Community Gmail server with auto-auth: search, read and send email via MCP.
Nothing to run from here
ReplyLayer 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.
96/100
trust score
8
downloads a day
local
or remote
2k
downloads in 30d
What it does
Safe email for AI agents: a mailbox with inbound scanning, outbound checks and human review.
Quickstart
# 1 — add it to Claude Code
claude mcp add --scope user replylayer-mcp '--' npx -y replylayer-mcp
# 2 — ask your agent for something these tools do
# add_inbound_allowlist_entry, add_inbound_allowlist_bulk, add_inbound_blocklist and 41 more
ReplyLayer costs nothing on mcprush: there is no plan to choose here, no cap to set and nothing mcprush can bill you for.
PublisherVerified · since 2026
Installation
Where are you running it?
Every route below installs the same thing and ends at the same approval screen. This one can run either way — at an address its publisher runs, or as a process under your own user, with your files and your network — and this marketplace is in the path of neither.
Claude Codedeclared by this listing
This is a public server: you run it yourself and this marketplace is not in the path. You need Node.js — the current LTS (24) is the safe choice; npx comes with it. Claude Code registers it in one command; --scope user makes it available in every project. The '--' is in quotes so that PowerShell passes it on.
claude mcp add --scope user replylayer-mcp '--' npx -y replylayer-mcpReconnect, or start a new session, and the tools appear in the model’s tool list.
Or connect to its publisher’s address
Its publisher also runs it at https://api.replylayer.ai/v1/mcp: nothing to start on your machine, and this marketplace is not in the path.
claude mcp add --scope user --transport http replylayer-mcp https://api.replylayer.ai/v1/mcpFree 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
Nothing is required, but it can take REPLYLAYER_API_KEY (ReplyLayer API key. Optional if you ran `npx -y replylayer-mcp init`, which stores it locally.) if you want to set 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
44 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.
Add a sender (email or @domain) to the inbound allowlist for a mailbox. When the mailbox is in allowlist mode, only senders matching an entry on this list will be delivered; everyone else lands in state="firewall_blocked". Accepts an exact email or @domain pattern. Idempotent — repeat adds return
No parameter schema on file.
Bulk-add up to 1000 sender email addresses or @domain patterns to the inbound allowlist for a mailbox in a single call. Useful for onboarding/migration when importing an existing trusted-senders list. Returns a partial-success response: added[], already_existed[], invalid[] (with per-email reason),
No parameter schema on file.
Add an email or domain pattern to the inbound (incoming) blocklist. Mirrors add_suppression for the inbound side: customer-protective do-not-receive list. Accepts an exact email (alice@spam.com) or @domain (@spam.com). Server forces reason="manual" + source="customer". Idempotent — repeat adds ret
No parameter schema on file.
Bulk-add up to 1000 sender email addresses or @domain patterns to the account-wide inbound blocklist in a single call. Useful for onboarding/migration when importing an existing spam/blocked-sender list. Returns a partial-success response: added[], already_existed[], invalid[] (with per-email reason
No parameter schema on file.
Add a person to the account's recipient list so you can email them. Only for accounts on the free Sandbox; other accounts don't use this list and get an error here.
No parameter schema on file.
Add an email address OR a domain pattern to the do-not-contact list. Accepts two input forms: an exact email (alice@corp.com) or a bare-domain pattern (@corp.com) that blocks every address at that domain. Once added, all future sends matching the entry from this account will be blocked. This tool is
No parameter schema on file.
Bulk-add up to 1000 email addresses or @domain patterns to the do-not-contact list in a single call. Useful for onboarding/migration when importing an existing suppression/unsubscribe list. Returns a partial-success response: added[], already_existed[], invalid[] (with per-email reason), and a count
No parameter schema on file.
Approve a state="pending_review" message and dispatch it. Wire status reports the dispatch outcome ("sent" or "blocked" — release-style); audit_log + message.review.approved webhook fire regardless of dispatch outcome. Auth: admin API keys + session only — agent keys 403. Optional reason text (max 5
No parameter schema on file.
