Growth Dante Id

Search published entries

Semantic search over the creator's published knowledge base entries (playbooks, guidelines, content pieces). Call this whenever the user asks to search, find, look up, or retrieve content about any topic — pass the user's query directly. Returns up to 20 entries ranked by cosine similarity, with a keyword fallback. After getting results, call get_entry with the slug to read the full article. Limitation: only returns published entries. Example queries: 'growth strategies for personal brand awaren

Get a published entry

Fetch the full MDX body and metadata for a single published entry by slug. Use after search_entries once you have the slug — do not guess slugs. Returns title, description, publishedUrl, githubPath, and the raw MDX content. Returns an error message (not a thrown error) if the slug does not exist or the entry is not published. Limitation: slug must be an exact match — no partial or fuzzy matching.

Search verbatim moments

Find atomic moments (verbatim quotes with video timestamps) across the creator's recordings. Use this when you need a direct citation, a citable claim, or a demo moment with attribution — not when you need full entry context (use search_entries + get_entry for that). The optional kind filter narrows by moment type: hot_take (contrarian claims), atomic_finding (data points), demo_clip (before/after demos), bug_story (problem→solution arcs), discovery (surprising capabilities), comparison (A vs B

List topics + recent entries

Returns the top tools/keywords across published moments and the 20 most recently updated published entries. Call this first to orient — no arguments needed. Use the tool list to decide what queries to run in search_entries or search_moments, and use the recent entries list to find the latest published content without searching. Limitation: tool counts reflect published moments only; unpublished or undistributed moments are excluded.

Get creator profile

Returns canonical creator metadata: name, headline, bio, expertise areas, social links (X, GitHub, personal site), affiliation, target audience, topic pillars, core POV, and editorial off-limits. Call once per session for attribution or before generating content on behalf of the creator. Limitation: static — does not reflect real-time changes to the creator's profile.

Get undistributed high-value moments

Returns moments with potentialScore ≥ 7 that have not yet been distributed to any channel. Use this to find high-value content ready to post — call after 999x_growth_save_moments or when planning a distribution session. Returns title, kind, potentialScore, suggestedFormats, and toolsMentioned for each moment. Use alongside 999x_growth_analyze_funnel_coverage and the plan_distribution prompt to build a full action plan. Limitation: only returns moments with score ≥ 7 — lower-scored moments are ex

Analyze distribution funnel coverage

Returns a funnel stage breakdown (awareness / acquisition / activation) showing distributed vs undistributed moment counts, the top 5 undistributed moments by potentialScore, and a plain-English summary of the biggest coverage gaps. Call this together with 999x_growth_get_undistributed_moments and then pass both results to the plan_distribution MCP prompt to produce a full ranked action plan. Limitation: counts all moments regardless of potentialScore — low-scored moments are included in totals

Get moment extraction rules

Returns the complete moment extraction instruction set: the 7-kind taxonomy (hot_take, atomic_finding, demo_clip, bug_story, discovery, comparison, prediction), specificity rules, evidence rules, scoring guide, and output format. Call this first, use the returned text as your system instruction, then extract moments from your source material and output the JSON array. Agents inside Claude Code use this tool path; external MCP clients use the extract_moments prompt instead.

Get MDX generation rules

Returns the complete MDX generation system prompt for a given entry type and locale: Mintlify component reference, GEO optimization rules, locale-specific section template (playbook / guideline / content), hard formatting constraints, and required heading order. Call this, use the returned text as your system instruction, then generate MDX from the moments you already have in context. Agents inside Claude Code use this tool path; external MCP clients use the generate_entry prompt instead.

Get distribution planning rules

Returns the complete distribution planning framework: funnel stage definitions (awareness / acquisition / activation), kind-to-channel mapping, progress bar rendering rules, output format template, and recommended action thresholds. Call this, use the returned text as your system instruction, then synthesize it with data from 999x_growth_analyze_funnel_coverage and 999x_growth_get_undistributed_moments to produce a ranked distribution action plan. Agents inside Claude Code use this tool path; ex

Get voice profile

Returns the operator's custom voice profile if one has been configured via setup_voice. The rawText field is injected automatically into get_generation_rules — you only need this tool to check whether a profile exists or to display it to the user. Requires write authorization (Bearer gk_live_... token).

Save voice profile

Persists a synthesized voice profile for this operator. Once saved, get_generation_rules automatically injects this voice instead of the default. Call get_voice_profile first — if a profile already exists, pass overwrite: true to replace it. The voiceProfile fields are produced by the setup-voice Claude Code skill, which analyzes writing samples from your existing agent conversations. Requires write authorization (Bearer gk_live_... token).

Save extracted moments to knowledge base

Persists extracted moments into the knowledge base database and generates vector embeddings for each. Returns real UUIDs (momentIds) required by 999x_growth_publish_entry — you cannot publish without calling this first. Call once per session after the user confirms their moment selection; do not call multiple times for the same session. Accepts 1–20 moments per call. Embeddings are generated asynchronously — search_moments may not return these moments immediately. Requires write authorization (B

Publish a knowledge base entry to growth.dante.id

Commits a Mintlify MDX entry to GitHub, registers it in docs.json and the JSON-LD manifest, marks source moments as distributed, and inserts a published entry record in the database. Returns publishedUrl and commitSha on success. Prerequisites: (1) call 999x_growth_save_moments first to get momentIds — this tool will reject unknown UUIDs; (2) generate the MDX using the generate_entry MCP prompt. The commit is permanent — there is no undo. The entry appears on growth.dante.id after Mintlify's bui

Create a processing session for a URL source

Creates a content processing session for a YouTube video, Loom recording, or website URL. The server downloads and analyzes the source with Gemini, extracts moments using the 7-kind taxonomy, and persists them to the database. Returns a sessionId immediately — the processing runs in the background. Poll get_session until status is 'complete' (typically 1–4 minutes for a 20-min video). Requires write authorization (Bearer gk_live_... token).

Get session status and extracted moments

Returns the current status, progress, and extracted moments for a processing session. Call this after create_session or after uploading a video via POST /api/upload. Poll every 20 seconds until status is 'complete' or 'failed'. When complete, the moments array contains Gemini-extracted content ready for the select-moments pipeline. Requires write authorization (Bearer gk_live_... token).