Name: Lunara
Author: DarkMatteros

Lunara MCP

Subtitle engine for AI agents: one upload gives a written track and a translation, timings exact.

Tool-list grade A Remote · Streamable HTTP OAuth 2.1 20 tools v1.0.1

No reviews yet by DarkMatteros The publisher’s own endpoint — this marketplace is not in the path

93/100

tool-list score

tools declared

remote

at its publisher

—

calls last 7d

What it does

Subtitle engine for AI agents: one upload gives a written track and a translation, timings exact.

Quickstart

# 1 — add it to Claude Code
claude mcp add --scope user --transport http lunara-mcp https://lunaraai.cc/mcp

# 2 — ask your agent for something these tools do
#   lunara_list_tools, lunara_get_job, lunara_get_srt and 17 more

Its address takes AUTHORIZATION (Bearer token. Either an OAuth 2.1 access token obtained through https://lunaraai.cc/.well-known/oauth-protected-resource, or a static `lnr_live_…` API key…) in the Authorization header: add it to the entry your client keeps for it.

Lunara costs nothing on mcprush: there is no plan to choose here, no cap to set and nothing mcprush can bill you for.

Publishernot claimed

DarkMatteros1 server

1 category

All 1 server

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. Nothing here runs on your machine and nothing runs on ours: this server answers at its publisher’s own address, and what you install is the connection straight to it.

Claude Codedeclared by this listing

This is a public server: it answers at its publisher’s own address and this marketplace is not in the path. 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. Its publisher asks you to sign in with OAuth: run /mcp in Claude Code and pick this server, and it opens a browser for that once.

claude mcp add --scope user --transport http lunara-mcp https://lunaraai.cc/mcp

Reconnect, or start a new session, and the tools appear in the model’s tool list.

Free on mcprush, and it answers at its publisher’s own address: this marketplace is not in the path, so there is no key of ours to hold and no card to add — the only sign-in is its publisher’s own, with OAuth.

What it asks of you

It reaches a system of yours, so it needs your own credential rather than the publisher’s: Authorization header — Bearer token. Either an OAuth 2.1 access token obtained through https://lunaraai.cc/.well-known/oauth-protected-resource, or a static `lnr_live_…` API key minted at https://lunaraai.cc/settings.. You set it in your own client’s config, on your machine — this marketplace never holds it.

What it writes

One config entry pointing straight at the publisher’s own address — this marketplace is not in the path, and nothing from this listing executes on your machine.

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.

Only tools that write or reach the network

Return every subtitle tool Lunara exposes, with the minutes each one costs. Call this first when you are unsure which tool fits.

No parameter schema on file.

Return the state of a subtitle job: status, progress, whether the SRT is ready, and how many cues it has. Poll this after lunara_upload_video.

Parameter	Type	Required	Notes
jobId	string	yes	The job id returned by lunara_upload_video.

Return the subtitle file for a finished job. One line per cue. Use track="secondary" for the translated track (English by default).

Parameter	Type	Required	Notes
jobId	string	yes	The job id.
track	string	no	primary = the written-language track; secondary = the translation.

Return subtitle cues for a job as timestamps plus text, so you can inspect or reference individual lines before proposing edits.

Parameter	Type	Required	Notes
jobId	string	yes	The job id.
offset	integer	no	Skip this many cues. Default 0.
limit	integer	no	How many cues to return. Default 100.

Run the full eight-layer subtitle pipeline on an audio or video URL you can already host. Produces a written-language track and a translated track. Prefer lunara_upload_video when the file is local.

Parameter	Type	Required	Notes
audioUrl	string	yes	A URL Lunara can download (public or pre-signed).
primaryLang	string	no	Language of the written track. Default yue-Hant.
secondaryLang	string	no	Language of the translated track. Default en.
spokenLang	string	no	What the speaker is speaking, e.g. yue-Hant-HK. Defaults to primaryLang.

