Connecting an AI app

You can connect an AI app, such as Claude, to your FilmScorecard account and work through your festivals with it. It can read almost everything your own screens show you: a season's submissions and any one of them in full, the season overview, the judges with their invitations and requests, filed scorecards and the problems judges report, awards, announcements, judge pay and scorecard sales, imports and season settings, and, across your organization, the judges directory, your own Inbox, orders, promo codes, the library, organization settings and the audit log. It can change three things: move films between In Consideration, Selected and Not Selected, set your own flag, and mark films Priority. It acts as you, and it reaches exactly what you reach on your own screens.

It is for the people who run festivals. It works only in an organization FilmScorecard has switched AI access on for, and it is not offered to judges or filmmakers at all.

Before you start

Your organization needs AI access switched on. FilmScorecard switches it on one organization at a time, and every organization starts with it off. There is no setting for it on your own screens. If you want it, write to us through the contact form.
You can tell whether it is on from your own account screen: on the Connections tab, each app's row opens to the steps to connect it, under How to connect, only when your organization has AI access. When it does not, the rows do not open, and the head of the AI apps card says so and offers Request AI access, which opens our contact form with the request already written. See Connections.
You need an AI app that can connect to a remote server with a sign-in. Claude, ChatGPT, Claude Code and Cursor each have a page below. AI apps call this an MCP server, a connector or a custom connector.

Connecting

The Connections tab on your account screen walks you through this for each app, with the address and any command ready to copy. See Connections.

Each app has a page of its own, with every step from start to finish, what to try first, and what to do if something goes wrong:

Connecting Claude: claude.ai and the Claude desktop app.
Connecting ChatGPT: on the web, in developer mode.
Connecting Claude Code: one command in a terminal.
Connecting Cursor: a few lines in Cursor's mcp.json.
Connecting another AI app: any app that can use a remote MCP server with a browser sign-in, and how to tell whether yours can.

Connecting is the same three moves in every app. You give the app FilmScorecard's address, the app sends you to FilmScorecard in your browser, and you press Allow there. There is no key, token or password to copy anywhere, and the app never sees your sign-in code: FilmScorecard gives it a pass of its own, which you can take back at any time from the Connections tab.

Every app needs the same address:

https://www.filmscorecard.com/api/mcp

Then, in your browser

If you are not signed in to FilmScorecard, you get the log-in screen first. Sign in with the emailed code as usual (see Logging in), and you come straight back.
The connection page asks whether to let the app in. It names the app, the address it will act as, and what it can do. Check the name is the app you just added. If you did not just start connecting an app, press Deny. See The connection page.
Press Allow. Your browser goes back to the app, which is now connected, and the app reads Connected on the Connections tab.

The connection page

The page has the log-in screen's plain frame: the FilmScorecard wordmark, the question, and the Help · Terms · Privacy · Contact links at the foot. There is no navigation on it. It shows:

On the page	What it is
The heading	The app's name, as the app gave it when it asked, with "wants to connect to your account." under it.
Website	The app's web address. Left out when the app did not give one.
Will act as	The address you sign in with. Everything the app does, it does as you.
It can	"read and update festival data you already have access to"

Then two buttons:

Allow connects the app and sends your browser back to it. It reads "Allowing…" while it works.
Deny sends your browser back to the app with a refusal. Nothing is connected. It reads "Denying…" while it works.

If you have already allowed that app and not revoked it since, the page does not ask again. Your browser goes straight back to the app.

What the page can say instead

The page says	What it means
"This connection request is incomplete. Start again from your AI app."	The address is missing the part the app adds to it. Most often the page was opened from a bookmark or typed by hand. Start the connection again from the app.
"This connection request has expired or was already answered. Start again from your AI app."	A connection request does not last long, and it can be answered only once. Start again from the app.
"AI access isn't enabled for your organization.", with Deny and no Allow	None of the organizations you belong to has AI access switched on, or you are signed in on an account that belongs to no organization at all, such as a judge's. Press Deny so the app stops waiting for an answer. The same sentence appears if AI access was switched off between the page opening and your pressing Allow.
"That didn't go through. Try again."	Your answer did not reach FilmScorecard's sign-in service. The buttons stay where they are, so press again. If it keeps happening, start again from the app.
"Not available in a support session."	Only FilmScorecard's own platform administrator ever sees this, while acting as somebody. See Support sessions.

