crm-records
Skill0 starsChanged 10 days agoMITView on GitHub
How CRM records are created and kept: contact, company, deal, associations, required fields, the deal stage ladder with real IDs, and the custom properties the post-call loop writes. Use for creating or updating a contact, qualifying a lead, creating a deal, or looking up a stage ID. The one home for pipeline and stage IDs.
In this repository12 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- call-debriefSkill
- company-contextSkill
- humanize-copySkill
- meeting-prepSkill
- pipeline-analysisSkill
- product-contextSkill
- sales-call-scorerSkill
- sales-intakeSkill
- sales-playbookSkill
- setupSkill
- weekly-exec-updateSkill
The file
Files
CRM records
Written for HubSpot; the shape holds for any CRM. Replace every <ID> with your portal's
values and verify them against the live portal before the first unattended run.
One creator per object
| Object | Created by | Never created by |
| Contact | booking receiver, inbound form, enrichment | the post-call loop |
| Deal | CRM workflow when a meeting is booked, or the founder by hand on a plan with no workflows | any agent reading email or transcripts |
| Task | sales-intake, people | — |
Two creators for one object is how duplicate deals happen.
Card route and owners
<!-- Filled at setup step 2. --><!-- card_route is workflow (the CRM posts the call card) or slack-connector (sales-intake posts
it; for CRM plans without workflows). -->card_route: workflowsales_channel: #CHANGE-MEdefault_owner: NAME(owns tasks on an unowned deal)
Pipeline
Deals pipeline <PIPELINE_ID> is the only pipeline the post-call loop touches.
Stage ladder
| Stage | ID | Probability | Enters when | Scorer call type |
| Lead | <ID> | 0.05 | Created, not yet qualified | Discovery |
| Qualified | <ID> | 0.10 | Qualified and first meeting booked | Discovery |
| Demo | <ID> | 0.20 | First meeting held | Demo |
| Trial | <ID> | 0.40 | Trial access granted | Proposal |
| Proposal Sent | <ID> | 0.60 | Proposal delivered | Proposal |
| Verbal | <ID> | 0.80 | Option chosen, paperwork pending | Negotiation |
| Closed Won | <ID> | 1.0 | Signed | — |
| Stalled | <ID> | 0 | No motion, not dead | infer from the call |
| Closed Lost | <ID> | 0 | Dead, with a reason | — |
Stages are moved by a person. Agents propose. The one exception: a completed DocuSign envelope
moves its deal to Closed Won through the signature receiver (close/CLOSE.md).
Required fields
Contact: first name, last name, email, company, job title, lead source, buying role.
Company: name, domain, industry, size.
Deal: name (<Company> — <offer>), pipeline, stage, owner, amount if known, close date.
Associations: deal ↔ company ↔ contacts, all three.
Properties the post-call loop writes
| Property | Type | Rule |
description | text | Append a dated activity-log entry per call. Never replace. This is the record |
hs_next_step | text | Replace. Must carry owner and date |
blockers | text | Replace. Point in time |
last_call_at | date | Writing it fires the Slack card workflow |
last_call_score | number | 0–100 |
last_call_grade | enum | A–F |
last_call_tier | enum | same_day / standard / escalate |
last_call_headline | text | [call type] ([duration]) with [attendees] - [next step, owner, date] |
The five last_call_* fields restate the log entry so a workflow can render them. The log
wins if they disagree.
Activity-log entry format
M/D CALL NOTE: [call type] ([duration]) with [attendees].
Call score: [N]/100 ([grade]).
- What happened, in three to five bullets
- Decisions and signals, with short quotes
NEXT: [owner] [action] by [date]Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

