Anthra developer platform

Build with Anthra

Anthra exposes two programmatic surfaces on top of the same studio that powers /dashboard:

REST API at /api/v1/* — JSON in, JSON out (plus streamed text/plain for chapter generation).
MCP server at /api/mcp — connect Claude, Cursor, or any Model Context Protocol client with a personal access token and Anthra appears as a set of callable tools.

Both surfaces use the same authentication, the same scopes, and the same underlying database — you can mix them freely.

Start here

Quickstart — mint a key, create a book, write its first chapter end-to-end with curl.
Authentication & keys — tokens, scopes, expiry, revocation.
Connect via MCP — Claude Desktop, Cursor, and Claude Code configs.

What you can do

Capability	REST	MCP tool	Scope
List / read books	`GET /v1/books`	`list_books`, `get_book`	`books:read`
Create or edit a book	`POST /v1/books`, `PATCH /v1/books/{id}`	`create_book`, `update_book`	`books:write`
Generate outline	`POST /v1/books/{id}/outline`	`generate_outline`	`generation:run`
Write / rewrite a chapter	`POST …/chapters/{cid}/write` (stream)	`write_chapter`, `rewrite_chapter`	`generation:run`
Narrate with ElevenLabs	`POST …/chapters/{cid}/narrate`	`narrate_chapter`	`generation:run`
Publish / unpublish	`POST /v1/books/{id}/publish`	`publish_book`, `unpublish_book`	`publish:write`
Reader analytics	`GET /v1/books/{id}/analytics`	`get_reader_analytics`	`analytics:read`
Account + token usage	`GET /v1/account`, `GET /v1/usage`	`get_usage`	`account:read`
Planning surface (theme, promise, comp, locks)	`GET / PATCH /v1/books/{id}/plan`	`get_plan`, `set_plan`	`books:read` · `books:write`
Characters (want / fear / lie)	`GET / POST /v1/books/{id}/characters`	`list_characters`, `create_character`	`books:read` · `books:write`
Writer's bible (characters, plot, world, glossary…)	`GET /v1/books/{id}/documents`, `GET / PUT /v1/books/{id}/documents/{kind}`	`list_book_documents`, `read_book_document`,`write_book_document`, `draft_book_document`	`books:read` · `books:write`
Beat sheets (Save the Cat, Freytag, …)	`GET /v1/books/{id}/beat-sheets`	`list_beat_sheets`	`books:read`
Sensitivity flags	`GET / POST /v1/books/{id}/sensitivity`	`run_sensitivity_scan`, `list_sensitivity_flags`, `dismiss_sensitivity_flag`	`generation:run` · `analytics:read`
Background jobs	`GET /v1/jobs`, `GET /v1/jobs/{jobId}`	`list_jobs`, `get_job_status`	`account:read`
Narration voice directory	`GET /v1/voices`	`list_voices`	`books:read`
Batch narration (whole book)	loop `POST …/chapters/{cid}/narrate`	`narrate_book`	`generation:run`
Cover art / chapter illustrations	`POST /v1/books/{id}/cover-image`, `POST /v1/books/{id}/back-cover`, `POST /v1/books/{id}/spine`, `POST /v1/books/{id}/cover/compose`, `POST /v1/chapters/{cid}/image`	`generate_cover_art`, `generate_back_cover`,`generate_spine`, `compose_cover_spread`,`generate_chapter_image`, `choose_cover_color`	`generation:run`
Chapter restructuring	`PATCH /v1/books/{id}/chapters/{cid}`	`list_chapters`, `update_chapter`, `split_chapter`, `merge_chapters`, `reorder_chapters`, `set_chapter_meta`, `list_chapter_revisions`, `restore_chapter_revision`	`books:read` · `books:write`
Manuscript export (EPUB / DOCX / PDF / Markdown)	`POST /v1/books/{id}/produce`	`export_book_markdown`, `produce_book`, `get_production_job_status`	`generation:run`
Translation (clone into another language)	`POST /v1/books/{id}/translate`	`translate_book`	`generation:run`
Editorial & quality (lint, critique, continuity, drift, repetition, cliches)	—	`lint_prose`, `critique_chapter`, `check_continuity`, `find_repetition`, `suggest_cliche_replacements`, `spoken_readability`, `style_drift`, `set_forbidden_words`	`books:read` · `generation:run`
Prose analysis (filter words, adverbs, POV/tense, rhythm, clichés)	—	`find_filter_words`, `analyze_adverb_density`, `analyze_sensory_balance`, `check_pov_slippage`, `check_tense_consistency`, `analyze_sentence_rhythm`, `detect_semantic_cliches`	`books:read` · `generation:run`
Specialist & critique panel (round-table, plot doctor)	`POST /v1/tools/round_table_critique`	`call_specialist`, `round_table_critique`, `skeptic_verify`, `draft_tournament`, `long_horizon_plot_doctor`	`generation:run`
Reader feedback (test-reader panel, emotional arc, hooks)	`POST /v1/tools/test_reader_panel`	`test_reader_panel`, `emotional_arc`, `score_hook`, `score_ending`	`generation:run`
Worldbuilding & continuity (rules, timeline, foreshadowing)	—	`check_continuity`, `check_scene_consistency`, `register_foreshadowing`, `list_unpaid_foreshadowing`, `refresh_timeline`, `list_object_trail`	`books:read` · `generation:run`
Persona & structural editing (track-changes, POV/tense flip)	`POST /v1/tools/edit_chapter_with_persona`	`edit_chapter_with_persona`, `track_changes_edit`, `sentence_variants`, `flip_chapter_tense`, `shift_chapter_pov`, `voice_transplant_paragraph`	`generation:run`
Collaboration (assignments & activity feed)	—	`list_collaborators`, `invite_collaborator`, `remove_collaborator`, `assign_chapter`, `list_chapter_assignments`, `list_activity_feed`	`books:read` · `books:write`
Author preferences	—	`list_author_preferences`, `set_author_preference`	`account:read` · `books:write`