What a connected app can do

Thirty-four things: thirty-one that read and three that change something. They are listed in The 34 tools, grouped by the screen each one reads. Your app shows them by the names in those tables when it asks your permission or tells you what it did.

The three that change something are marked as doing so, and the other thirty-one are marked as only reading. AI apps can use that mark to ask you before they change anything.

It reads what your own screens show you, and changes only what you could change there. An organization, festival or season you cannot open is invisible to it, exactly as it is to you. A tool whose screen needs a permission needs the same permission here, and a section your screen leaves out for your role is left out of the app's answer too: see Answers that leave a section out. A change it makes is the same change the Submissions screen makes, and your colleagues see it on their own screens when their list next refreshes. See The list keeps itself up to date.

Every app is offered all 34, whatever your role. A tool your role does not reach is refused when the app calls it, with one sentence naming what is missing. See When the app is refused.

Season tools and organization tools

Most tools read one season, and name it by its organization, festival and season. Nine read your organization as a whole, across all its festivals, and name only the organization: the ones under Across your organization. List my festivals names nothing, and is where the app learns the rest.

Lists come a page at a time

Every list comes 25 rows at a time, or up to 50 if the app asks, in the order its screen opens in, and the app pages through the rest. Each answer says where it is, for example "Showing 25 of 36 judges. Pass nextCursor to read the next page."

An empty list is an answer, not a refusal. Most say so in words, for example "There are no judges to show."; List submissions says "Showing 0 of 0 submissions."
What the app narrows by is applied before the count, so the total is what matched, not the whole season.
A list the app does not narrow opens the way its screen opens, which is not always everything. Each area below says where it differs.
A single record comes whole, up to its screen's own limits: the latest 500 messages of a conversation, the first 1,000 recipients of an announcement, the latest 50 saves in a submission's history.

The 34 tools

In the Who can use it column, "anyone who can open the season" means every festival role, Viewers included, as well as organization owners and org admins. Where a permission is named, it is the tick box of that name on Festival access, and it is the same one the screen asks for.

Your festivals

In your app	What it does	Who can use it
List my festivals (`list_my_festivals`)	Lists your organizations that have AI access, your role in each, and their festivals and seasons.	Anyone connected

List my festivals is where an app starts. Everything else names an organization, and most tools a festival and a season as well, and this is where the app learns them. It returns each of your organizations that has AI access, your organization role there, the festivals in it you can open, their seasons, and the season each festival opens on. Archived seasons are left out.

The one-line answer reads, for example, "1 organization with AI access, 3 festivals in all." When none of your organizations has AI access, the list is empty and the app is told "AI access isn't enabled for any organization you belong to."

A season's submissions

In your app	What it does	Who can use it
List submissions (`list_submissions`)	Reads one season's submissions, as the Submissions screen lists them, with the Submissions export's columns.	Anyone who can open the season
Get submission (`get_submission`)	Reads one submission in full, as its detail panel shows it.	Anyone who can open the season
Set selection status (`set_selection_status`)	Changes something. Moves films to In Consideration, Selected or Not Selected. Previews first.	Anyone who can change the season
Set my flag (`set_flag`)	Changes something. Sets or clears your own flag on one film.	Anyone who can open the season
Set priority (`set_priority`)	Changes something. Marks one film Judge this first, or clears the mark.	Anyone who can change the season

List submissions reads the same films the Submissions screen lists, with withdrawn and disqualified ones left out, in the order the screen opens in: highest Overall first, with the title breaking ties.

The app can narrow the list by any of these, together:

Status: In Consideration, Selected or Not Selected, as many as it likes.
Category: one category.
Flag: the films you have flagged, or the ones you have not.
Priority: the films marked Judge this first, or the rest.
Type: films or screenplays.
Search: the title or the submitter's name, the same rule as the screen's search box. A tracking number does not match.

