Save winning plays as skills

---
name: Weekly territory refresh
description: Loads the rep's own CRM accounts, scores them, and returns
  the 20 to work this week with the reason for each. Use for "weekly
  refresh", "what should I work this week", or Monday planning.
---

## Goal
A ranked list of 20 accounts from the rep's own territory to work this
week, each with its score reason and last activity.

## Inputs
- CRM — default: the team's connected CRM
- how many — default: 20

## Steps
1. Load the rep's own accounts from the CRM (their book, modified in the
   last 180 days).
2. Add the owner, last activity, next activity, and open deals columns.
3. Score the accounts against the team's default scoring profile.
4. Take the top N by score. If fewer than N have a score, say how many
   were skipped for lack of data and stop.

## Output
A table: Account, Score, Top reason, Last activity, Open deals. Then one
line naming the three accounts that moved most since last week.

## Guardrails
Read-only. Never enrich phones or emails. Never delete rows.

---
name: Closed-lost re-engagement
description: Builds a dialable list of decision makers at accounts lost
  in a time window, with the loss reason beside each person. Use for
  "re-engage closed-lost" or "win-back list".
---

## Goal
A people list of 2 to 3 decision makers per closed-lost account, each
carrying the deal's loss reason, amount, and close date, ready to export.

## Inputs
- time window — default: last 12 months
- minimum deal amount — default: none
- titles — default: VP of Sales, Director of Sales Development, SDR Manager

## Steps
1. Load closed-lost deals from the CRM for the window. Apply the minimum
   amount if given.
2. Remove accounts that have an open deal today.
3. Find the target titles at each remaining account.
4. Bring the loss reason, deal amount, and close date onto each contact.
5. Stop and tell the user the contact count. Offer phone enrichment; do
   not run it.

## Output
A people table with Account: Loss Reason, Account: Deal Amount, and
Account: Close Date columns, grouped by loss reason in the summary.

## Guardrails
Never enrich phones or emails without approval. Never change CRM records.

---
name: Hiring-signal prospecting
description: Finds companies hiring for a role in the last 60 days that
  do not run a product in our category, sizes the pool first, and lands
  a scored account list. Use for "hiring signal list" or "who's hiring
  SDRs".
---

## Goal
A scored list of companies with an active hiring signal for our buyer's
team and no incumbent product, capped at a size the rep can work.

## Inputs
- role being hired — default: SDR
- product category to exclude — default: dialer software
- geography — default: United States
- headcount band — default: 100 to 2,000
- cap — default: 50

## Steps
1. Size the search for free. If the pool is under 20, widen recency to 90
   days and the headcount band, then size again.
2. Run the hiring search with the stack-gap exclusion. Keep the Open Roles
   count on every row.
3. Check which accounts are already in the CRM and add the owner column.
4. Remove accounts owned by someone else.
5. Score the rest.

## Output
A company table sorted by score: Company, Open Roles, Score, CRM owner.
State the pool size you found and how many were removed as owned.

## Guardrails
Do not find contacts or enrich anything. Stop at the account list.

---
name: Pre-call account brief
description: One-screen briefing on a company before a call, from our
  own data first. Use for "brief me on", "prep me for my call with", or
  "what do we know about".
---

## Goal
A briefing a rep can read in 60 seconds before dialing.

## Inputs
- company name — required

## Steps
1. Read the CRM record and its associated contacts, open deals, and
   recent activity.
2. Read our call history with the account: dials, connects, dispositions,
   and the last three call summaries.
3. Read any battle card that names a competitor on the record.
4. Check the account's current hiring activity.
5. Only if steps 1 to 3 return nothing, offer external research; do not
   run it unasked.

## Output
Sections in this order: Relationship (owner, stage, last touch), People
we know (name, title, last disposition), Open deals, Recent calls (three
lines), Hiring right now, Suggested opener (one sentence).

## Guardrails
Internal data first. Never create or update any record.

---
name: New-in-seat outreach
description: Finds leaders who started a target role in the last 90 days
  at companies matching our ICP, with the account's CRM status beside
  each. Use for "new in seat", "new leaders", or "recently hired VPs".
---

## Goal
A people list of recent movers into buying roles, ranked, with a note on
whether their new company is already a customer or an open opportunity.

## Inputs
- titles — default: VP of Sales, CRO, Head of Revenue
- job-change window — default: 90 days
- ICP filters — default: US, 200 to 2,000 employees, B2B software

## Steps
1. Find people with the titles who changed jobs within the window at
   companies matching the ICP filters.
2. Check whether each person's company is in the CRM and add the account
   owner and deal stage.
3. Let people scoring run, then list the top 25.

## Output
A people table: Name, New title, Company, Started, CRM status, Score.
Flag any mover whose new company is an existing customer.

## Guardrails
Never enrich phones or emails without approval.

---
name: Look-alike expansion from wins
description: Turns last year's closed-won accounts into a scored list of
  look-alike companies not yet in the CRM. Use for "expand from wins" or
  "companies like our customers".
---

## Goal
A scored company list of look-alikes to our best customers, excluding
anything we already have in the CRM.

## Inputs
- seed window — default: closed-won in the last 18 months
- how many — default: 100
- constraints — default: same country as the seeds

## Steps
1. Load closed-won accounts for the window.
2. Find look-alikes seeded from that whole table, with the constraints.
3. Remove any that are in the CRM or on the seed list.
4. Score the rest against the team's default scoring profile.

## Output
A company table sorted by score, with the firmographic columns, and a
one-line note on which seed accounts the top results most resemble.

## Guardrails
Stop at the account list. Do not find contacts or enrich.