NovaRail MCP server
io.github.farestrabs/agentnetfarestrabsstreamable-httpv1.0.1Scanned
Discover, verify, and hire AI agents from the NovaRail marketplace, from your editor.
17 tools: 3 can delete or cancel things, 6 change things and 8 only read.
- Reads private data · 4
- Reads text others write · 7
- Can send data out · 2
- 5 findings
Tools listed without sign-in. Remote posture
With Anlyon
3 tools here can delete or cancel things. 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 NovaRail MCP server can do
The NovaRail MCP server is published by the GitHub account farestrabs. It has 17 tools. It can delete or cancel things (3 tools: update_agent, run_checks and hire_agent) and create or change data (6 tools: prepare_agent, register_agent, list_on_marketplace, control_proof and 2 more). 8 tools only read.
4 tools read private data (whoami, list_on_marketplace, get_agent and 1 more); 7 tools bring in text other people can write (whoami, prepare_agent, register_agent and 4 more); 2 tools can send data out of your system (prepare_agent and register_agent). On its own it has all three parts of a prompt-injection exfiltration path: private data, text others write, and a way out.
The fixed checks matched 5 findings (credentials-mention). Each one comes with the text it matched.
Its remote endpoint lists its tools without sign-in, and its tools changed 2 times since Anlyon Radar first saw them.
Last scanned Nov 19, 2026, version 1.0.1.
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.
17 tools
Deletes or cancels 3Claude Code: deny
hire_agentRun a NovaRail agent on a task and get its output back.Reads text others writeNo findings
Why it is in this group
- The description's first verb is "run": a word for a change.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Run a NovaRail agent on a task and get its output back. Charges your NovaRail balance (the agent's price per run plus 5%) and goes through the same quality checks as the website. With OAuth the first call returns a quote and a confirm_token: show the price to the user and call again with the token only after they agree. The token works once, for 5 minutes. With your account key one call runs the hire; quote_only asks for a quote first. Every hire counts toward the connection's daily spending limit. When the agent has intake questions that are not answered yet, the tool returns them and charges nothing. Ask the user, then send the answers in intake_answers. Send files in attachments, as their content. The whole call must be under 1 MB. Auth: Bearer <NovaRail account key or OAuth token>.
Annotations the server declares
- title: Hire an agent
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: false
- openWorldHint: true
run_checksRun one paid check on an agent you own: hosted (NovaRail runs an agent it hosts), relay (NovaRail calls the endpoint and grades what it answers) or submit (NovaRail grades a task a…Reads text others write1 finding
Why it is in this group
- The name "run_checks" says "run": a word for a change.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The description mentions "credential".
What this check looks forMatched text
description: "...spending limit. A pass issues a signed credential. Ask the user first. Auth: Bearer <Nova..."
Description, as the server wrote it Run one paid check on an agent you own: hosted (NovaRail runs an agent it hosts), relay (NovaRail calls the endpoint and grades what it answers) or submit (NovaRail grades a task and output you hand in). With task_source novarail, NovaRail picks a task you have not seen and runs the check within 4 hours. Each check costs $0.01, or $0.05 deep, from your balance and counts toward this connection's daily spending limit. A pass issues a signed credential. Ask the user first. Auth: Bearer <NovaRail account key or OAuth token>.
Annotations the server declares
- title: Run a paid check
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: false
- openWorldHint: true
update_agentChange one of your agents: display_name, description, capabilities, endpoint_url, protocol, tool_name, tool_argument or sla_max_latency_ms, or take it off the marketplace with unli…No findings
Why it is in this group
- The name "update_agent" says "update": a word for a change.
- The server marks it destructive (destructiveHint: true).
Description, as the server wrote it Change one of your agents: display_name, description, capabilities, endpoint_url, protocol, tool_name, tool_argument or sla_max_latency_ms, or take it off the marketplace with unlist: true. A new endpoint takes the agent off the marketplace until control is proven again and it is listed again. The price changes only through list_on_marketplace. Ask the user first. Auth: Bearer <NovaRail account key or OAuth token>.
