Dalea MCP server

list_brands

readidempotent

List the user's brands with niche and connected platforms, and which one is active. Use the returned id as brandId for other tools.

No parameters.

get_brand_profile

readidempotent

Get a brand's full profile: basics (name, language, strategy summary), the onboarding brief (niche, audience, pain, promise, tone, pillars, taboo topics), connected social accounts with follower counts, business segment and subscription status. Call this first to understand who the creator is before generating or reviewing content.

get_brand_kit

readidempotent

Get the brand's visual identity kit (colors, fonts, logo, palette, voice description) used by the video editor and carousels. Returns kit: null when none was generated yet.

get_account_overview

readidempotent

Get the brand's connected accounts with followers, followersGained30d (null when there is no metric snapshot at least 30 days old), total videos, reach, and lastSyncedAt.

list_videos

readidempotent

List the brand's published videos with metrics and classification (funnelStage, format, communicationStyle, mediaType). Filters: platform, since/until (ISO), funnelStage, format, minViews, mediaType (video|image|carousel). sortBy: views | engagement | published_at (desc). limit default 20 max 100, offset for paging. Captions are truncated to 300 chars; use get_video for full detail.

get_video

readidempotent

Get one of the brand's videos by platformVideoId: full metrics (incl. averageWatchTime, reach, saves), plus the AI analysis and transcript when they exist (null otherwise).

get_performance_summary

readidempotent

Totals for videos published in the last N days (7|30|90, default 30): videoCount, views, likes, comments, shares, saves, averageEngagementPct; the same for the previous window; deltaPct per metric (null when the previous value is 0); and the topVideo by views.

get_top_performers

readidempotent

Rank the brand's videos published in the last N days (default 30) by metric: views | engagement | saves | shares | likes. direction top (default) or bottom. limit default 10 max 100.

get_metrics_daily

readidempotent

Daily aggregated metrics per connected account between from and to (YYYY-MM-DD, inclusive; range capped to the last 180 days before `to`). One row per account per day: date, platform, totals and averageEngagementPct. Optional platform filter.

get_niche_benchmarks

readidempotent

Get median/p75/p90 view and engagement benchmarks for the brand's niche. With compare=true also returns `brand` (medianViews, p75Views, medianEngagementPct, sampleSize) computed from the brand's own videos of the last 90 days, or null when it has none.

get_voice_profile

readidempotent

Get the brand's AI-derived voice profile (status, confidence, profile, exemplars) and form profile (status, profile, stats). Each is null when it has not been generated yet.

set_video_classification

writeidempotent

Manually classify one of the brand's published videos (by platformVideoId): funnel (tofu | mofu | bofu), format (taxonomy slug, e.g. talking_head, tutorial, listicle) and/or communicationStyle (style slug). Provide at least one; pass null to clear an axis back to unclassified. Only the classification columns change (marked as manual so auto-classification will not overwrite them); metrics, analysis and the video itself are NOT touched, and nothing is published. Returns the video's resulting classification.

list_contents

readidempotent

List the brand's content cards (board + calendar) as summaries: id, title, idea, status, platforms, postedAt, mediaType, category, funnelStage, format, updatedAt. Filters: status (idea | draft | edited | posted), platform (tiktok | instagram), mediaType (short_video | carousel | short_text), from/to (planned date range, YYYY-MM-DD or ISO), search (matches title or idea). Paginate with limit (default 50, max 100) and offset. Use get_content for the full script.

get_content

readidempotent

Get one content card's full detail by id: the script (hook, body, cta, caption), planning fields (status, platforms, postedAt, mediaType, category, funnelStage, format, communicationStyle, targetDurationSeconds) and how many references and AI generations it has. Use list_content_references / list_content_generations / get_refine_chat to drill into those. The card's url opens it in Dalea: share it so the user can review and edit it there.

list_content_generations

readidempotent

List the AI script generations saved for a content card (newest first): title, hook, cta and the alternative body/caption options each run produced. Use it to see what was already tried before regenerating.

list_content_references

readidempotent

List the references (videos, links, files) attached to a content card, in order: id, referenceType, referenceUrl, position, title.

get_refine_chat

readidempotent

Get the last N messages (default 20, max 100) of the refine chat attached to a content card, as { role, content } pairs, plus the total message count. Use it to recover what the creator asked for before editing the script.

