AI: Article Series planner (folder of documents) #13

Activity

1. Adron

   added

   parityWeb/API feature-parity workWeb/API feature-parity work

   P1Table stakesTable stakes

   area:aiArea: aiArea: ai

   on Sep 15, 2026
2. Adron

   added a parent issue on Sep 15, 2026

   • Epic: AI Writing Assistance & Generation (entire pillar missing) #8
3. Adron commented on Sep 16, 2026

   

   Adron

   on Sep 16, 2026

   MemberAuthor

   More actions

   Live-verified corrections from the service layer (#9 / PR #137) — read before building this

   Five things contradict what the AI epic and these issues assume.

   1. quota is not the same object on both endpoints.
   GET /api/ai/status → {usedToday, dailyLimit, remaining}. POST /api/ai/suggest → {usedToday, dailyLimit} with no remaining (verified twice). So AiQuota.Remaining is int?, and a post-suggest quota chip cannot read .Remaining — bind RemainingOrComputed instead. #10's "remaining quota is visible wherever an AI action is offered" has to account for that.

   2. There is no provider picker to build.
   AI is Anthropic-only, app-wide; per-user provider keys were removed server-side 2026-09-05. The provider field on /generate is audit-ledger-only and does not select anything, and the openai/gemini entries in defaultModels are dormant and unreachable. Don't surface a provider choice.

   3. crossPost has six fields, not two — crossPostToBluesky, selectedMastodonIds, crossPostToLinkedIn, selectedLinkedInTargets, crossPostToTwitter, linkedInLinkAsFirstComment. Both crossPost and scheduleImmediately are message_series-only (relevant to #12).

   4. A sixth 422 code exists that the epic didn't list: refused. Treat it like invalid_ai_output. Modelled and marked retryable.

   5. Quota accounting is asymmetric, and it shapes the UI.
   Input-validation 422s cost nothing (usedToday stayed 0 across two), but any call that reaches the model costs a unit — including failures. So client-side pre-flight is load-bearing, not politeness: enforce the word caps, the 10-word series gate (#12/#13), a missing powered_document source ref (#15), and the non-persistable-artifact rule (#11) before calling. And no auto-retry — a retry burns another unit.

   Also: context.spacingMinutes is accepted but ignored (series rows always land 4 minutes apart), so it's deliberately not exposed.

   Note on the exception type

   The code is surfaced on a new AiApiException, not on InterlinedApiException — the latter is sealed and was being rewritten in #130. Folding them together is flagged as a follow-up in the code. Catch AiApiException in these UI issues.

   Unverified

   POST /api/ai/generate was never called (it persists content to a shared test account). Its envelope is contract-transcribed, flagged unverified in doc comments, and keeps the raw created element for read-after-write. Whichever of #12–#15 lands first should verify it against a real account and tighten the model.
4. Adron commented on Sep 16, 2026

   

   Adron

   on Sep 16, 2026

   MemberAuthor

   More actions

   ⚠️ Correction to the quota rule I posted earlier on this issue

   My earlier comment said: "input-validation 422s cost nothing … but any call that reaches the model costs a unit." The first half is too optimistic, and #10/#14/#15's probing proved it (PRs #173/#174/#177).

   An unrecognized context.mode is not cheap-rejected — it reaches the model and bills a unit, returning 422 invalid_ai_output.

   So the accurate rule is narrower: a 422 is free only for validation the server actually performs. Anything it doesn't recognise as invalid gets forwarded to the model and charged. Observed on the real quota counter:

   call	cost
   powered_template /suggest	1
   powered_document /suggest (mode: from_article)	1
   mode: "not_a_mode"	1 — unexpectedly billed
   4 reference/scheme rejection probes	0

   The rejections that were free came back with prose specific enough to show verbatim: "List not found.", "A list must be selected.", "Document not found.", "Only http(s) URLs are supported."

   Practical consequence for this issue: client-side pre-flight must validate against a closed set of known-good values, not merely catch obvious omissions. Sending an unknown enum value is not a free mistake. And still no auto-retry.

   Also useful here

   The powered_template artifact shape is now pinned (it was contract-transcribed only):

   kind: "list"
   { title, description,
     dsl: { name, description,
            fields: [{ key, type, label, required, displayOrder,
                       options?, defaultValue?,
                       visibility?{ condition{ field, operator, value } } }] },
     rows: [ { <fieldKey>: value } ] }     // flat maps; null is a legal cell value
   

   Four members were on record nowhere in the repo: the dsl wrapper's name/description, and per-field displayOrder/options/defaultValue/visibility. The model attaches rules the editor doesn't surface (e.g. visibility.condition), which makes preserving unparsed members across an edit load-bearing — a naive re-serialize would silently drop them.

   powered_document returns {kind, title, markdown, outline[], isPublic} and matched the transcribed model exactly.

   Still unverified for all AI issues

   POST /api/ai/generate has never been called (it persists to a shared account), so {listId}/{documentId} field names are unconfirmed — every confirm path does read-after-write plus an explicit "don't retry, you'd duplicate" when no id comes back. Also unobserved: subscriber:false, providers:[], either 429, and a real research_url fetch.