Planning surface (added 2026-06)

The studio's planning workbench is now fully programmable. Both REST and MCP can:

Read and write the theme statement, reader promise, comp authors, beat sheet, POV / tense locks, and POV depth dial.
Manage the character roster with Truby-style want / fear / lie axes and the inter-character conflict matrix.
Generate and apply outline variants, grade the current outline, run the premise-sharpening interview, pick one of three premise pitches, auto-detect genre, and run era research with citation links.
List tone / mood references and roll the outline back via snapshots.

REST tool gateway (added 2026-06)

Beyond the idiomatic routes above, every tool in the toolkit is now reachable over a generic gateway — the same ~239 tools the MCP server and in-app agent share, at full parity. Two endpoints:

GET /api/v1/tools — the live catalog: every tool name, description, required scope, and JSON Schema for its arguments. This is the authoritative, always-current list.
POST /api/v1/tools/{tool} — invoke any tool by name with a JSON body matching its schema. Scope-gated exactly like the idiomatic routes.

# Discover every tool + its JSON Schema
curl "$ANTHRA_BASE/tools" \
  -H "authorization: Bearer $ANTHRA_TOKEN"

# Invoke one by name
curl -X POST "$ANTHRA_BASE/tools/generate_outline" \
  -H "authorization: Bearer $ANTHRA_TOKEN" \
  -H "content-type: application/json" \
  -d '{ "bookId": "…uuid…" }'

The gateway is the fastest way to reach newer capabilities — prose analysis, the specialist panel, reader feedback, worldbuilding & continuity, structural editing, and collaboration — without waiting for a dedicated route. See the tool reference for the catalog grouped by capability. OpenAPI is published at v1.1.0.

Versioning

The REST surface is namespaced by version (/api/v1). Anthra will introduce a new namespace before making any breaking change; v1 is stable and additive-only.