create_content

write

Create a new content card on the board/calendar (no AI). Requires a title or idea. status: idea | draft | edited | posted (default idea). scheduledDate (YYYY-MM-DD or ISO timestamp) sets the calendar date. The card's url opens it in Dalea: share it so the user can review and edit it there.

update_content

writeidempotent

Edit a content card's editorial fields (no AI). Provide at least one of title, idea, hook, body, cta, caption, platforms, mediaType, category, funnelStage. Does NOT change the board column, order, or date, use move_content / schedule_content for those. The card's url opens it in Dalea: share it so the user can review and edit it there.

move_content

writeidempotent

Move a content card on the kanban board: change its column (status) and/or its order within a column. Provide at least one of status or placement. placement top|bottom positions within the target column (default bottom); the exact order index is managed automatically. status: idea | draft | edited | posted. Note: the 'posted' column does not publish anything.

schedule_content

writeidempotent

Set the planned calendar date for a content card (the day it should be posted). Pass YYYY-MM-DD for a day (stored at noon UTC so the day is stable in every timezone) or a full ISO timestamp when the exact time matters. This only plans the date; it does not publish.

unschedule_content

writeidempotent

Clear a content card's planned calendar date (sets posted_at to null).

delete_content

writedestructiveidempotent

Delete a content card from the board/calendar by id. The card disappears from the app (soft delete).

reorder_contents

writeidempotent

Set the order of cards inside one kanban column (status: idea | draft | edited | posted). orderedIds (1-200, all must belong to that column) become positions 0..n-1 in the given order; cards of the column not listed keep their relative order after them. Does NOT change any card's status, date or fields, and does not publish anything. Returns the full column order as { id, position }.

duplicate_content

write

Create a copy of a content card (same title, idea, script, caption, platforms, mediaType, format, category, funnelStage, communicationStyle, targetDurationSeconds) at the bottom of the idea column, with no planned date. The original is NOT changed; references, generations and refine chat are NOT copied; nothing is published. Returns the new card and duplicatedFromId. The card's url opens it in Dalea: share it so the user can review and edit it there.

bulk_create_contents

write

Create 1-20 content cards in one call (no AI). Each item has the create_content shape (title or idea required; optional status, scheduledDate, platforms, mediaType, category, funnelStage, hook, body, cta, caption). Cards are created in order at the bottom of their column; nothing is published. Not idempotent: calling twice creates the cards twice. Returns { contents, created }. The card's url opens it in Dalea: share it so the user can review and edit it there.

list_brainstorm_suggestions

readidempotent

List the brand's active brainstorm suggestions with reasons. Optional status filter: pending | added (contentId points to the created content) | dismissed.

accept_brainstorm_suggestion

write

Accept a brainstorm suggestion: creates a content card in the idea column from its title, idea and funnel stage (no AI), then marks the suggestion as added with contentId. scheduledDate (YYYY-MM-DD or ISO) overrides the suggestion's suggested date as the planned date. The suggestion's text is NOT changed; nothing is published. Fails if the suggestion was already added. Returns { suggestion, content }.

dismiss_brainstorm_suggestion

writeidempotent

Mark a brainstorm suggestion as dismissed so it leaves the pending list. Does NOT delete it, does NOT touch any content card already created from it, and publishes nothing. Returns { suggestion }.

list_insights

readidempotent

List the brand's AI-generated insight tiles.

get_homepage_insights

readidempotent

Get the latest homepage insight cards generated for the brand (insights array, how many videos they were based on, language).

list_saved_references

readidempotent

List the brand's saved reference videos (with catalog videoId, note and whether an analysis exists, use get_catalog_video for the transcript) and creator inspirations (creatorId for get_creator / get_creator_analysis).

get_creator_analysis

readidempotent

Get the stored deep analysis for a saved creator by creatorId.

get_calendar

readidempotent

Get the brand's content calendar for a date range as one entry per day (empty days included), each with the cards planned for that day (id, title, status, platforms, mediaType, postedAt). from/to are YYYY-MM-DD (UTC); default is today through the next 7 days. Use it to see what is scheduled and find free days before schedule_content.

