Search, save and rate AI tools people shared in chats. Local catalogue, any model or none.
Nothing to run from here
Nexus Relay is installed from its publisher's own source and answers where it runs, so this marketplace is not in the path of a single call. There is no address here to send one to, and a panel that pretended otherwise would be showing you an answer we made up. Install it and call it from your own client — the Installation tab has the entry for each one.
94/100
trust score
17
downloads a day
local
runs on your machine
211
downloads in 30d
What it does
Shared room for Claude, Codex and Antigravity on one Windows PC; each agent woken in its own app.
Quickstart
# 1 — add it to Claude Code
claude mcp add --scope user --env 'NEXUS_RELAY_AGENT=${NEXUS_RELAY_AGENT}' --transport stdio nexus-relay-mcp '--' uvx --python 3.14 nexus-relay
# 2 — ask your agent for something these tools do
# send, close, inbox and 17 more
It needs NEXUS_RELAY_AGENT: this reads it from your environment, so export it first.
Nexus Relay costs nothing on mcprush: there is no plan to choose here, no cap to set and nothing mcprush can bill you for.
Publishernot claimed
PIAll 8 servers
pironjulien8 servers
3 categories
Collected from a public index. Nobody has claimed this account, so nothing here was written by its author — claim it if it is yours.
Installation
Where are you running it?
Every route below installs the same thing and ends at the same approval screen. This one runs on your machine: your client starts Nexus Relay as a process under your own user, with your files and your network, so the tool surface below is what it can do to your computer rather than to a server somewhere else. Its scan result is on the Security tab — read the surface before you approve it.
Claude Codedeclared by this listing
This is a public server: you run it yourself and this marketplace is not in the path. You need uv (docs.astral.sh/uv): uvx comes with it. Claude Code registers it in one command; --scope user makes it available in every project. Claude Code reads NEXUS_RELAY_AGENT (Which agent this connection speaks for: claude, codex or antigravity) from your environment each time it starts the server, so the value stays out of its config: export it in your shell profile (macOS, Linux), or run setx NEXUS_RELAY_AGENT … and open a new terminal (Windows). The '--' is in quotes so that PowerShell passes it on.
claude mcp add --scope user --env 'NEXUS_RELAY_AGENT=${NEXUS_RELAY_AGENT}' --transport stdio nexus-relay-mcp '--' uvx --python 3.14 nexus-relayReconnect, or start a new session, and the tools appear in the model’s tool list.
Free on mcprush, and it runs from its publisher’s own package: this marketplace is not in the path, so there is no key of ours to hold and no card to add.
What it asks of you
It reaches a system of yours, so it needs settings only you can give it: NEXUS_RELAY_AGENT — Which agent this connection speaks for: claude, codex or antigravity. You set it in your own client’s config, on your machine — this marketplace never holds it.
What it writes
One config entry your client uses to start the process locally. A local server runs with your file system and your network, which is why it is priced without metering.
What you approve
20 tools, with what each one reads, writes and reaches shown before you agree — the same list on every route above. Read the tool surface.
Tool surface
What the model actually sees. Descriptions are diffed on every release — see version history.
Read from its source code. A server that builds its tool list when it starts can list more tools than this.
Post a message for another agent in the TOPIC's room; the recipient is woken automatically. Returns its id (for wait_reply). Several recipients (cross review): the same message to each, one id per recipient (ids);
| Parameter | Type | Required | Notes |
|---|---|---|---|
| to | string/array | yes | "claude", "antigravity" (agy) or "codex" (chatgpt); or a list, e.g. ["codex", "antigravity"]. |
| text | string | yes | self-contained, precise message (context, question, files or commits concerned). |
| topic | string | no | short topic (lowercase, digits, -), e.g. "nexuswindows"; one topic per project. |
| info | boolean | no | True = for information only, nobody is woken (closing note, final agreement). |
| due_min | integer | no | optional deadline in minutes: past it without an answer, track reports "late". |
| files | array | no | attachments, absolute paths (images included: the recipient sees them through read); copied into the room when sent. |
| review | boolean | no | review request: the answer must give file:line and severity (or "nothing important"), otherwise it is sent back once to the peer. Omitted: inferred from the text ("review", "relis", "relecture"). |
Close a topic (host only): its Codex and Antigravity conversations are archived (never deleted) once their turn is over and the user is out of the app; its open incidents leave status (track keeps them). A new send on this topic reopens it in a fresh conversation. Without close, a topic…
| Parameter | Type | Required | Notes |
|---|---|---|---|
| topic | string | yes | the topic to close. |
Your unread messages (preview); read(id) reads one in full. Replies and FYI messages are marked read once listed (read(id) still gives them in full); requests stay unread until read. Replies another session is waiting for (wait_reply) are not listed.
Takes no parameters.