For each film the app is given its title and tracking number, whether it is a film or a screenplay, the submitter's name and email address, its categories, its selection status, whether you have flagged it, whether it is marked Priority, how far its scoring has got (all scores in, partly scored, or not started), its score and the colour the score is shown in, how many approved scorecards it has, the record's own status word, which of the tagline, synopsis, poster and statement its record is missing, the date it was submitted, and its running time or page count.

These are the Submissions export's columns, so the app can answer anything the file would. See The two CSV exports.

The score is what approved work has banked so far: a whole number out of 100, the same figure as the list's Overall column and the export's Score, and no figure at all until approved work counts. It is not a forecast, so a film waiting on two of its three judges reads low. See Overall is points banked, not a forecast.

One submission in full

Get submission reads one submission as its detail panel shows it:

the same row List submissions gives;
the whole record: tagline, synopsis and statement, the film's details and credits, the submitter's name, email address, phone, city and region, and the trailer, website and Instagram links;
each category it competes in, with the scorecards that category applies and how its 100 points are divided;
every filed scorecard the panel lists, those still waiting for approval included and marked as waiting, with the judge's name, the points and notes on each question, and the written feedback;
the press kit items, the questions the filmmaker asked the judges, and where the scorecard release stands;
whether a screener link and a poster are on file (never the link or the image);
the record's edit history: the latest 50 saves, each on its own, with who saved and each field's before and after. Only the saves: the rest of the Audit tab — judging, status changes, emails, the filmmaker's visits and actions — is not in the answer.

Someone who holds View revenue and payouts also gets the Scorecard access card: whether the filmmaker has unlocked their scorecard, the receipt number, when, how many of their categories are unlocked, and whether it was refunded. Anyone else gets everything else. See Answers that leave a section out.

A submission that is not on the season's list, withdrawn and disqualified ones included, is answered "No submission with that id in this season."

Moving films between statuses

Set selection status moves one film or many, up to 500 at once, and it always previews first.

The app names the films and the status. Nothing changes. FilmScorecard answers with each film that would move, its status now and after, and a line such as:

Preview only, nothing was changed: 3 films would move to Selected. Call again with confirm set to true to apply it.

Any film that is not on that season's list is counted instead: "1 id is not in this season's list."
The app asks again, this time confirming. Now the films move, and the answer counts only the films that actually changed:

Moved 3 films to Selected.

or, when some of them did not:

Moved 2 of 3 films to Selected. The others were already Selected, are not in this season's list, or you cannot change them.

FilmScorecard asks the app to show you the preview before it confirms, but it cannot make the app wait for you. Read what your app says it is about to do before you approve it.

If you can only view the season, the preview is refused as well as the move. The app is told "You can view this season but not change it.", and nothing is previewed or changed.

It is the same move as the Submissions screen's status items and bulk bar. See Selecting rows and changing status in bulk.

Your flag

Set my flag sets or clears your own flag on one film, straight away, with no preview. It is the flag in the Submissions screen's Flag column, so it is private to you like any other flag. See Flags are private. It needs only access to the season, so a Viewer can set their own flag too.

The app is told "Flagged the film for you." or "Cleared your flag on the film."

Priority

Set priority marks one film Judge this first, or clears the mark, straight away, with no preview. It is the same mark as the Judge this first switch on a submission's Settings tab: your whole team sees it, and every judge who can take the film is offered it before the others. See Priority: judge this first.

The app is told "Marked the film as priority." or "Removed priority from the film." Marking a film that is already marked changes nothing, and still answers that it is marked. If you can only view the season, it is told "You can view this season but not change it." and nothing changes, just as the Priority card is not drawn for a Viewer.