Start a persisted subtitle job from a URL you already host — no upload needed. Returns a job id; poll lunara_get_job every 15–30 s until status is "completed", then call lunara_get_srt. Prefer this over lunara_upload_video whenever the media is reachable by URL.

Parameter	Type	Required	Notes
audioUrl	string	yes	Direct HTTP(S) URL to the audio or video file. Must be downloadable by our server.
filename	string	no	Display name for the result. Defaults to the URL basename.
primaryLang	string	no	Language of the written track. Default yue-Hant.
secondaryLang	string	no	Language of the translated track. Default en.
spokenLang	string	no	What the speaker is speaking, e.g. yue-Hant-HK. Defaults to auto-detect.

Reserve an upload slot and return the object key plus a signed URL per part. After uploading the bytes call lunara_finish_upload. Use this for a file you hold locally.

Parameter	Type	Required	Notes
projectId	string	no	Project to upload into. Omit to let Lunara reuse the most recent project for this organisation.
clientOperationId	string	no	Idempotency key. Reuse the same value to retry without creating a second upload.
filename	string	yes	Original filename, used for the download name.
mime	string	yes	MIME type, e.g. audio/mpeg, video/mp4.
sizeBytes	integer	yes	Exact file size in bytes.
primaryLang	string	no	Written-track language. Default yue-Hant.
secondaryLang	string	no	Translated-track language. Default en.

Tell Lunara the bytes are uploaded so the pipeline can start. Returns the job id to poll with lunara_get_job.

Parameter	Type	Required	Notes
videoId	string	yes	The video id from lunara_upload_video.
uploadId	string	yes	The multipart upload id.
key	string	yes	The object key.
durationMs	integer	yes	Media duration in milliseconds.
parts	array	yes	Uploaded parts, each { PartNumber, ETag }.

Return the proper nouns Lunara has learned for this org, so names are spelled consistently across tracks.

Parameter	Type	Required	Notes
limit	integer	no	Default 100.

Run speech recognition only — no rewrite, no translation. Returns timed segments. Use this when you want the transcript itself, or when you plan to pick a later stage yourself.

Parameter	Type	Required	Notes
audioUrl	string	yes	Direct HTTP(S) URL to the audio or video file.
spokenLang	string	no	What the speaker is speaking, e.g. yue-Hant-HK. Defaults to auto-detect.

Translate the text of an existing job’s cues into another language. Every cue boundary and timestamp is preserved; only the text changes. Use this instead of translating yourself so the timing contract holds.

Parameter	Type	Required	Notes
jobId	string	yes	The job whose cues to translate.
targetLang	string	yes	Target language.
track	string	no	Which track to read. Default primary.

Turn spoken Cantonese cue text into written Chinese. Every cue boundary and timestamp is preserved. Deliberate mispronunciations are never corrected, and proper nouns are protected.

Parameter	Type	Required	Notes
jobId	string	yes	The job whose cues to rewrite.
track	string	no	Which track to read. Default primary.

Approximate split, not audio-grounded segmentation: cue text is divided at clause punctuation and each piece receives a share of the original span proportional to its character count. The words are not changed. Timings are estimated, so treat them as a reading aid rather than a re-alignment — use lunara_start_subtitle_job if you need timing derived from the audio.

Parameter	Type	Required	Notes
jobId	string	yes	The job whose cues to re-cut.
maxCueMs	integer	no	Duration ceiling per cue in ms. Default 7000.
track	string	no	Which track to read. Default primary.

Find candidate proper nouns in a job’s cues, ranked by frequency, with the first cue they appear in. Suggestions for a glossary, not corrections — nothing is rewritten.

Parameter	Type	Required	Notes
jobId	string	yes	The job whose cues to scan.
limit	integer	no	Maximum terms to return. Default 50.