search_catalog_videos

readidempotent

Search the shared reference catalog of top-performing creator videos. Defaults to the brand's niche and content language (language 'all' disables the language filter). query understands format names, funnel phrases (tofu/mofu/bofu) and presentation modes (text on screen / spoken), and matches the rest against captions and handles. format is a taxonomy slug. sortBy: views (default) | virality (engagement rate) | recent. Creators the brand hid are excluded. Paginate with limit (max 100) and offset.

get_catalog_video

readidempotent

Get one catalog video by id with its stored analysis (transcript and virality score) when one exists.

search_creators

readidempotent

Search catalog creators by name or handle (min 2 chars). Optional niche tag and platform filters. Returns each creator with their platform accounts and follower counts.

get_creator

readidempotent

Get a catalog creator by id: profile, AI summary, platform accounts and their 5 most-viewed catalog videos.

list_niche_patterns

readidempotent

List the recurring script patterns detected in a niche's viral videos (name, description, structure, average virality, usage count). Defaults to the brand's niche.

list_creator_facts

readidempotent

List facts the app has learned about the creator (story, pillars, arguments, restrictions, preferences, products, milestones, voice, audience, goals). status defaults to active. limit max 100.

get_script_preferences

readidempotent

Get the brand's script preferences: content pillars with weights, preferred formats, default duration, approved and banned hooks, standing instructions.

list_restrictions

readidempotent

List things the brand never wants in its content (restrictions), with where each one came from.

get_creator_memory

readidempotent

Get the AI-maintained memory of who this creator is (free text plus structured fields). Fields are null when no memory was generated yet.

get_creator_context

readidempotent

Get the creator interview outcome: phase, origin story, 90-day goals, what content already works, struggles, themes and voice. Never includes the interview transcript.

list_macro_topics

readidempotent

List the brand's macro content topics grouped from its creator facts, with usage counts and the fact ids behind each one.

add_restriction

write

Add a restriction: something the brand never wants in its content. Text is whitespace-normalized and capped at 280 characters; an identical existing restriction is not duplicated. Returns the full restriction list. Does not touch creator facts or script preferences.

remove_restriction

writedestructiveidempotent

Permanently delete one restriction by id (from list_restrictions). Returns the remaining restriction list.

update_script_preferences

writeidempotent

Update the brand's script preferences. Only the keys present in patch are replaced (a list key replaces the whole list); everything else is kept. formats must be taxonomy slugs (talking_head, voiceover_broll, listicle, tutorial, storytelling, reaction, pov, street_interview, day_in_life, podcast_clip, green_screen). defaultDurationSeconds 5-600 or null. Restrictions live elsewhere (add_restriction). Returns the full preferences document.

add_creator_fact

write

Record a fact about the creator in the learning ledger (source manual, status active). category defaults to preferencia. If an active fact with the same text already exists it is returned instead of duplicated (created: false). Does not regenerate the creator memory.

archive_creator_fact

writeidempotent

Archive a creator fact by id so it stops feeding generation. The row is kept (status archived), not deleted. Returns the updated fact.

delete_saved_reference

writedestructiveidempotent

Permanently remove a saved reference video (id from list_saved_references). The catalog video itself and any inspirations are untouched.

hide_creator

writeidempotent

Hide a creator (by handle, with or without @) from this brand's reference catalog searches. Already-hidden creators are left as is. Does not delete saved references from that creator.

unhide_creator

writeidempotent

Show a previously hidden creator (by handle) in this brand's reference catalog again. No-op when the creator was not hidden.

update_brand_kit

writeidempotent

Update the brand's visual identity kit: colors (primary, accent, background, foreground, palette), fonts (fontFamily, headingFont), logoUrl, voiceDescription. Only the keys in patch change; source becomes manual and sourceUrl is preserved. Fails when the brand has no kit yet unless the patch carries every required field.

update_strategy_brief

writeidempotent

Update the brand's strategy brief (onboarding brief). Only these keys can change: niche, niches, topics, profileType, contentGoal, icpWho, icpWants, mainPain, promiseLine, toneStyle, contentPillars, tabooTopics, mediaTypes. Keys not in patch are kept; offer, site and connected-platform data are never touched. Returns the brief's editable keys.

list_trend_feed

readidempotent

List trending topic clusters matched to the brand, best fit first: fit score and reasons, a suggested angle, and the cluster's title, description, category, platforms, momentum, tags, post count and average virality. limit max 100.

