Companion to sumble-account-scoring: plan and rebalance sales territories. Interviews the user about their segments (default Enterprise + Commercial) and whether the segment line is a hard rule or should be calibrated from their data, pulls territory ownership from the CRM, and pulls per-rep×account activity (calendar meetings, Gong/Fireflies/Granola calls, Salesforce email) from whatever MCPs are connected. Generates a self-contained, zero-dependency Python + HTML/JS app at territory_planning/{company}/ with per-segment book-strength heatmaps (Capture plus each rep's top 25 accounts by average in-segment rank, and a matching coverage table), a granular account view, highlights for accounts that are not being worked / in the wrong segment / strong-but-unallocated / double-allocated / strong-but-idle (with a live top-N strong-account cutoff), and a suggest→accept/reject→export flow that writes an actions.csv of approved owner changes.
Guide a seller through researching and prospecting accounts on the Sumble MCP. Asks up front whether they're working a specific account or brainstorming, and which deliverable — outreach sequences, an account plan (own prep, SDR-to-AE handoff, AE-to-manager, or QBR prep), a deck, or call prep (they name who they're meeting; builds a brief from Sumble data + previous touchpoints); for a plan or deck, also asks format and an example. For brainstorm, ranks their Sumble territory/org list on ICP fit/score and recent triggers via SearchSignals (champion moves, new hires on tracked tech, hiring/tech trends), so a hot signal surfaces an account even when its score is middling. Builds a cached Sumble profile from GetMyCompanyProfile plus any sales plays / persona profiles. Then researches one account at a time — internal context (Gong/Fireflies/Granola/CRM/marketing), the Sumble overview (tech, teams, people, headcount, hiring signals, ICP fit), and recommended teams + people for land or expansion.
Build an account-scoring AND/OR whitespace web app powered by Sumble data and optional first-party data. One skill, three objectives: score the accounts in your CRM, find whitespace (high-ICP-fit orgs NOT in your CRM), or both in one ranked sheet. Interviews the user about their ICP, pulls data from internal systems and possibly Sumble MCP, and generates a self-contained, zero-dependency Python + HTML/JS app at account_scoring/{company}/ with real-time slider re-weighting, an account-category column + filters (customer / allocated / unallocated / whitespace), and an evaluation mechanism to tune the score. Outputs a config file describing the scoring method and a Python script that applies it across all accounts.
Build a CRM-cleaning review web app powered by Sumble data. Finds (1) potential duplicate accounts — CRM accounts that resolve to the SAME Sumble organization (Sumble's matcher is the only duplicate evidence; no domain or name-similarity matching) — and (2) missing or conflicting parent/subsidiary links, by matching every account to Sumble's organization graph via POST /v6/organizations (attributes incl. parent_id/subsidiary_ids — no SQL, no RunSqlQuery). Interviews the user about their CRM source and which checks to run, pulls the account list from internal systems or a CSV, and generates a self-contained, zero-dependency Python + HTML/JS review app at crm_cleaning/{company}/ with accept/reject/skip per finding, survivor selection for merges, and an actions.csv export of every approved CRM change.
Build a people-scoring web app powered by Sumble data and optional first-party data. Interviews the user about their ICP (job functions + skills via GetMyCompanyProfile), then builds a SMALL calibration sample from ~5 user-named companies via the unified v6 REST endpoints (POST /v6/people) so the app is ready in minutes. Optional gold set (closed-won opportunity contacts) drives an Evaluation tab and a conservative regularized weight fit. Generates a self-contained, zero-dependency Python + HTML/JS app at people_scoring/{company}/ with real-time slider re-weighting, a score.csv superset sheet, plus a production score_leads.py that applies the calibrated weights to an entire enriched CRM.
Use Sumble MCP for account research, prospecting, people/job/technology searches, organization/contact list workflows, and Sumble MCP prompt or docs authoring. Use when Codex should follow Sumble MCP tool sequencing, query rules, and credit-management guardrails.