Returns a step-by-step guide written to be read by a human, not by you. Call this when someone asks "how do I use this", "what can it do", "how do I get started", or when you need to explain the workflow rather than execute it. Pass journey="first_subtitle" for the end-to-end path, "api_key" for credential setup, "languages" for what is supported, or "limits" for what it cannot do. Answer the pers

Parameter	Type	Required	Notes
journey	string	no	Which journey to explain. first_subtitle: upload to finished file. api_key: getting a credential. languages: what it does and does not handle. limits: honest boundaries. edit_subtitles: fixing a line

How many transcription jobs are running and whether a new one can be accepted. Call this before lunara_start_subtitle_job or lunara_subtitle_from_url if you are submitting several jobs — when saturated, wait the retryAfterSeconds rather than retrying immediately.

No parameter schema on file.

Run the contract gate over both tracks: ASR text and timings byte-exact, cue length bounds, no empty cues. Returns pass/fail per rule. Free, because refusing a broken track should never cost the caller money.

Parameter	Type	Required	Notes
jobId	string	yes	The job to validate.
track	string	no	Which track to check. Default both.

Return duration, dimensions, codecs and container for a URL, plus whether the audio is mono 16 kHz already. Call this before choosing an ASR engine.

Parameter	Type	Required	Notes
mediaUrl	string	yes	URL to probe.

Report how much of the source timeline the subtitles cover, and list suspicious gaps longer than a second. Use before shipping a long video.

Parameter	Type	Required	Notes
jobId	string	yes	The job to inspect.

Per-cue reading speed, displayed length and characters-per-second, with the cues that exceed comfortable limits. Mirrors the metrics shown in the editor.

Parameter	Type	Required	Notes
jobId	string	yes	The job to measure.
track	string	no	Default primary.

Free

$0 forever

Every tool — any call limit is its publisher’s, not ours
No card, and mcprush charges nothing for it
Runs at its publisher’s own address — this marketplace is not in the path
Nothing to cap, because mcprush bills nothing

Nothing here is billable: mcprush charges nothing to install Lunara or to call it, at any volume.

ATool-list grade · score 93 / 100

Release 1.0.1 · scanned 30 Oct 2026

Graded from the tools this server lists, not from its source code.

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 score93

Findings0 threat · 1 capability

Capabilitymoderate

Coveragemanifest

Verificationnot in this report

Surface digest5c02bcdec9104d7fee03fdff94cca0da2afc82f551081a8434f640daf7a2a8bd

Score vectorthreat 100 → client 93

Term	Level	What it prices	Points
capability-exposure	moderate	capability blast radius (moderate) — client exposure if the model is manipulated	−3
coverage-honesty	manifest	inspection depth (manifest) — how much of the target the scan could see	−4

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.

The scan read each tool’s name, description, input schema and annotations exactly as the server listed them the last time it answered us; no code was reviewed. We ask the server again about once a month, and a list that changed is graded again.

Tool surfaceread

Implementation sourcenot available

Package metadatanot available

Live transportnot available

The implementation source was not read — this reflects the declared tool metadata only.

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
lunara_list_tools	no tags
lunara_get_job	no tags
lunara_get_srt	no tags
lunara_list_cues	no tags
lunara_start_subtitle_job	external-sink	external-sink: declares readOnlyHint: false, openWorldHint: true
lunara_subtitle_from_url	external-sink	external-sink: declares readOnlyHint: false, openWorldHint: true
lunara_upload_video	state-write	state-write: declares readOnlyHint: false
lunara_finish_upload	state-write	state-write: declares readOnlyHint: false
lunara_get_glossary	no tags
lunara_transcribe	external-sink	external-sink: declares readOnlyHint: false, openWorldHint: true
lunara_translate	state-write	state-write: declares readOnlyHint: false
lunara_rewrite	state-write	state-write: declares readOnlyHint: false
lunara_segment	state-write	state-write: declares readOnlyHint: false
lunara_glossary_extract	no tags
lunara_guide	no tags
lunara_get_queue	no tags
lunara_qa_contract	no tags
lunara_probe_media	untrusted-input	untrusted-input: declares openWorldHint: true (interacts with entities outside the server)
lunara_get_coverage	no tags
lunara_readability	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 digest5c02bcdec9104d7f

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.