Block a state="quarantined" INBOUND message (→ terminal blocked, no delivery). INBOUND-ONLY: the tool pre-fetches the message and refuses an outbound row (manage outbound quarantines from the dashboard). Optional reason is persisted to the audit log. Mailbox-bound agent keys can only block messages
No parameter schema on file.
No parameter schema on file.
Soft-delete a draft (sets state="deleted"; hides it from list_drafts). The row remains in the database and the raw MIME body persists in storage for operator recovery. If the draft was scheduled for future dispatch, the schedule is cancelled and a message.schedule_cancelled webhook fires. This delet
No parameter schema on file.
Permanently delete a message: soft-deletes the row and purges its raw MIME and attachment derivatives from object storage. Works on any direction (inbound or outbound) in a deletable state; draft, scheduled, and dispatching rows are refused with a conflict. Idempotent (re-deleting returns success).
No parameter schema on file.
Deny a state="pending_review" message; terminal block (no dispatch). Audit_log + message.review.denied webhook fire on commit. Auth: admin API keys + session only — agent keys 403. Optional reason text (max 500 chars) is persisted to messages.review_reason and included in the audit/webhook payloads.
No parameter schema on file.
Get account usage, limits, storage quota state, and storage byte breakdown.
No parameter schema on file.
Get the current send-budget quota (sends today, effective daily limit, remaining, reset time, bound mailbox IDs). Works with agent-scoped keys. Includes a warmup object while a new paid account ramps on the shared domain.
No parameter schema on file.
Get a derived text preview of an inbound message attachment. Returns extracted text for plain-text, CSV, PDF, and Office (DOCX/PPTX/XLSX) attachments on mailboxes with attachment_exposure_mode="derived_content". Content is hard-capped at 20 000 characters; check preview.truncated and preview.char_co
No parameter schema on file.
Read a specific draft by ID. Returns the full draft including body, scan results, scheduled-send fields, and sub-address fields if set.
No parameter schema on file.
Get malicious link scanning (URL reputation) activation status
No parameter schema on file.
Read a full email thread as one ordered conversation (every message in the thread, oldest → newest, safe-view only). Pass the thread_id from list_threads or a message's thread_id. A thread key can resolve in more than one mailbox — a NOT_FOUND error means the thread is either absent or the key is
No parameter schema on file.
List the recipient allowlist for a given mailbox. Only meaningful when the mailbox is in recipient_policy_mode=allowlist; in blocklist mode the allowlist is not consulted. Use this to check what addresses are approved before attempting a send. Entries may be exact emails (alice@corp.com) OR domain p
No parameter schema on file.
Inspect sends this agent (or a peer agent on the same mailbox) attempted that were rejected by the recipient allowlist (403 RECIPIENT_NOT_ON_ALLOWLIST). Default returns an aggregated top-N summary with one row per recipient and sending key (the key id is not shown; actor_type says admin or agent) wi
No parameter schema on file.
List drafts in a ReplyLayer mailbox. Drafts are scan-then-review-then-send entries; use create_draft to make one and send_draft to dispatch it.
No parameter schema on file.
List per-mailbox inbound allowlist entries (senders permitted in allowlist mode). Mirrors list_allowlist for the inbound side. Read-only. Returns up to limit rows (default 100, max 500); pass cursor to paginate.
No parameter schema on file.
List account-wide inbound sender blocklist entries (the do-not-receive list). Mirrors list_suppressions for the inbound side. Returns up to limit rows (default 100, max 500); pass cursor to paginate. Read-only.
No parameter schema on file.
List incoming senders rejected by the inbound firewall for a mailbox. Default aggregated view groups by (sender, matched_field, mode, reason_code) ordered by most recent. Pass aggregate=false for raw per-attempt history with tuple-cursor pagination. within_days filters by recency (1..365). Read-only
No parameter schema on file.
List all mailboxes on the account. Returns name, full address, status, and the recipient-visible From identity (effective_from_display) for each mailbox.
No parameter schema on file.
List / search messages in a ReplyLayer mailbox. Filter by keyword search, sender, date range, direction, and status — including the held-mail states for triage: status="quarantined" or status="pending_review" (then release_quarantined_message / approve_review). Page older by passing before=<id of th
No parameter schema on file.
