on the last entry's modified_at (ISO-8601) and name, passed as modified_before and name_after. Send both: modification times are not unique, and a timestamp alone skips the rest of a tied group.
read_fileRead a file from a session's machine; returns base64 content and metadata.Reads private dataNo findings
Why it is in this group
- The name "read_file" says "read": a word for reading.
What it can reach
- private-dataThe name "read_file" says "read" and "file": it reads files.
- private-dataThe description's first sentence says "read" and "file": it reads files.
Description, as the server wrote it Read a file from a session's machine; returns base64 content and metadata. ``session_id`` is the id ``run_agent`` returns, or an env-manager session id.
wait_for_sessionPoll a session: with ``wait`` long-poll for the answer, else return the current snapshot.No findings
Why it is in this group
- The name "wait_for_session" says "wait": a word for reading.
Description, as the server wrote it Poll a session: with ``wait`` long-poll for the answer, else return the current snapshot.
This is not a safety rating, and no finding is not a safety guarantee. The groups and findings come from fixed checks of the tool definitions the server returned on the date shown, not from its code or what it does when called. How the checks work.
Exfil surface
Which tools read private data, which read text someone else can write, and which can send data out. One of each is a path a prompt injection can use.
2 paths in this server alone from private data, through text someone else wrote, to a way out. Denying 1 outbound tool breaks every one.
- Reads private data2
- Reads files, mail, messages, databases, secrets, calendars or repositories.
- Reads text others write1
- Brings in text someone else can write: web pages, inbound email, messages, issues and comments.
- Can send data out1
- Can carry data to people or systems: it sends, posts or publishes, or takes a recipient, channel or URL.
{
"permissions": {
"deny": [
"mcp__hai-agents__send_message"
]
}
}Other servers in the same agent add their own tools to these paths. Audit your whole setup to see them together.
Remote posture
What anyone on the internet sees at the remote endpoint: an anonymous initialize and tools/list, and the public OAuth metadata.
Looked at Nov 20, 2026, 19:05 UTC, unchanged since Oct 22, 2026. No credentials were sent and no tool was called. What is looked at.
- Reachable
- yes
- HTTP status of an anonymous initialize
- 200
- tools/list served without credentials
- yes
- Tools that change things, listed without credentials
- 5
- Protocol version
- 2025-06-18
- WWW-Authenticate on 401
- not observed
- resource_metadata in WWW-Authenticate
- not observed
- RFC 9728 protected resource metadata
- not observed
- Authorization servers
- not observed
- Legacy HTTP+SSE transport
- no
- Dynamic client registration
- not observed
- Client ID metadata documents
- not observed
- HighThe whole server
The endpoint served tools/list without credentials, and the list has 5 tools that change things (1 destructive, 1 irreversible, 3 write). Whether a call also works without credentials was not tested: tools are never called.
What this check looks forMatched text
https://agp.hcompany.ai/mcp: tools/list returned 9 tools to an anonymous client, including cancel_session (destructive), send_message (irreversible), run_agent (write), share_session (write), write_file (write)
Findings
Matches of fixed checks, with the text they matched.
No check matched anything in these tool definitions. That is not a safety guarantee.
History
Changes to its tool definitions and to its remote endpoint, newest first.
Tools3 changed.~3 changedView diff
- Changed:list_filesOnly readsdescriptioninputSchema
- Changed:read_fileOnly readsdescription
- Changed:write_fileChanges thingsdescription
Install
Pinned to version 1.0.0. Values in angle brackets are yours to fill in.
claude mcp add --transport http hai-agents https://agp.hcompany.ai/mcp --header 'Authorization: <your-authorization>'claude mcp add --transport http hai-agents https://agp.hcompany.ai/mcp --header 'Authorization: <your-authorization>'Run in a terminal.
Claude Code permissions
Allow the reads, ask before a change or a way out, deny the deletes. Paste into .claude/settings.json.
{
"permissions": {
"allow": [
"mcp__hai-agents__wait_for_session",
"mcp__hai-agents__list_agents",
"mcp__hai-agents__list_files",
"mcp__hai-agents__read_file"
],
"ask": [
"mcp__hai-agents__run_agent",
"mcp__hai-agents__send_message",
"mcp__hai-agents__share_session",
"mcp__hai-agents__write_file"
],
"deny": [
"mcp__hai-agents__cancel_session"
]
}
}Tools that only read are allowed, tools that change things ask first, tools that delete or cancel are denied. A read that can send data out (it takes a URL, say) asks first too. It assumes the server is installed as hai-agents. Review it before you use it.
Capability tags behind 3 rules
- mcp__hai-agents__send_message outbound
- mcp__hai-agents__list_files private-data, untrusted-input
- mcp__hai-agents__read_file private-data
Pin
Keep the decision: fail CI when the tools change, follow every change, or show a badge.
Pin in CI
Fail a build when the tools you reviewed are not the tools the server serves today.
curl -s "https://radar.anlyon.com/api/v1/servers/io.github.hcompai/hai-agents/verify?hash=14050e65854bb7cd03f3039241b56e264573f2cc4640e4d40c88ed420b9948e8" | jq -e '.changed == false'Pinned hash 14050e65854b, from Nov 19, 2026, 20:17 UTC.
Watch
Get told when its tools change: a new tool, a new description, a new class.
https://radar.anlyon.com/servers/io.github.hcompai/hai-agents/changes.xmlAn Atom feed. Works in any feed reader or Slack's RSS app.
Badge
[](https://radar.anlyon.com/servers/io.github.hcompai/hai-agents)Source: MCP Registry · repository. Synced Oct 21, 2026, 18:35 UTC.
Computer-Use Agents API MCP server: questions and answers
Answered from the scan above, not from the publisher's description.
Is the Computer-Use Agents API MCP server safe?
- No scan can promise that. What its definitions show: 9 tools, 5 of which can change, send or delete something, and 0 findings from fixed checks. Allow the reads, ask before each change, and deny the deletes; the permissions block on this page does that.
Can the Computer-Use Agents API MCP server delete data?
- Yes. 1 tool can delete, cancel or revoke something: cancel_session.
Can the Computer-Use Agents API MCP server do things that cannot be undone?
- Yes. 1 tool can send, pay, publish or merge: send_message.
What tools does the Computer-Use Agents API MCP server have?
- 9 tools: 1 destructive, 1 irreversible, 3 write, 4 read. cancel_session, send_message, run_agent, share_session, write_file, wait_for_session, list_agents, list_files and 1 more.
Does the Computer-Use Agents API MCP server need sign-in or OAuth?
- Its remote endpoint listed its tools to an anonymous client, without sign-in.
Does the Computer-Use Agents API MCP server send data to outside services?
- 1 tool can carry data out (it sends, posts or publishes, or takes a recipient, channel or URL): send_message.
Has the Computer-Use Agents API MCP server changed its tools?
- Yes, once since Anlyon Radar started tracking it, most recently on Nov 5, 2026. Every change is on this page and in its Atom feed.