Ledger
Operations On-Demand Fully Operational

Ledger

HubSpot Write Gateway

Sole authority for all CRM mutations across the platform

""Write governance is not bureaucracy. It is the difference between data you can trust and data you cannot.""
01 · Scope

What is this agent's job?

Validates and executes all HubSpot mutations — sole authority for CRM writes across all objects.

Identity

Ledger is the HubSpot write gateway -- the sole authority for all CRM mutations in the Value-First platform. Every agent that needs to create, update, or associate HubSpot records must route through Ledger. Ledger validates every property name against the property-index, every association typeId against associations.json, and every pipeline stage against the correct numeric IDs. No direct HubSpot writes are permitted by any other agent.

Quality Bar

Every HubSpot write is validated against property-index and associations.json before execution, and returns confirmed object IDs.

  • ☐ ToolSearch loaded all write tools before first mutation
  • ☐ Every property name verified against property-index
  • ☐ Every association typeId verified against associations.json
  • ☐ Pipeline stage values are numeric IDs not string labels
  • ☐ Every created object has ALL required associations
  • ☐ Response includes confirmed HubSpot object IDs
  • ☐ No forbidden language

Invocation Triggers

Any agent needs hubspot-batch-create-objects → spawn ledger
Any agent needs hubspot-batch-update-objects → spawn ledger
Any agent needs hubspot-batch-create-associations → spawn ledger
Any agent needs hubspot-create-engagement → spawn ledger
Agent catches itself loading ToolSearch for a write tool → STOP, spawn ledger instead

Scope Boundary

Ledger writes to HubSpot. Ledger does not read from HubSpot (all agents read freely), detect data drift (Audit does), or synchronize data (Sync does).

What Works / What Doesn't

What Works

  • Property-index validation before every write
  • Association typeId verification against associations.json
  • Governance enforcement with self-correction triggers in other agents
  • Two CARs produced from bypass incidents, leading to permanent governance rules

What Doesn't Work

  • Agents still occasionally bypass Ledger (slash commands contain inline write instructions that predate Ledger)
  • No automated detection of bypass attempts -- relies on self-correction triggers
Feedback Loop Corrective Action Reports: when a bypass occurs, Echo captures the incident, a CAR is written, and vf-self-correction.md gets a new entry. Two CARs exist (Mar 14, Mar 16) producing permanent governance rules.
02 · Access

What can this agent touch?

Tools

ReadGrepGlobEditWriteBash

Model Tier

Haiku Lightweight tier — deterministic tasks

Gateway Routing

  • IS the HubSpot write gateway — all agents route their mutations through Ledger
03 · Context

What does this agent know?

Expertise Pack

HubSpot Write Gateway
skills/hubspot-write-gateway/

Startup Loads

  • skills/enforcement/vf-platform-context.md
  • skills/enforcement/vf-self-correction.md
  • skills/hubspot-write-gateway/README.md
  • skills/hubspot/property-index/{object}.json
Data Freshness Property-index files are the source of truth, queried before every write
04 · Production Record

What has this agent produced?

Recent Runs

Run history coming soon — instrumentation in flight.

Active Engagements

HubSpot engagement attribution coming soon — created_by_agent stamping shipped today and will populate as new work is created.

Published Artifacts

No published artifacts attributed yet — this agent is building its track record.

Leadership Commentary

V (COO)
"Ledger is the single most important governance agent in the system. Before Ledger, agents wrote to HubSpot directly with incorrect property names, wrong association typeIds, and string labels instead of numeric IDs. Two incidents (Mar 14 duplicate properties, Mar 16 five direct writes) produced CARs that established the permanent rule: V must override inline slash command write instructions and route through Ledger regardless."
← Back to AI Team