notolens

Add an asset to a monitor: an official domain to watch for conflicting registrations, a mark text for trademark office filings, or a brand keyword for app store listings. This is confirmed work: show the user the exact type, value, config and reference choice first and pass confirm true, because the first scan is scheduled as soon as the asset exists and archiving (dashboard-only) only stops future scans, it does not undo a scan that already completed. Required config per type: domain needs config.brand, trademark needs config.offices (subset of US, EM, WO; classes optional, Nice 1-45), app needs config.stores (subset of google_play, app_store). Domain assets also need an explicit reference choice (capture, upload or skip); trademark and app accept an optional upload staged with prepare_asset_upload. A capture always goes through preview_domain: pass its capture_token so the stored baseline is exactly the render the user approved, and the response includes that screenshot as image content; when the preview rendered nothing (captured false) pass skip instead, leaving reference_stored false. site_result carries the probe outcome recorded at preview time. For domain assets, the response's match_scope reports the matching scope: 'distinctive' names are matched across TLDs while 'generic' common terms are only watched on the asset's own suffix. next_scan_at reports when the first scan is due, or null while the monitor is paused. Errors name the cause: invalid input, plan limit reached, or a duplicate already watched by the monitor.

Append an entry to a takedown's timeline. comment records progress or correspondence ('registrar replied, ticket #123'); action logs a step the user performed from the match's possible next steps and requires that step's catalog code, the same codes get_match lists under actions. Steps only log on an open takedown; comments also work on closed ones. Entries are append-only: a mistaken one is retracted with retract_takedown_event, never edited.

Create a monitor: the container that watches one brand's conflicting domain registrations, trademark filings and app listings. Name it after the brand. The monitor starts empty and dormant; add assets one at a time with add_asset and the first scan starts on its own. Plan limits apply and the error names the limit that was hit.

Get the account context for this connection: the signed-in user, the organization this token operates on, its subscription plan and limits, digest notification settings, and whether this connection holds the write scope (write_access). A good first call to confirm which account and organization the agent is connected to. A null plan means the organization has no paid subscription; send the user to billing_url to subscribe or upgrade. On a read-only connection, new_monitor_url and settings_url are the handoff links for setup and preferences.

Get the full curated detail for one match: triage status, risk assessment, the source record (DNS, site, registration and network data for domains; register record for trademarks; store listing for apps), possible next steps in actions (enforcement guides with steps and contacts; options for the user to decide on, never advice, and nothing is sent or filed by notolens), and whether a screenshot capture exists (fetch it with get_match_screenshot). reappeared marks a domain that came back after an absence.

Get a download link for the evidence PDF of one match: the same document the dashboard exports for enforcement work such as takedown requests and disputes, compiling the match details, the observed source record and captured images. It carries observed facts only: no risk score, assessment summary or possible next steps. download_url needs no sign-in and expires at expires_at (a few minutes); hand it to the user or fetch it before then, and call again for a fresh link.

Get the screenshot captured for a domain match, as image content: the top of the rendered page, the same view the risk assessment reads. Only domain matches with a reachable site have one. The full-page capture is on the match's dashboard_url.

Get the full detail for one takedown: status, which attempt it is out of the match's total, whether it can be reopened, the append-only timeline (lifecycle entries plus comments and logged steps; retracted entries stay listed, muted) and the match it tracks with its record and possible next steps. The mine flag on an event marks entries the signed-in user authored, which are the ones retract_takedown_event can mute.

List matches: conflicting domain registrations, trademark filings or app listings found by the monitors. Results are filtered to exactly one status (default inbox) and ordered by assessed risk, worst first. Combine status with min_risk/max_risk or verdict to answer questions like "watched matches with risk above 5". To triage the results, call update_matches with their ids or the same filters.

List the organization's brand monitors. Each monitor watches domains, trademark offices and app stores for conflicts with one brand. Only active assets are listed; archived ones are hidden, same as the dashboard.

List monitoring scans: each scan is one kind-scoped check of a monitor's sources (domains, trademarks or apps). Shows when monitoring last checked, what it found, and any asset failures.

List takedowns: self-guided enforcement attempts the user works against matches (notolens never submits anything itself). Filter to the open queue (the default) or the closed history; each item links back to its match, and get_takedown shows the full timeline.