The season overview

In your app	What it does	Who can use it
Get season overview (`get_season_overview`)	Reads the season's home page and the counts its navigation shows.	Anyone who can open the season

It holds what the Season overview draws for that season:

the season's status, time zone and year, its judges per category and its judge compensation;
once judging has been opened, the four judging cards: what needs attention (unresolved screener problems and how long the oldest has waited, flagged scorecards, scorecards awaiting approval and the submissions they sit on), judging progress, judge completion, and submissions by category and by status;
the Revenue card's figures, for someone who holds View revenue and payouts: the gross, the net, the processing figure, and one fee that is the card's Platform fee and Processing added together — everything the sales gave up — so the gross is always that fee plus the net;
the import history: every batch with its counts;
the counts the navigation carries: awards still waiting for a decision, scorecards pending in the Review Queue, categories nobody can judge, and your own unread Inbox messages;
while judging has not been opened, what is left before it can be: which of the three requirements are met (submissions imported, judging dates set, payouts connected), the dates, how many submissions and categories the season has, the judge invitations sent and accepted, and any import in progress. This is there whether or not you dismissed the Finish setting up strip on the page.

Judges

In your app	What it does	Who can use it
List judges (`list_judges`)	Reads the season's Roster: each judge's roles, status, whether they have asked to leave, pay mode, what they may be assigned, any film of theirs in the season, when they were invited and accepted, when they last filed, scorecards filed and awaiting approval, and FilmFreeway. Also the season's categories with what each pays per film.	Anyone who can open the season
Get judge (`get_judge`)	Reads one judge's Roster panel: recent activity, the categories their scope names, every scorecard they have filed with its score and review state, their own per-category rates, and their directory profile if they are listed.	Anyone who can open the season
List judge invitations (`list_judge_invites`)	Reads Invites sent: who was invited and by whom, the rates offered, the roles and scope, when it was sent and what came back. Also the season's call budget.	Manage judges and assignments
List judge requests (`list_judge_requests`)	Reads Requests: each applicant, the call, the role, the posted, asked and countered rate per category, any counter and the judge's answer, their note, pace and away dates, and the categories they applied for. Also the call budget.	Manage judges and assignments

Each list opens as its tab does. List judges gives the judges on the Roster (active and paused), List judge invitations the pending ones, and List judge requests the open ones (new, shortlisted and countered). The app can ask for the others: invited, declined, removed and revoked judges, or archived ones (on an earlier season of this festival and not this one); accepted, declined or revoked invitations; accepted or declined requests.
List judges can search a judge's name, email address or JDG- number. The other two cannot.
Rates are not behind View revenue and payouts here. The Roster shows each category's per-film pay to everyone who can open the season, and List judges and Get judge carry the same figures. See Judge pay.

Scorecards and the problems judges report

In your app	What it does	Who can use it
List scorecards (`list_scorecards`)	Reads the Review Queue's filed scorecards: the film, the judge, the kind of scorecard and its categories, the score, the review state, flag notes and return reasons, and the written feedback. Also the counts on the queue's tabs.	Anyone who can open the season
Get scorecard (`get_scorecard`)	Reads one filed scorecard as its panel shows it: the questions with their guidance and options, the judge's answers or marks, the filmmaker's own questions and the judge's answers to them, the written feedback, award recommendations, the judge's original answers wherever a coordinator has edited them, and who approved it and when.	Anyone who can open the season
List issues (`list_issues`)	Reads the Issue tab: the films judges could not play or reported another problem with, what the judge wrote, when and how long it has waited, where it stands, any resolution note, and whether the film can still be sent back to that judge. Also the counts by state.	Anyone who can open the season
Get issue (`get_issue`)	Reads one issue, whether the film has a screener link on file, and who settled it.	Anyone who can open the season

List scorecards gives every review state unless asked otherwise, oldest filed first, the same order as the Review Queue. The screen opens on Pending; the app does not. It can narrow to pending (awaiting review or flagged), returned or approved, and by kind of scorecard, by judge, by category, to one submission, or by a search of the title, the judge's name or the tracking number.
List issues gives the open ones unless asked otherwise, newest reported first, as the tab does. It can ask for returned, dismissed or all, and search the same way.
Reading the queue needs no permission, as on the screen. Approving and settling do, and the app can do neither. See Review Queue and Screener problems your judges raise.