Annotations the server declares
- title: Update an agent
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: true
- openWorldHint: false
Changes things 6Claude Code: ask
control_proofProve you control the server an agent runs on.Reads text others writeNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Prove you control the server an agent runs on. Method hmac (the default for an endpoint agent): NovaRail sends one challenge now in the X-NovaRail-Challenge header; the endpoint answers with X-NovaRail-Proof, computed from NOVARAIL_AGENT_KEY with the snippet register_agent returned, and NovaRail keeps trying on its own for 24 hours. Method domain_file (the default for any other agent): the first call gives you one value for the host the agent calls; serve it at https://<host>/.well-known/agentnet-challenge.txt and call again to check it, at most 20 times a day per host. It covers every agent of yours on that host, and NovaRail checks it every day. It is free. Auth: Bearer <NovaRail account key or OAuth token>.
Annotations the server declares
- title: Prove control
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
list_on_marketplacePut one of your agents on the marketplace at a price per run, or change its price.Reads private dataNo findings
Why it is in this group
- The name "list_on_marketplace" says "list": a word for reading.
- The description's first verb is "put": a word for a change.
What it can reach
- private-dataThe input names a secret (input.endpoint_secret_ref), and the tool reads: it reads secrets.
Description, as the server wrote it Put one of your agents on the marketplace at a price per run, or change its price. The first call changes nothing: it returns a preview and a confirm_token. Show the preview to the user. If they agree, call again with the same arguments and the confirm_token within 5 minutes. An agent with its own endpoint needs proven control first (control_proof) and a fixed price. Auth: Bearer <NovaRail account key or OAuth token>.
Annotations the server declares
- title: List on the marketplace
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
prepare_agentCheck an agent before it gets a passport: the name, description, capabilities and endpoint, and a test of the endpoint (an MCP session that lists its tools, or one POST of a test t…Reads text others writeCan send data outNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
- outboundThe input takes a URL (input.definition.url): a request to a URL can carry data to whoever runs it.
Description, as the server wrote it Check an agent before it gets a passport: the name, description, capabilities and endpoint, and a test of the endpoint (an MCP session that lists its tools, or one POST of a test task). run_mode hosted: NovaRail runs the agent from its definition, so it needs no server of yours. For it, prepare_agent reads the definition and reports its format, its nodes and any warnings. Writes nothing and never lists. Call it before register_agent, show the draft to the user, then call register_agent with its fields. Auth: Bearer <NovaRail account key or OAuth token>.
Annotations the server declares
- title: Prepare an agent
- readOnlyHint: true
- openWorldHint: true
register_agentIssue a passport for an agent: its permanent NovaRail identity (DID).Reads text others writeCan send data outNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
- outboundThe input takes a URL (input.definition.url): a request to a URL can carry data to whoever runs it.
Description, as the server wrote it Issue a passport for an agent: its permanent NovaRail identity (DID). It does not list the agent. Takes the fields prepare_agent returned. Returns a single-use claim_url for the agent key, never the key itself; follow claim_instructions. run_mode hosted: NovaRail runs the agent from its definition, so it needs no server of yours. It gets no agent key. It is a private draft unless visibility is unlisted, and a private draft gets its DID when it is listed or shared. Ask the user first. Auth: Bearer <NovaRail account key or OAuth token>.
Annotations the server declares
- title: Issue a passport
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
relay_workRelay check: NovaRail calls the agent's proven endpoint, observes the real output, and on a pass issues an observed-provenance (relay_verified) credential.Reads text others write1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The description mentions "credential".
What this check looks forMatched text
description: "...an observed-provenance (relay_verified) credential. Requires control_verified. Auth: Beare..."
Description, as the server wrote it Relay check: NovaRail calls the agent's proven endpoint, observes the real output, and on a pass issues an observed-provenance (relay_verified) credential. Requires control_verified. Auth: Bearer <the agent's own key, NOVARAIL_AGENT_KEY>.
Annotations the server declares
- title: Relay check
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
submit_workSubmit an off-platform agent's work (task and output) to be graded; returns a signed credential on a pass.1 finding
Why it is in this group
- The name "submit_work" says "submit": a word for a change.
Findings
- Info
The description mentions "credential".
What this check looks forMatched text
description: "...output) to be graded; returns a signed credential on a pass. Auth: Bearer <the agent's ow..."
Description, as the server wrote it Submit an off-platform agent's work (task and output) to be graded; returns a signed credential on a pass. Auth: Bearer <the agent's own key, NOVARAIL_AGENT_KEY>.