List recipients with confirmation status. Sandbox-tier accounts can only send to confirmed recipients.
No parameter schema on file.
List addresses on the do-not-contact list (suppressions). Includes both system-added entries (hard bounces, spam complaints, RFC 8058 unsubscribes) and customer-added entries (reason="manual", source="customer"). Useful for verifying that an address is or is not blocked before attempting to send. En
No parameter schema on file.
List conversation threads in a mailbox (most-recent first), each with participants, unread_count, and last_message_at. Then use get_thread to read one. Page older by passing before_ts = the oldest last_message_at from the previous page. Filter with starred / has_inbound.
No parameter schema on file.
Mark a single message as read. Eligible only for inbound, visible rows. Outbound / deleted / firewall_blocked rows return a 200 no-op with the existing read_at. Idempotent: a second call returns the same read_at. Use this after read_message to advance the customer's read state — read_message itself
No parameter schema on file.
Bulk-mark every visible inbound unread message in a thread as read. Returns marked_count of rows newly stamped this call. Excludes outbound, deleted, and firewall_blocked rows. Idempotent: a thread with everything already read returns marked_count: 0. Use this after fetching a full thread (e.g. via
No parameter schema on file.
Read a specific message by ID. Returns the default plaintext body projection, attachment metadata/status fields, and scan results for human handoff.
No parameter schema on file.
Release a state="firewall_blocked" message back into scanner processing. Returns 202 immediately; actual scanner verdict (available/quarantined/blocked) is observed by re-reading the message detail. Mailbox-bound agent keys can only release messages whose mailbox they are bound to. Releasing does NO
No parameter schema on file.
Release a state="quarantined" INBOUND message back into the inbox (→ available). INBOUND-ONLY: the tool pre-fetches the message and refuses an outbound row (releasing an outbound quarantine would re-send it via the provider — manage those from the dashboard). Optional reason is persisted to the audi
No parameter schema on file.
Suppression removal is intentionally NOT available via MCP. Undoing a do-not-contact entry is a per-human admin decision — use the CLI (rly suppressions remove), an SDK admin key, or the dashboard.
No parameter schema on file.
Reply to a message. The reply is sent from the same mailbox that received the original message, threading correctly. Returns the message ID, status, and the Governed Email-Effect fields: email_effect (effect_status = sent | held_for_review | held_infrastructure | blocked, plus releasable/terminal/re
No parameter schema on file.
Report and block an INBOUND message: blocks a held (quarantined) message and adds the resolved sender to your account-wide inbound blocklist (idempotent). INBOUND-ONLY: the tool pre-fetches the message and refuses an outbound row (an outbound message has no inbound sender to block). Optional reason
No parameter schema on file.
No parameter schema on file.
Send an email from a ReplyLayer mailbox — a fresh send, or continue an existing thread with thread_id. Returns the message ID, status (sent, quarantined, blocked, or pending_review), and the Governed Email-Effect fields: email_effect (effect_status = sent | held_for_review | held_infrastructure | bl
No parameter schema on file.
Star or unstar a message. Pass starred=true to star, starred=false to unstar. Idempotent — re-starring an already-starred message is a no-op. Works for both inbound and outbound messages.
No parameter schema on file.
Star or unstar a thread. Pass starred=true to star, starred=false to unstar. The mailbox parameter accepts a mailbox name or UUID. Idempotent — re-starring an already-starred thread is a no-op.
No parameter schema on file.
No parameter schema on file.
Long-poll for new messages in a mailbox. Returns when a new message arrives or the timeout expires.
No parameter schema on file.