Awards

In your app	What it does	Who can use it
List awards (`list_awards`)	Reads the Awards screen: the category awards and your library's awards, each with its `AWD-` code, whether its ranks are released, how many entries are still being scored — not yet fully judged — whether it is waiting or completed, who decided and when, and its whole pool: every entrant's categories, how it got in, its result, any honorable mention, judges' recommendations and score.	Anyone who can open the season
Get award (`get_award`)	Reads one award's pool page: the award and its entrants, best score first, how many library categories are eligible for it, and the submissions that could still be added.	Anyone who can open the season

Category awards come first, then the others, each A to Z, as the two tabs. As on the screen, a category award is listed only when its category has entries. The app can ask for one kind, and search an award's name, its AWD- code or a current winner's title.
A category award's pool is its category's entries, so Get award gives it no eligible categories and no submissions to add. See Award pools.

Announcements

In your app	What it does	Who can use it
List announcements (`list_announcements`)	Reads the Announcements list: the `ANN-` code, subject, audience, sender, how many judges it reached, draft or sent, whether it is archived, and when it was sent and last updated.	Manage judges and assignments
Get announcement (`get_announcement`)	Reads one announcement in full and, once it is sent, who it reached and when each of them read it, with the read and unread counts.	Manage judges and assignments

List announcements gives drafts and sent ones, newest update first, as the screen does. Archived ones come only when the app asks for them. It can search the subject or the ANN- code. Past 1,000 announcements the answer says the list stopped there, as the screen's footer does.
A draft has no recipients yet, so Get announcement gives it none. A sent one lists up to 1,000 recipients by name; its counts always cover everybody. See Who has read an announcement.

Judge pay and scorecard sales

In your app	What it does	Who can use it
List judge payouts (`list_judge_payouts`)	Reads Payouts, Ledger: for each judge, approved pay, approved and unpriced scorecards, what has been paid, what is outstanding, and their pay state. Also the season's figures, and the filed scorecards still awaiting approval with what approving them would add.	View revenue and payouts
Get judge payout (`get_judge_payout`)	Reads one judge's pay panel: their totals, every approved scorecard with the rate stamped on it, those scorecards grouped by category and rate, and every payment logged to them, voided ones included, with what it covers and whether a receipt is on record.	View revenue and payouts
List judge payments (`list_judge_payments`)	Reads Payouts, Payments: each payment's judge, amount, date paid, method, reference, what it covers, who logged it, whether a receipt is on record, and who voided it and why. Also the count and total of the payments listed.	View revenue and payouts
List scorecard sales (`list_scorecard_sales`)	Reads the scorecard-sales export row by row: when, the `SC-` receipt number, the submission, what was paid, the fee, the net, the currency, whether it was a sale, a comp or a free unlock, and any promo discount. Refunds are left out, as they are from the file.	View revenue and payouts

The figures are the screen's own, to the cent. Amounts reach the app as whole numbers of cents, named as cents, so it has nothing to round. A negative outstanding figure is a credit, and a voided payment is counted and never added, exactly as on Paying your judges and The payments log.
List scorecard sales gives one fee per sale, not two. The export splits what was taken from a sale into Platform fee and Processing; the app is given the two together as a single fee — the whole amount FilmScorecard kept — and the net after it. So its fee matches the file's Platform fee only where your organization pays no share of processing, and its net always matches the file's Net. See The two CSV exports.
The ledger opens largest outstanding first, lists every judge who accepted, removed and revoked ones included, and can be narrowed by pay state or a search of the judge's name or email address. The payments open newest paid first, and can be narrowed by method, by active or voided, or a search of the judge's name or the reference.
A volunteer season has no ledger. List judge payouts answers "This season's judges volunteer, so there is no pay ledger.", with no judges and every figure at zero, and Get judge payout is refused with the same sentence. List judge payments still lists any payments that were logged.
Get judge payout answers only for a judge on the ledger. Anyone else is "No judge with that id in this season."