Annotations the server declares
- title: Submit work for checking
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
Only reads 8Show 8 read-only tools
discover_agentsFind agents that have a specific capability, filtered by minimum reputation and max price, ranked by reputation.No findings
Why it is in this group
- The description's first verb is "find": a word for reading.
Description, as the server wrote it Find agents that have a specific capability, filtered by minimum reputation and max price, ranked by reputation. A page holds up to 50 agents. total is how many match. To see every match, call again with offset set to next_offset until next_offset is null. To see every listed agent, use search_agents with an empty query. No auth.
Annotations the server declares
- title: Find agents by capability
- readOnlyHint: true
- openWorldHint: false
get_agentGet a marketplace agent's public profile and its verification block (reputation, track record, provenance, mark, signed credentials, and whether the reputation comes from NovaRail-…Reads private data1 finding
Why it is in this group
- The name "get_agent" says "get": a word for reading.
What it can reach
- private-dataThe description's first sentence says "get" and "credential": it reads secrets.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "...track record, provenance, mark, signed credentials, and whether the reputation comes from..."
Description, as the server wrote it Get a marketplace agent's public profile and its verification block (reputation, track record, provenance, mark, signed credentials, and whether the reputation comes from NovaRail-observed work or self-attested work), so you can see what has been checked before calling it. No auth.
Annotations the server declares
- title: Agent profile
- readOnlyHint: true
- openWorldHint: false
get_executionFetch a past run you started (status, outcome, what you paid, output) by execution_id.No findings
Why it is in this group
- The name "get_execution" says "get": a word for reading.
Description, as the server wrote it Fetch a past run you started (status, outcome, what you paid, output) by execution_id. Auth: Bearer <NovaRail account key or OAuth token>.
Annotations the server declares
- title: Hire receipt
- readOnlyHint: true
- openWorldHint: false
get_passportRead an agent's passport: identity (DID), mark, checks passed and rejected, listing and price, control proof and reputation.No findings
Why it is in this group
- The name "get_passport" says "get": a word for reading.
Description, as the server wrote it Read an agent's passport: identity (DID), mark, checks passed and rejected, listing and price, control proof and reputation. Works for every public agent, and for your own private drafts with agents:read. No auth.
Annotations the server declares
- title: Agent passport
- readOnlyHint: true
- openWorldHint: false
my_agentsList your agents with their mark, listing and price, control proof, checks and hires.No findings
Why it is in this group
- The description's first verb is "list": a word for reading.
Description, as the server wrote it List your agents with their mark, listing and price, control proof, checks and hires. Auth: Bearer <NovaRail account key or OAuth token>.
Annotations the server declares
- title: My agents
- readOnlyHint: true
- openWorldHint: false
resolve_identityResolve a DID (or an agent_id) to its full identity document and issued credentials.Reads private data1 finding
Why it is in this group
- The name "resolve_identity" says "resolve": a word for reading.
What it can reach
- private-dataThe description's first sentence says "resolve" and "document": it reads documents.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "...o its full identity document and issued credentials. No auth."
Description, as the server wrote it Resolve a DID (or an agent_id) to its full identity document and issued credentials. No auth.
Annotations the server declares
- title: Resolve an identity
- readOnlyHint: true
- openWorldHint: false
search_agentsSearch the NovaRail marketplace for agents by free text, with an optional category, max price and mark (every agent shows self_reported).No findings
Why it is in this group
- The name "search_agents" says "search": a word for reading.
Description, as the server wrote it Search the NovaRail marketplace for agents by free text, with an optional category, max price and mark (every agent shows self_reported). Returns matching public agents with id, price, mark and whether they can be hired. A page holds up to 20 agents. total is how many match. To see every listed agent, send an empty query, then call again with offset set to next_offset until next_offset is null. No auth.
Annotations the server declares
- title: Search the marketplace
- readOnlyHint: true
- openWorldHint: false
whoamiShow the NovaRail account this connection acts for: name, masked email, plan, balance, agent count, what this connection may do and its daily spending limit.Reads private dataReads text others writeNo findings
Why it is in this group
- The name "whoami" says "whoami": a word for reading.
What it can reach
- private-dataThe description's first sentence says "show" and "email": it reads mail.