get_trend_cluster

readidempotent

Get one trend cluster by clusterId, plus the brand's fit (score, reasons, suggested angle, content guide) when the cluster is in the brand's feed.

list_carousel_drafts

readidempotent

List the brand's carousel drafts from the Carousel Studio (title, slide count, last edit). Use get_carousel_draft for the slide texts. Each draft has a url that opens it in the Dalea Carousel Studio: share it when the user wants to see or edit a carousel.

get_carousel_draft

readidempotent

Get one carousel draft with its slides (role, image prompt, headline/body/CTA texts). No image files are returned; share the url so the user can see and edit the carousel in the Dalea Carousel Studio.

list_carousel_posts

readidempotent

List carousel/video publish jobs sent to Instagram or TikTok: status, scheduled time, per-platform result (permalink or error). Optional status filter.

list_video_projects

readidempotent

List the brand's video editor projects (name, status queued|processing|preview_ready|ready|failed, language, enabled edits, durations). Each project has a url that opens it in the Dalea video editor: share it when the user wants to watch or adjust a video.

get_video_job_status

readidempotent

Get the latest render job of a video editor project: status, stage, progress percent, error code, plus the url of the project in the Dalea video editor.

search

readidempotent

Search the user's Dalea workspace: planned/written contents, the reference video catalog, and creators. Returns ids to pass to fetch. Use for open-ended questions when you do not know which specific tool applies.

fetch

readidempotent

Fetch the full text of a search result by id (content:<id> = script with hook/body/cta/caption; catalog:<id> = reference video caption + transcript; creator:<id> = creator bio and summary).

sync_metrics

write6/day

Fetch fresh metrics from every connected TikTok/Instagram account (platform APIs + scrape). Consumes the daily quota (6/day). Returns per-account counts.

No parameters.

generate_title

write50/day

AI-generate a working title. Pass contentId to title an existing content (saves it and returns { content }) or idea for a free-form idea (returns { title }). Consumes the daily quota (50/day, one short AI call).

regenerate_brainstorm

write5/day

Replace the brand's active brainstorm set with freshly generated suggestions (optional manualContext steers them). Consumes the daily quota (5/day, several AI calls). Returns { suggestions } like list_brainstorm_suggestions.

generate_insights

write5/day

Regenerate the brand's homepage insights from its synced videos (needs 10+ videos). Consumes the daily quota (5/day, three AI calls). Returns { insights } like get_homepage_insights.

save_reference_by_url

write30/day

Save a TikTok/Instagram video or creator profile URL as a brand reference (scrapes it if unknown; video analysis runs in the background). Consumes the daily quota (30/day). Poll with list_saved_references.

run_creator_deep_analysis

write10/day

Run (or return the cached) AI deep analysis of a saved creator by creatorId. Consumes the daily quota (10/day, one long AI call). Returns { analysis } as JSON text; read it later with get_creator_analysis.

generate_scripts

write10/day

Generate a full AI script (hook, CTA, body and caption options; carousel slides for carousels) for 1-5 content cards and save it to each card. Uses the brand's memory, references, sources and web search when relevant. A card generated in the last 60 seconds is returned as-is (status skipped). Consumes daily quota (10/day). durationSeconds sets the target spoken length. bodyOptions/captionOptions are suggestions to apply with update_content. Each result has a url to the card in Dalea: always share it so the user can open, review and edit the script there.

generate_hook_options

write30/day

Generate up to 3 AI hook (opening line) suggestions for a content card, using the brand's context, memory and references. Returns options only, nothing is saved; apply one with update_content. Consumes daily quota (30/day). objective defaults to engagement.

generate_body_options

write30/day

Generate up to 3 AI body (script development) suggestions for a content card, using the brand's context, memory and references. Returns options only, nothing is saved; apply one with update_content. Consumes daily quota (30/day). objective defaults to engagement.

generate_cta_options

write30/day

Generate up to 3 AI call-to-action suggestions for a content card, using the brand's context, memory and references. Returns options only, nothing is saved; apply one with update_content. Consumes daily quota (30/day). objective defaults to engagement.

generate_caption_options

write30/day

Generate up to 3 AI caption suggestions for a content card, using the brand's context, memory and references. Returns options only, nothing is saved; apply one with update_content. Consumes daily quota (30/day). objective defaults to engagement.