Imports and season settings

In your app	What it does	Who can use it
List imports (`list_imports`)	Reads the Imports screen: each finished, failed or stopped import with its file name, dates, who ran it, rows read, films imported and duplicates skipped, whether it was rolled back and any error; and the import running now, or failed and waiting for a retry, with its progress.	Anyone who can open the season
Get season settings (`get_season_settings`)	Reads the season's Settings screen: the festival's identity, the season's dates, the judging window and rules, the section weights, the scorecard price and release rule, submitter invitations, the attached categories with their per-film judge rates, the attached awards, the open judge call, and how far scorecards and invitations have got.	Edit festival settings and team, and a role that can change the season

Imports come newest first, a page at a time. The file itself never leaves; see What the app is never given.
Get season settings asks what the Settings screen asks: the permission, and a role that can change the season. A Viewer is refused even with the permission ticked, because the Settings screen refuses them too. See Festival settings.

Across your organization

These nine name only the organization, and cover every festival in it.

In your app	What it does	Who can use it
List the judges directory (`list_judge_directory`)	Reads the Judges directory as your organization sees it: each listed judge's public profile, whether they have judged for you, their newest invitation or place with you, and your own private note on them.	Anyone in the organization
List inbox conversations (`list_inbox_threads`)	Reads your own Inbox: each conversation's judge, festival, whether you archived it, the newest message, and how many of the judge's messages you have not read. Also your unread total.	Anyone in the organization, for their own conversations
Get inbox conversation (`get_inbox_thread`)	Reads one of your own conversations: its latest 500 messages, oldest first, whether the judge has blocked you, whether you can reply, and how many you have not read.	Anyone in the organization, for their own conversations
List orders (`list_orders`)	Reads Orders: every scorecard unlock across your festivals, newest first, with the buyer, submission and category, festival and season, paid, refunded, comp or free, the amount, any promo code and discount, and — where one applies — the fee, the part of it that is your processing share, and the net. Also the screen's figures over every order that matches.	Owner or Org admin
List promo codes (`list_promo_codes`)	Reads Promo codes: each code's discount, the festivals it covers, its limits and dates, its status, its completed and held uses, and its five latest orders.	Owner or Org admin
Get library (`get_library`)	Reads your library: categories (type, length limits, default rates, the scorecard in each slot, how many festivals run it), scorecards (each question's label, answer type and points, how many categories use it) and awards (eligible categories and use), archived ones included and marked.	Anyone in the organization
Get scorecard template (`get_scorecard_template`)	Reads one scorecard from your library as the builder shows it: every question with its answer type or points, guidance, options, and whether it appears on the scorecard a submitter buys, and which feedback questions it asks.	Anyone in the organization
Get organization settings (`get_org_settings`)	Reads Organization settings: the profile, everyone in the organization with their role, status, Last active date and festival access, pending invitations, and the payout account with your fee terms — the platform fee rate, whether it is charged on what each order pays or on the full price, and your share of processing.	Owner or Org admin
List audit log (`list_audit_log`)	Reads Settings, Audit log: every access change, newest first, with when, who, the change in a sentence, and the festival it touched or "Organization".	Owner or Org admin

The Inbox tools read only your own conversations. A colleague's conversation with the same judge is not yours, and is answered "No conversation with that judge in this organization." Reading never marks anything read; see Reading changes nothing. Whether you can reply follows the strips described in When you cannot write.
The directory opens with the judges who have scored the most films first. The app can narrow it to judges who have or have not judged for you before, who are or are not available now, by country or by genre, or search a name, location or genre. It carries no email address, as the directory itself does not, and no photo. See Judges directory.
Orders can be narrowed by festival, season, status and dates, and searched by the buyer's name or email address, the title, the category or the SC- receipt number. A date the app names is a whole day in UTC. The screen's own date choices follow your browser's calendar, so near midnight the two can disagree about which day an order belongs to. See Orders.
An order's fee is the whole amount FilmScorecard kept, the platform fee and your processing share together, with the share given beside it. The Orders screen shows them as two columns, Fee and Processing, so the screen's Fee is the app's fee less the share. The figures over every matching order give one fee total, the two together.
Promo code statuses are worked out on your organization's calendar, as the screen does. See Promo codes.
The library is your organization's templates, not a season's copy. Editing a library category or scorecard changes what Get library and Get scorecard template say, and never what a season tool says, because a season keeps its own copy. Get scorecard template reads a template; a scorecard a judge filed is Get scorecard. See Categories and scorecards.
The payout account is as FilmScorecard last recorded it. Get organization settings never asks Stripe, so the Payments tab, which does, can be a step ahead. The fee terms are the ones that apply to new orders; an order already taken keeps the fee it was charged. See Connecting payouts.