Stage an image upload for an asset reference: a mark image, an app icon or a domain screenshot the user supplies. Returns a signed URL; POST the raw image bytes to it (PNG, JPEG, WebP, GIF or AVIF, up to 8 MB), then pass upload_token back verbatim as reference.upload_token on add_asset. The link is bound to this organization and expires within a few minutes.

Render a domain before adding it as an asset on a monitor with a capture reference. A preview exists only for a planned add_asset: it needs the monitor, passes the same checks add_asset would (valid domain, plan slot free, not already watched) and takes confirm once the user has asked for the add. Returns the screenshot as image content so it can be shown to the user, plus a capture_token bound to that domain with its expiry in expires_at (a few minutes). Show the image to the user and pass that token as reference.capture_token on add_asset only once they approve it; to retry, call preview_domain again. captured marks a usable capture: a favicon-only render has a token but no image, while a failed or blocked render returns the outcome with neither. Requires the mcp:write scope, since it only makes sense when setting up monitoring.

Retract a comment or logged step on a takedown. The entry stays in the append-only timeline, muted. Only the author can retract it: the mine flag on the event says whether the signed-in user authored it. Lifecycle entries (started, succeeded, unsuccessful, dropped, reopened) are never retractable.

Send a message to the notolens team: a question, product feedback, or a bug report about notolens itself (the service, the dashboard or these tools), not about a match. The message goes out as an email from the signed-in user, and replies arrive at the account email (reply_to). Send only what the user asked to send: show them the exact topic and text first and pass confirm true once they approve it. Include what the user did and what they expected, plus any dashboard_url the message is about; never include passwords, tokens or other secrets. Each user can send a few messages a day; past that the call fails with a limit error, so do not retry it. Works on read-only connections because it changes nothing in the account.

Start a self-guided takedown on a match: opens an enforcement attempt the user works through manually, using the match's possible next steps as the guide; notolens never submits or sends anything itself. Requires an open match (inbox or watching); a closed match errors. Starting moves the match to in_takedown and arms recurring record refreshes so a subject that stays up keeps being probed; it stays in_takedown until the attempt closes or the match is resolved or dismissed. Safe to repeat: an already-open takedown is returned with created=false. Record work performed with add_takedown_event, close the attempt with update_takedown, and use get_match_evidence for the PDF to attach when filing.

Get match counts by triage status and kind, plus how many are new (undecided inbox matches first found by the latest completed scan of their kind) and how many scans are active (active_scans). A good first call to understand the current state.

Change the triage status of matches. This is the triage action behind requests like "dismiss everything in the inbox below risk 5" or "move all dismissed back to inbox". Select matches in exactly one of two ways: match_ids for explicit picks (ids come from list_matches), or filter with the same fields as list_matches plus a required status naming the bucket to move out of (in_takedown selects matches with an open takedown). Filter moves apply immediately to every match currently matching, capped at 500 per call; repeat for larger sets. Call list_matches with the same filters first to review the set. Dismissing or resolving cancels any open assessment and closes any open takedown. Matches in in_takedown are skipped by moves to inbox or watching: close the takedown with update_takedown instead. in_takedown is never a target; start_takedown moves a match there. watching and in_takedown arm recurring record checks, daily for domain and app matches and weekly for trademark, and every check writes a record and queues a reassessment: use watching only for subjects whose changing state would change the decision, not for undecided matches, which sit in inbox unchecked. dismissed is permanent for trademark and app matches, which never resurface; domain matches reappear automatically if the subject resurfaces, so confirm with the user before dismissing trademark or app matches.

Pause or resume a monitor. Pausing stops scans and record refreshes but keeps the monitor and its matches; resuming restarts scanning within plan limits. Deleting a monitor is dashboard-only because it removes its matches too.

Change a takedown's status. Close an open attempt with succeeded (the subject is down; resolves the match), unsuccessful (tried and failed) or dropped (abandoned); both non-success closes return the match to inbox for fresh triage and stop its recurring checks. Pass open to reopen the latest closed attempt, moving the match back to in_takedown and re-arming its check; reopening fails when a newer attempt exists or the subject reappeared after it closed; start a fresh attempt with start_takedown then.