Validate Formhaus form definitions and simulate multi-step paths against the real engine.
Nothing to run from here
Callwright 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
9
downloads a day
local
runs on your machine
205
downloads in 30d
What it does
A voice agent that places phone calls (reservations, appointments, inquiries) on your behalf, shaped by LLM direction + grounding. Runs fire-and-forget on Retell; an MCP server lets an LLM shape the grounding before each call.
Quickstart
# 1 — start it (keep it running): npx -y callwright
# 2 — add it to Claude Code
claude mcp add --scope user --transport http callwright-mcp http://localhost:8787/mcp
# 3 — ask your agent for something these tools do
# get_setup_status, configure, list_scenarios and 16 more
It needs RETELL_API_KEY: set it in the terminal you start it from. Windows PowerShell says 'running scripts is disabled on this system'? Run Set-ExecutionPolicy -Scope CurrentUser RemoteSigned once, or type npx.cmd instead of npx (codex.cmd, gemini.cmd, grok.cmd likewise). PowerShell 7 needs neither.
Callwright costs nothing on mcprush: there is no plan to choose here, no cap to set and nothing mcprush can bill you for.
Publishernot claimed
TOAll 1 server
topness1 server
1 category
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 Callwright 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 Node.js — the current LTS (24) is the safe choice; npx comes with it. This package runs as a local HTTP server, not over stdio: start it in a terminal with npx -y callwright and keep it running, with RETELL_API_KEY (Your Retell secret API key (stays server-side)) set in that terminal, then add its address, http://localhost:8787/mcp. Claude Code registers it in one command; --scope user makes it available in every project. The transport below is http, as the publisher's own manifest declares it.
claude mcp add --scope user --transport http callwright-mcp http://localhost:8787/mcpReconnect, 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 your own credential rather than the publisher’s: RETELL_API_KEY — Your Retell secret API key (stays server-side). 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
19 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.
Report what callwright has configured (from-number, agent, your name/callback, fact keys) and what's missing. Call this first if you're unsure whether setup is complete; if basics are missing, ask the user for them and call configure.
No parameter schema on file.
Save the user's standing profile so calls need less per-call input. Set name and callback_number (the always-safe basics). Optionally save standing facts (durable PII like service_address, member_id — a STORE the LLM later draws minimal subsets from). This is also the SAVE TARGET for the lazy-save…
No parameter schema on file.
List known scenario profiles (recommended details, default flexibility, aliases). Use to learn what grounding a given call type benefits from before constructing place_call.
No parameter schema on file.
Place a call on the user's behalf. DRY-RUN by default: returns a read-back (including exactly which personal data types will be available to the agent) and does NOT dial. Review it with the user, then call again with confirm:true to actually place the call. DATA MINIMIZATION: in principal.facts…
No parameter schema on file.
Fetch a call's outcome, post-call analysis, and transcript by call_id. The outcome object is authoritative when Retell's raw success boolean conflicts with custom analysis. Use after place_call (give it ~30-60s to complete + analyze).
No parameter schema on file.
List recent calls in this Retell workspace (most recent first).
No parameter schema on file.
Show recent automatic call retries (no-answer/busy/failed re-dials), most recent first. Retries default to at most 1 per call and never auto-escalate. Each entry shows the original call, the attempt number/cap, the new call_id (if placed), and status (scheduled/placed/failed).
No parameter schema on file.
Set up the optional 'text me when the call is done' feature. Saves the user's mobile number as the default notify target, provisions the one-time SMS summary agent if needed, and (with test:true) sends a test SMS now so the user can confirm delivery. Per-call SMS is still opt-in via place_call…
No parameter schema on file.
Improve the system from a completed call. If the call_type matches a profile, it's ENRICHED (deferred questions become recommended_details, value-baked ones rejected). If it has NO profile, the (generalized) scenario is STAGED as a candidate; once seen >=2 times it becomes a PROPOSAL you can…
No parameter schema on file.
Show staged scenario candidates (unmatched call types seen but not yet a profile) with their occurrence counts, variant names, and the questions agents had to defer. Candidates seen >=2 times are 'ready' to propose as a new profile (create_profile) or merge into an existing one (add_scenario_alias).
No parameter schema on file.