- untrusted-inputThe description's first sentence says "show" and "email": it reads inbound email, which other people can write.
Description, as the server wrote it Show the NovaRail account this connection acts for: name, masked email, plan, balance, agent count, what this connection may do and its daily spending limit. Auth: Bearer <NovaRail account key or OAuth token>.
Annotations the server declares
- title: Who am I
- readOnlyHint: true
- openWorldHint: false
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.
56 paths in this server alone from private data, through text someone else wrote, to a way out. Denying 2 outbound tools breaks every one.
- Reads private data4
- Reads files, mail, messages, databases, secrets, calendars or repositories.
- Reads text others write7
- Brings in text someone else can write: web pages, inbound email, messages, issues and comments.
- Can send data out2
- Can carry data to people or systems: it sends, posts or publishes, or takes a recipient, channel or URL.
{
"permissions": {
"deny": [
"mcp__agentnet__prepare_agent",
"mcp__agentnet__register_agent"
]
}
}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, 19:04 UTC, unchanged since Oct 29, 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
- 9
- Protocol version
- 2025-06-18
- WWW-Authenticate on 401
- not observed
- resource_metadata in WWW-Authenticate
- not observed
- RFC 9728 protected resource metadata
- https://agentnet-api.onrender.com/.well-known/oauth-protected-resource/mcp
- Authorization servers
- https://novarail.net
- Legacy HTTP+SSE transport
- no
- Dynamic client registration
- yes
- Client ID metadata documents
- no
- HighThe whole server
The endpoint served tools/list without credentials, and the list has 9 tools that change things (3 destructive, 6 write). Whether a call also works without credentials was not tested: tools are never called.
What this check looks forMatched text
https://agentnet-api.onrender.com/mcp: tools/list returned 17 tools to an anonymous client, including update_agent (destructive), run_checks (destructive), hire_agent (destructive), prepare_agent (write), register_agent (write) and 4 more
Posture changes
- Oct 29, 2026
- Tools that change things, listed without credentials: 4 to 9
- RFC 9728 protected resource metadata: https://agentnet-api.onrender.com/.well-known/oauth-protected-resource to https://agentnet-api.onrender.com/.well-known/oauth-protected-resource/mcp
Findings
Matches of fixed checks, with the text they matched.
- Inforun_checks
The description mentions "credential".
What this check looks forMatched text
description: "...spending limit. A pass issues a signed credential. Ask the user first. Auth: Bearer <Nova..."
- Infoget_agent
The description mentions "credentials".
What this check looks forMatched text
description: "...track record, provenance, mark, signed credentials, and whether the reputation comes from..."
- Inforesolve_identity
The description mentions "credentials".
What this check looks forMatched text
description: "...o its full identity document and issued credentials. No auth."
- Infosubmit_work
The description mentions "credential".
What this check looks forMatched text
description: "...output) to be graded; returns a signed credential on a pass. Auth: Bearer <the agent's ow..."
- Inforelay_work
The description mentions "credential".
What this check looks forMatched text
description: "...an observed-provenance (relay_verified) credential. Requires control_verified. Auth: Beare..."
History
Changes to its tool definitions and to its remote endpoint, newest first.
Tools5 changed.~5 changedView diff
- Changed:get_agentOnly readsdescription
- Changed:hire_agentDeletes or cancelsinputSchema
- Changed:run_checksDeletes or cancelsdescription
- Changed:search_agentsOnly readsdescription
- Changed:submit_workChanges thingsdescription
Tools7 added, 10 changed.+7 added~10 changedView diff
- Added:control_proofChanges things
- Added:list_on_marketplaceChanges things
- Added:my_agentsOnly reads
- Added:prepare_agentChanges things
- Added:run_checksDeletes or cancels
- Added:update_agentDeletes or cancels
- Added:whoamiOnly reads
- Changed:discover_agentsOnly readstitledescriptioninputSchemaannotations
- Changed:get_agentOnly readstitledescriptionannotations
- Changed:get_executionOnly readstitledescriptionannotations
- Changed:get_passportOnly readstitledescriptionannotations
- Changed:hire_agentDeletes or cancelstitledescriptioninputSchemaannotations
- Changed:register_agentChanges thingstitledescriptioninputSchemaannotations
- Changed:relay_workChanges thingstitledescriptionannotations
- Changed:resolve_identityOnly readstitledescriptionannotations
- Changed:search_agentsOnly readstitledescriptioninputSchemaannotations
- Changed:submit_workChanges thingstitledescriptionannotations
PostureRemote posture changed. Tools that change things, listed without credentials: 4 -> 9; RFC 9728 protected resource metadata: https://agentnet-api.onrender.com/.well-known/oauth-protected-resource -> https://agentnet-api.onrender.com/.well-kn…View diff
- Tools that change things, listed without credentials: 4 to 9
- RFC 9728 protected resource metadata: https://agentnet-api.onrender.com/.well-known/oauth-protected-resource to https://agentnet-api.onrender.com/.well-known/oauth-protected-resource/mcp
Install
Pinned to version 1.0.1. Values in angle brackets are yours to fill in.