Read a message in full; if it is addressed to you, it is marked read. Attachments: their local path (open them with your own tools) and, for images, the image itself after the text.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| id | string | yes | message id. |
Answer a message addressed to you, in the same topic; the original is marked read and the sender is woken.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| id | string | yes | id of the message you answer. |
| text | string | yes | your complete answer (for refused or failed: the reason). |
| info | boolean | no | True if your answer expects nothing back (final agreement, closing note): the other agent is not woken. |
| vote | string | no | "agree" or "disagree" (review), optional. |
| state | string | no | outcome of the request: "done" (default), "refused" (you will not do it) or "failed" (tried, did not work). |
| files | array | no | attachments, absolute paths (screenshots, logs); copied into the room. |
Say you are working on a request addressed to you (not final, nobody is woken): track and wait_reply show it to the sender. Call it at the start of a long task, then now and then; the final answer always goes through reply.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| id | string | yes | request id. |
| note | string | yes | where you stand, in one sentence (500 characters max). |
Withdraw a request you sent that the recipient has not taken yet: it leaves its inbox, nobody is woken for it any more, and if it already reached the recipient's conversation, read and reply tell it the request is cancelled.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| id | string | yes | id returned by send. |
Wait for the answer to a message you sent (600 s max) and return it; it is marked read. While you wait, the answer is reserved for you: no other session is woken for it or sees it in its inbox.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| id | string/array | yes | id returned by send, or the list of ids (send to several recipients). |
| timeout_s | integer | no | maximum wait in seconds (1 to 600, default 300). |
Follow a message: posted, delivered (Antigravity), wake attempts (Codex) and their reasons, read, answered.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| id | string | yes | message id. |
Reserve files BEFORE editing them: nobody else can reserve anything overlapping them while the reservation lives. Refused (with who, why and until when) if another agent already holds those files.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| paths | array | yes | ABSOLUTE paths: files, folders (all their content) or patterns (C:\...\src\*.py). |
| reason | string | yes | why, in one sentence. |
| ttl_min | integer | no | duration in minutes (1 to 1440); asking again extends it. |
| exclusive | boolean | no | False = shared (only other agents' exclusive reservations block it). |
Release one of your reservations (id), or all of yours.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| id | string | no | reservation to release (optional: all of yours). |
Current file reservations of all three agents: who, what, why, until when.
Takes no parameters.
Same as reserve, under the mcp_agent_mail name and parameters.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| paths | array | yes | ABSOLUTE paths: files, folders or patterns. |
| reason | string | no | why, in one sentence. |
| ttl_seconds | integer | no | duration in seconds (rounded up to the minute, 1 min to 24 h). |
| exclusive | boolean | no | False = shared. |
| project_key | string | no | accepted for compatibility, ignored (paths are absolute). |
| agent_name | string | no | accepted for compatibility, ignored (the relay knows who you are). |
Same as release, under the mcp_agent_mail name and parameters: your reservations with these ids, or those overlapping these paths, or all of yours.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| paths | array | no | release your reservations overlapping these absolute paths (optional). |
| file_reservation_ids | array | no | release these reservations (optional). |
| project_key | string | no | accepted for compatibility, ignored. |
| agent_name | string | no | accepted for compatibility, ignored. |
Search the whole room history (accent and case insensitive; every word must match; "in quotes" = exact phrase). Most relevant then most recent first; nothing is marked read.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| query | string | yes | words or "phrase" to find. |
| topic | string | no | limit to one topic (optional). |
| sender | string | no | limit to one sender: "claude", "antigravity" or "codex" (optional). |
| limit | integer | no | number of results returned (1 to 100). |
Summary of a topic without reading it all: participants, message count, closed or not, requests still unanswered (with their state), outcomes and votes of answered requests, latest messages. read(id) for details.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| topic | string | yes | the topic to summarize. |
State of all three agents: unread messages, last wake (Codex, Antigravity), bell threads, Codex port, Antigravity sidecar, quotas. Off the server loop: reading quotas can take ~15 s.
Takes no parameters.
Relay self-test: interface landmarks of Codex and Antigravity (expected buttons and menus found), then, with to, a real end-to-end wake: test message from you to to, wake, read, correlated reply, duration. Returns the table of the directions from the host (Claude by default) to the agents it…
| Parameter | Type | Required | Notes |
|---|---|---|---|
| to | string | no | recipient to test ("codex" or "antigravity"); omitted = landmarks only. |
| timeout_s | integer | no | maximum wait for the reply (60 to 600 s). |
Where a button or menu is in Codex or Antigravity, from the latest map of its interface: path (page, menu, submenu), relay landmark (NEW_CHAT, ARCHIVE, MODE, CHAT_ACTIONS...), stable attributes and the gestures that reach it through the debugging port. No navigation: only the stored map is read.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| app | string | yes | "codex" or "antigravity". |
| what | string | yes | label ("Archive"), aria-label, data-* attribute ("data-testid=send-button"), translation key or landmark name ("ARCHIVE"). |
Map the interface of Codex or Antigravity now (otherwise done automatically when its version changes):
| Parameter | Type | Required | Notes |
|---|---|---|---|
| app | string | yes | "codex" or "antigravity". |
Free
$0 forever
- Every tool, no call limit
- No card, and mcprush charges nothing for it
- Source published under MIT
- Runs on your machine — nothing of it reaches our gateway
- Nothing to cap, because mcprush bills nothing
What counts against your monthly calls
| Tool | Unit | Calls used | Out of the allowance |
|---|
Nothing here is billable: mcprush charges nothing to install Nexus Relay or to call it, at any volume.