What the app is never given

Screener links and passwords. Get submission and Get issue say whether a screener link is on file. A press kit item that checks the screener says whether it is satisfied, never what it holds.
Images. A submission says whether it has a poster, and an award's winner whether its card shows one. A payment says whether a receipt is on record. Your festival's settings say whether a logo is uploaded. Nobody's photo is ever included.
Import files. An import's counts, dates, file name, who ran it and its error line, never the file's rows, its columns, how they were matched, or a correction made while checking it.
Any file at all. No CSV export, pay statement, receipt, import error report or poster reaches the app. What the exports hold is in its answers instead: the submissions export's columns in List submissions, the scorecard-sales file in List scorecard sales, the orders in List orders, the audit log in List audit log, and an announcement's read receipts in Get announcement.

Reading changes nothing

The only things a connected app changes are the three above. Reading, by any of the other thirty-one, leaves everything as it was:

A conversation the app reads stays unread. Its count, the amber number on your Inbox row and the unread figure in Get season overview are exactly what they were. They clear when you open the conversation on the Inbox screen yourself. See Messaging your judges.
It is not a visit. Reading Organization settings through the app does not move your Last active date on the People tab.
Nothing is emailed, sent to a judge or queued.

FilmScorecard does keep a record of the call itself; see Every call is recorded.

Answers that leave a section out

When your screen leaves a section out for your role, the app's answer leaves it out too. The rest of the answer comes back as usual, and nothing in it says a section is missing, just as nothing on your screen does.

Tool	Left out	Unless
Get season overview	The Revenue card's figures	You hold View revenue and payouts on the season.
Get submission	The Scorecard access card	You hold View revenue and payouts on the season.
Get library	FilmScorecard's default list of categories and awards, with what your organization already holds of it	You are the organization's owner.

Three more sections are missing because the season or the record has none: the judging cards before judging opens, the setup once it has, and the recipients of an announcement that is still a draft.

When the app is refused

Every refusal reaches the app as one sentence, which it usually passes on to you:

The app is told	What it means
"That festival or season isn't available."	The organization, festival or season it named does not exist or is not yours to open, or the film it named for a flag or a priority change is not on that season's list. FilmScorecard gives the same answer to all of these, so nothing outside your reach can be discovered by asking.
"That organization isn't available."	The same, for the nine tools that name only an organization: it does not exist, is not yours, or is suspended.
"AI access isn't enabled for this organization."	The organization it named has AI access switched off. See When AI access is switched off.
"That needs "View revenue and payouts" in this season, and your role doesn't include it."	One of the four judge pay and scorecard sales tools, without that permission on the season.
"That needs "Manage judges and assignments" in this season, and your role doesn't include it."	List judge invitations, List judge requests or either announcement tool, without that permission on the season.
"That needs "Edit festival settings and team" in this season, and your role doesn't include it."	Get season settings, without that permission or without a role that can change the season.
"That needs the Owner or Org admin role in this organization."	List orders, List promo codes, Get organization settings or List audit log, from anyone else: Finance and Member are refused.
"You can view this season but not change it."	Set priority, or Set selection status (the preview as well as the move), from somebody whose access to the season is view-only. Nothing is previewed or changed. Set my flag is not refused.
"This season's judges volunteer, so there is no pay ledger."	Get judge payout on a volunteer season. List judge payouts gives the same sentence as its answer, with no judges.
"No submission with that id in this season.", "No judge with that id in this season.", "No scorecard with that id in this season.", "No issue with that id in this season.", "No award with that id in this season.", "No announcement with that id in this season."	The record it named is not in that season: it does not exist, belongs somewhere else, or has gone (a withdrawn submission, a voided scorecard). The same answer covers all of these.
"No scorecard template with that id in this organization."	The same, for Get scorecard template.
"No conversation with that judge in this organization."	You have no conversation with that judge in this organization. A colleague's conversation with them gets the same answer.
"Too many calls in the last minute. Wait a moment and try again."	You have reached the limit below. Nothing ran.
"That didn't save. Try again."	A flag or priority change did not save. Nothing changed.
"The change could not be saved. Try again."	A status move did not save. Nothing changed.
"Something went wrong. Try again."	Something failed on FilmScorecard's side. A read changed nothing, so asking again is safe. If the app was making a change, look at the Submissions screen before trying again.