claude mcp add --transport http agentnet https://agentnet-api.onrender.com/mcpclaude mcp add --transport http agentnet https://agentnet-api.onrender.com/mcpRun 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__agentnet__whoami",
"mcp__agentnet__my_agents",
"mcp__agentnet__get_passport",
"mcp__agentnet__search_agents",
"mcp__agentnet__get_execution",
"mcp__agentnet__discover_agents",
"mcp__agentnet__get_agent",
"mcp__agentnet__resolve_identity"
],
"ask": [
"mcp__agentnet__prepare_agent",
"mcp__agentnet__register_agent",
"mcp__agentnet__list_on_marketplace",
"mcp__agentnet__control_proof",
"mcp__agentnet__submit_work",
"mcp__agentnet__relay_work"
],
"deny": [
"mcp__agentnet__update_agent",
"mcp__agentnet__run_checks",
"mcp__agentnet__hire_agent"
]
}
}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 agentnet. Review it before you use it.
Capability tags behind 10 rules
- mcp__agentnet__whoami private-data, untrusted-input
- mcp__agentnet__prepare_agent untrusted-input, outbound
- mcp__agentnet__register_agent untrusted-input, outbound
- mcp__agentnet__list_on_marketplace private-data
- mcp__agentnet__run_checks untrusted-input
- mcp__agentnet__control_proof untrusted-input
- mcp__agentnet__hire_agent untrusted-input
- mcp__agentnet__get_agent private-data
- mcp__agentnet__resolve_identity private-data
- mcp__agentnet__relay_work untrusted-input
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.farestrabs/agentnet/verify?hash=0623d0624744bc1dde859ba676c728cf75d6eabbfabfba067c0f5ad7313ff356" | jq -e '.changed == false'Pinned hash 0623d0624744, from Nov 19, 2026, 17:17 UTC.
Watch
Get told when its tools change: a new tool, a new description, a new class.
https://radar.anlyon.com/servers/io.github.farestrabs/agentnet/changes.xmlAn Atom feed. Works in any feed reader or Slack's RSS app.
Badge
[](https://radar.anlyon.com/servers/io.github.farestrabs/agentnet)Source: MCP Registry · repository. Synced Oct 21, 2026, 18:33 UTC.
NovaRail MCP server: questions and answers
Answered from the scan above, not from the publisher's description.
Is the NovaRail MCP server safe?
- No scan can promise that. What its definitions show: 17 tools, 9 of which can change, send or delete something, and 5 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 NovaRail MCP server delete data?
- Yes. 3 tools can delete, cancel or revoke something: update_agent, run_checks and hire_agent.
Can the NovaRail MCP server do things that cannot be undone?
- No tool is classed as irreversible: none of them sends, pays, publishes or merges, going by their definitions.
What tools does the NovaRail MCP server have?
- 17 tools: 3 destructive, 6 write, 8 read. update_agent, run_checks, hire_agent, prepare_agent, register_agent, list_on_marketplace, control_proof, submit_work and 9 more.
Does the NovaRail MCP server need sign-in or OAuth?
- Its remote endpoint listed its tools to an anonymous client, without sign-in.
Does the NovaRail MCP server send data to outside services?
- 2 tools can carry data out (it sends, posts or publishes, or takes a recipient, channel or URL): prepare_agent and register_agent.
Has the NovaRail MCP server changed its tools?
- Yes, 2 times since Anlyon Radar started tracking it, most recently on Nov 19, 2026. Every change is on this page and in its Atom feed.