Runs the coordinated specialist passes (continuity, voice, craft, anti-slop) against pasted chapter text and returns a prioritized findings report. The anti-slop + craft passes work with no API keys configured; continuity + voice passes need a project selected.

Diffs a chapter against the project's entity database and prior chapter summaries, flagging evidence-chained contradictions (character, timeline, setting, plot).

Each major character reviews their own dialogue in a chapter, flagging off-voice, anachronistic-knowledge, or off-motivation lines with in-voice rewrites.

Runs a tournament of your candidates against a demographically varied synthetic reader panel and ranks them by win rate. Works with any active AI provider, including free tiers (Ollama/Gemini) — confidence may be lower on smaller models.

Runs a GEPA-style evolve loop: scores your passage, reflects on its weakest points, revises, and keeps only rounds that beat the running best. Works with any active AI provider, including free tiers (Ollama/Gemini) — the judge may be weaker on smaller models.

Runs the quality tools (revision, contradiction, character critique) against a chapter and learns durable lessons from recurring patterns. Learned lessons feed forward into future writing via the project's lesson store.

Lists sites registered in the site registry (GET /api/sites) and probes which deploy tools are installed on this machine (GET /api/site-deploy/doctor). Rendering and deploying happen elsewhere in the app — this card is read-only.

Builds the 90-day launch plan (pre-order, ARC seeding, launch day, follow-ups) for a project + target release date. Display-only — every external action in the timeline is gated behind the confirmation flow elsewhere in the app; nothing here executes anything.

Estimates cost + market ROI for translating a manuscript into one or more foreign-rights languages (POST /api/translation/plan). Display-only — this never calls /propose or /execute, so no translation work is kicked off from this card.

Assembles a BookBub Featured Deal submission draft (BookBub-style blurb, price point, rationale) from your Amazon listing details (POST /api/bookbub/draft). Produces a draft to review and paste into partners.bookbub.com manually — nothing is submitted from here.

Paste exported Amazon Advertising keyword performance data (one row per keyword) and get pause/bid recommendations (POST /api/ams/optimize). Never proposes a bid increase over 2x and never executes anything in AMS — recommendations only.

Browse the built-in beat-sheet library (GET /api/structures), check a pasted outline against a chosen structure's beats (POST /api/structures/check-outline), or get a genre-based recommendation (POST /api/structures/recommend). Advisory only — nothing here forces a structure on your project.

Tracks Sanderson-style "promises and payoffs": what the opening chapters implicitly promise the reader, and whether later chapters pay them off (GET/POST /api/projects/:id/plot-promises). Extraction and audit are analysis only — nothing is rewritten.

Builds a per-character style fingerprint from dialogue as chapters are written, then flags "drift" — moments where a character stops sounding like themselves (GET/POST /api/projects/:id/character-voices). Ingest and drift-detection are local analysis, no rewriting.

Cross-book memory for multi-project series: merges entity indexes, timelines, and contradictions across every book in a series (GET /api/series, GET /api/series/:id/report). Read-only — series are created/managed elsewhere in the app.