A request the app gets wrong, such as a missing season, a status that does not exist, or an id that is not in the right form, is turned away before anything runs, and the app is told which part was wrong.

120 calls a minute

Each person can make 120 calls in any one minute, across every app they have connected. The next is refused with "Too many calls in the last minute. Wait a moment and try again.", and nothing runs. The calls the limit turns away do not count toward it, so waiting is all it takes.

A preview and its confirm are two calls. Each page of a list is one, and so is each record read. A question about a large season can take many of them.

Every call is recorded

FilmScorecard keeps a record of every call your apps make: which app, which tool it used, the organization it reached and, for a season tool, the season, what it asked for, when, how long it took, and how it ended. Refused calls are recorded, and so are calls the limit turns away. A request turned away for being wrongly made, as above, is not.

No screen shows this record yet.

Disconnecting an app

Open Your profile in the menu behind your name, and choose the Connections tab.
Find the app's row on the AI apps card, and press the amber Disconnect beside the connection.
The dialog asks "Disconnect the app?" and says "Its next request is refused. To use it again you connect it again from the app." Press Disconnect.

The app is refused from its next call. To use it again, connect it again from the app, which brings you back to the connection page. The whole tab is described in Connections.

When AI access is switched off

FilmScorecard can switch an organization's AI access off at any time. From then on:

Every call naming that organization is refused with "AI access isn't enabled for this organization.", starting with the app's very next call, even in the middle of a conversation. That includes the nine tools that name only the organization.
List my festivals leaves the organization out.
The Connections tab asks you to request AI access on your account screen in that organization: its rows still name the apps, but open no steps and show no connections.
Your connection is not removed. The app still reaches any other organization of yours that has AI access, and if this one is switched back on, the app works there again without your connecting it again.

Nobody is emailed either way.

What an app cannot do

Any other change. It cannot withdraw, move or edit a submission or its categories, approve or return a scorecard, settle a reported problem, decide an award, log or void a payment, send an announcement, write to or invite a judge, or change a setting. The three changes above are all it makes.
Hand you a file. See What the app is never given.
Read your own account screen: your profile, the browsers you are signed in on, your notification choices or your connected apps.
Read anything a judge sees. Every tool reads a staff screen. No judge's queue, scoring screen, inbox or earnings is reachable.
Read the screens that exist only to make something: filling in a scorecard yourself, the import steps, the new season and new festival forms, or the seasons a submission could be moved to.
Know in advance which tools your role reaches. It is offered all 34 and finds out by calling one.
Connect for reading only. Every connected app can do all 34 things. There is no way to allow the reading and refuse the changes.
Show you what it has done. Every call is recorded, and no screen shows the record yet.
Connect as a judge or a filmmaker. Only the people who run an organization's festivals can connect an app.

Using an AI app: what to ask a connected app, how to get good answers, and example questions for each part of your work.
Your account settings: the Connections tab, and disconnecting an app.
Festival access: the roles and the permissions a tool asks for.
Submissions and Opening a submission: the list and the detail panel the submission tools read, and what each status, flag and Priority mark means there.
The Season overview: the page Get season overview reads, and the two CSV exports whose columns the app is given.
Judges, Invites sent and Judge requests: the three tabs the judge tools read.
Review Queue: the scorecards and issues the app reads.
Awards and Announcements.
Paying your judges and The payments log: the Payouts screens the pay tools read.
Imports and Festival settings.
Messaging your judges: the Inbox the app reads without marking anything read.
Orders, Promo codes and Organization settings: the owner and org admin screens.
Logging in: the emailed code you sign in with on the way to the connection page.
Contacting us: asking for AI access for your organization.