Create a NEW scenario profile (propose-and-confirm — call this only after reviewing a learn_from_call proposal with the user). GUARDS: the name must be a GENERAL scenario (not instance-shaped like 'dinner_for_2'); recommended_details must be GENERALIZED field keys ('destination', not…
No parameter schema on file.
Merge a call_type into an EXISTING profile by adding it as an alias (the anti-splintering 'merge, don't create' path). Use when a candidate / new call_type is really the same scenario as a profile you already have. Clears the matching candidate.
No parameter schema on file.
Discard a staged scenario candidate that is noise or a one-off (not a real recurring scenario). Removes it from the candidates store.
No parameter schema on file.
List available Retell voices (for picking a voice when adding a language). Optionally filter by accent (e.g. 'French', 'Japanese') or a free-text query (name/provider/accent). NOTE: Retell voices have an accent, not a hard language field, and most are multilingual — accent is a hint, not a…
No parameter schema on file.
List the call languages this server can conduct calls in: the built-in English base plus any added languages (with their agent, voice, Retell language code, and whether prompt/phrase assets are on the volume). Use to see what's available before place_call or add_language.
No parameter schema on file.
Add a NEW call language at runtime (no redeploy). Two-phase: call with just lang + display_name to get back the base prompt + English phrase block to TRANSLATE; then call again with prompt_text + phrases (translated, preserving every {{variable}} and {placeholder}) to create the language agent.…
No parameter schema on file.
Quality-gate a newly added language before production. DRY-RUN by default: returns a review card with the EXACT spoken opening (greeting + your ask + AI disclosure) and voicemail message — composed by the language's agent — plus a native-review checklist. To HEAR it live, pass to:<your phone> and…
No parameter schema on file.
Change an already-added language: replace its prompt_text and/or phrases (re-translated), and/or change its voice_id, Retell language code, interruption_sensitivity, or stt_mode. Only the provided fields change. Use list_languages to see what's registered.
No parameter schema on file.
Unregister an added language and delete its volume prompt + phrase assets. By default the Retell agent is LEFT in your dashboard (set delete_agent:true to delete it too). English cannot be removed.
No parameter schema on file.
Free
$0 forever
- Every tool, no call limit
- No card, and mcprush charges nothing for it
- Source published under ISC
- 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 Callwright 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 1.0.4 · scanned 14 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
Findings1 threat · 1 capability
Capabilitymoderate
Coveragesource
Verificationrepo
Surface digestc7fe2f5b96315a047ecc5028c3fcadcb2fd51c1fe762bbe44cec3d8d68ccfe60
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 (19 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 |
|---|---|---|
| get_setup_status | no tags | |
| configure | external-sink | external-sink: declares readOnlyHint: false, openWorldHint: true |
| list_scenarios | no tags | |
| place_call | external-sink | external-sink: declares readOnlyHint: false, destructiveHint: true, openWorldHint: true |
| get_call_outcome | untrusted-input | untrusted-input: keyword "fetch", declares openWorldHint: true (interacts with entities outside the server) |
| list_recent_calls | untrusted-input | untrusted-input: declares openWorldHint: true (interacts with entities outside the server) |
| list_retries | no tags | |
| configure_notifications | external-sink | external-sink: keyword "sms", declares readOnlyHint: false, openWorldHint: true |
| learn_from_call | external-sink | external-sink: declares readOnlyHint: false, openWorldHint: true |
| list_candidates | no tags | |
| create_profile | state-write | state-write: declares readOnlyHint: false |
| add_scenario_alias | state-write | state-write: declares readOnlyHint: false |
| reject_candidate | state-write | state-write: declares readOnlyHint: false, destructiveHint: true |
| list_voices | untrusted-input | untrusted-input: declares openWorldHint: true (interacts with entities outside the server) |
| list_languages | no tags | |
| add_language | external-sink | external-sink: declares readOnlyHint: false, openWorldHint: true |
| verify_language | external-sink | external-sink: declares readOnlyHint: false, openWorldHint: true |
| update_language | external-sink | external-sink: declares readOnlyHint: false, openWorldHint: true |
| remove_language | external-sink | external-sink: declares readOnlyHint: false, destructiveHint: true, openWorldHint: true |
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 digestc7fe2f5b96315a04
Scanned2026-10-14
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 1.0.4. Pin a release below and the command asks for exactly that one.
1.0.4
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 Callwright 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.