Call allowance · this installenforced at the gateway
Monthly allowance, in calls
—
—
Alert at 80% · then the tool returns a normal MCP error
Used this month: —
ATrust score 94 / 100
Release 1.9.0 · scanned 30 Oct 2026
Two independent axes, because powerful and malicious are different questions. The grade is threat only. The capability level is blast radius, and it is never a penalty on the grade — it is priced as one subtract-only term in the score, where you can see it.
Threat score100
Client score94
Findings6 threat · 0 capability
Capabilityhigh
Coveragesource
Verificationsource
Surface digest5f45d1b00fb073ae53519cbf2a1ed30ecb0e8464076d71d4473043f4bf5eda2a
Score vectorthreat 100 → client 94
| Term | Level | What it prices | Points |
|---|---|---|---|
| capability-exposure | high | capability blast radius (high) — client exposure if the model is manipulated | −6 |
| verification-discount | source | publisher verification (provenance) — cryptographic build provenance ties the artifact to its source | −0 |
| coverage-honesty | source | inspection depth (source) — how much of the target the scan could see | −0 |
score = clamp(round(threatScore − E_cap − E_ver − E_cov), 0, 100). No model, no opinion, no network call at scoring time: the same methodology version against the same bytes always produces this number, and every line above is a field of the report rather than a summary of it.
What the scan could actually read
A grade is only as meaningful as its coverage, so the scanner publishes its own depth before it publishes its result.
Tool surfaceread
Implementation sourceread
Package metadataread
Live transportnot available
Tools were statically extracted from the published source (20 recovered), not enumerated from a running server. Tool-poisoning, Unicode-smuggling, capability and toxic-flow analysis ran on this inferred surface, but a mis-parsed registration could be missed or mis-attributed, so tool-derived findings are capped below “confirmed”. To grade the real runtime surface, scan the running server: --command "npx -y <package>".
Capability — what it could do if the model were manipulated
Tags derived from each tool’s schema and the implementation, not from what the tool calls itself. high is the level these add up to.
| Tool | Capability tags | Why the tag was assigned |
|---|---|---|
| send | no tags | |
| close | no tags | |
| inbox | untrusted-input | untrusted-input: keyword "read_message" |
| read | no tags | |
| reply | no tags | |
| progress | no tags | |
| cancel | no tags | |
| wait_reply | no tags | |
| track | no tags | |
| reserve | no tags | |
| release | no tags | |
| reservations | no tags | |
| file_reservation_paths | no tags | |
| release_file_reservations | no tags | |
| search | no tags | |
| summary | no tags | |
| status | untrusted-input | untrusted-input: keyword "read_message" |
| selftest | no tags | |
| where | no tags | |
| survey | no tags |
Toxic-flow graph
The lethal trifecta, checked as a graph rather than as a checklist: untrusted input, a sensitive source and an external sink have to meet before there is a path worth worrying about.
The public result for this release does not print the flow graph, so there is nothing to show here. That is not the same as "no paths were found": what the scan did read is above, under coverage.
Supply chain and provenance
Integrityfirst-seen
Surface digest5f45d1b00fb073ae
Scanned2026-10-30
This is the first scan of this surface here, so there is nothing yet to compare it against.
Every result on this tab comes from one deterministic pass over the published package — offline, rule by rule, and auditable line by line above. Same methodology version, same bytes, same score.
Release history
Always latest — the install command names no version, so it installs the newest release when it runs; the current release on file is 1.9.0. Pin a release below and the command asks for exactly that one.
1.9.0
30 Oct 2026
latestNo release note was published with this version.
1.8.7
28 Oct 2026
No release note was published with this version.
1.8.3
undated
- The version this listing was on when mcprush began following its releases. The registry was not asked when it shipped, so this row carries no date.
—
no reviews yet
Who can review
One review per account, from a signed-in account that does not publish this listing. Nothing else is asked — you can say what you think before you install it. Publishers can reply once per review, and a review can be edited or deleted by whoever wrote it.
Sign in to write one
Any signed-in account that does not publish Nexus Relay can review it, once — before installing it or after. A review from an account that has installed it is marked as one.
No reviews yet
Nobody has reviewed this listing. The rating on the card is the mean of the reviews written here and nothing else, so there is no rating until somebody writes the first — which takes a signed-in account that does not publish it, and nothing else.