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 ReplyLayer 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 96 / 100
Release 0.15.1 · scanned 13 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 score96
Findings0 threat · 1 capability
Capabilitymoderate
Coveragesource
Verificationrepo
Surface digestad5d7e6a3cd8ae1a502aa2adabfd2be39eb82b27732bc75485c18bd599e38025
Score vectorthreat 100 → client 96
| Term | Level | What it prices | Points |
|---|---|---|---|
| capability-exposure | moderate | capability blast radius (moderate) — client exposure if the model is manipulated | −3 |
| verification-discount | repo | publisher verification (public source) — no provenance, but the source is public and inspectable | −1 |
| 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 (44 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. moderate is the level these add up to.
| Tool | Capability tags | Why the tag was assigned |
|---|---|---|
| add_inbound_allowlist_bulk | state-write | state-write: declares readOnlyHint: false, destructiveHint: true |
| add_inbound_allowlist_entry | state-write | state-write: declares readOnlyHint: false, destructiveHint: true |
| add_inbound_blocklist_bulk | state-write | state-write: declares readOnlyHint: false, destructiveHint: true |
| add_inbound_blocklist | state-write | state-write: declares readOnlyHint: false, destructiveHint: true |
| add_recipient | external-sink | external-sink: declares readOnlyHint: false, destructiveHint: true, openWorldHint: true |
| add_suppression | state-write | state-write: declares readOnlyHint: false, destructiveHint: true |
| add_suppressions_bulk | state-write | state-write: declares readOnlyHint: false, destructiveHint: true |
| approve_review | external-sink | external-sink: keyword "webhook", declares readOnlyHint: false, destructiveHint: true, openWorldHint: true |
| block_quarantined_message | state-write | state-write: declares readOnlyHint: false, destructiveHint: true |
| create_draft | external-sink | external-sink: declares readOnlyHint: false, destructiveHint: true, openWorldHint: true |
| delete_draft | external-sink | external-sink: keyword "webhook", declares readOnlyHint: false, destructiveHint: true, openWorldHint: true |
| delete_message | state-write | state-write: declares readOnlyHint: false, destructiveHint: true |
| deny_review | external-sinkstate-write | external-sink: keyword "webhook"; state-write: declares readOnlyHint: false, destructiveHint: true |
| get_account_usage | no tags | |
| get_agent_quota | no tags | |
| get_attachment_preview | no tags | |
| get_draft | no tags | |
| get_link_scanning_status | no tags | |
| get_thread | no tags | |
| list_allowlist_blocked_attempts | no tags | |
| list_allowlist | no tags | |
| list_drafts | no tags | |
| list_inbound_allowlist | no tags | |
| list_inbound_blocklist | no tags | |
| list_inbound_firewall_blocked_attempts | no tags | |
| list_mailboxes | no tags | |
| list_messages | no tags | |
| list_recipients | no tags | |
| list_suppressions | no tags | |
| list_threads | no tags | |
| mark_message_read | untrusted-inputstate-write | untrusted-input: keyword "read_message"; state-write: declares readOnlyHint: false, destructiveHint: true |
| mark_thread_read | untrusted-inputstate-write | untrusted-input: keyword "read_message"; state-write: declares readOnlyHint: false, destructiveHint: true |
| read_message | untrusted-input | untrusted-input: keyword "read_message" |
| release_firewall_blocked_message | state-write | state-write: declares readOnlyHint: false, destructiveHint: true |
| release_quarantined_message | state-write | state-write: declares readOnlyHint: false, destructiveHint: true |
| remove_suppression | no tags | |
| reply_to_message | external-sink | external-sink: declares readOnlyHint: false, destructiveHint: true, openWorldHint: true |
| report_and_block | state-write | state-write: declares readOnlyHint: false, destructiveHint: true |
| send_draft | external-sink | external-sink: declares readOnlyHint: false, destructiveHint: true, openWorldHint: true |
| send_email | external-sink | external-sink: keyword "send_email", declares readOnlyHint: false, destructiveHint: true, openWorldHint: true |
| star_message | state-write | state-write: declares readOnlyHint: false, destructiveHint: true |
| star_thread | state-write | state-write: declares readOnlyHint: false, destructiveHint: true |
| update_draft | external-sink | external-sink: declares readOnlyHint: false, destructiveHint: true, openWorldHint: true |
| wait_for_message | 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.
flow-1
Untrusted input can drive an external action even though no sensitive source is exposed.
Supply chain and provenance
Integrityfirst-seen
Surface digestad5d7e6a3cd8ae1a
Scanned2026-10-13
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 0.15.1. Pin a release below and the command asks for exactly that one.
0.15.1
undated
latestNo release note was published with this version.
—
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 ReplyLayer 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.