Where every agent starts.
One index of what's available and what it means, so the AI tools your team already uses give better answers from your Salesforce data.
A published table of contents.
The catalog is published per domain. An agent reads it first to know what is available: every entity, brief and skill that domain allows, each with descriptions, instructions and details.
From there it decides where to dig in — which entity to load and at what depth, which briefs contain the background, which skill fits the task. The index also says how: how to load entities, what filters are available, what to fetch before aggregating.
A purpose-built agent gets a narrow set. A conversational tool gets a broad one.
How definitions get writtenDomain
The index
what every agent sees first
Entities (6)
- Account 360
- Campaign 360
- Contact 360
- Lead 360
- Opportunity 360
- User 360
Briefs (1)
- RevWorks Sales Methodology
Skills (1)
- Pipeline Review
The index every agent sees first
~937 tokensContextWorks Catalog — revworks_sales domain
Sales context as RevWorks curates it: deals with Deal Health, methodology status and progression, plus accounts, people and campaigns through their default views.
Three of six entities shown. Briefs and skills follow in the same index.
The agent's instruction manual for your data.
Domain specific
Each agent sees only the context in its own domain. Nothing outside that domain is available to it.
Markdown or JSON
The agent asks for the format it needs. Markdown or JSON changes how the context is laid out, not what it contains.
From table of contents, to all the details
The index lists what exists. From it, an agent loads the entity at the level of detail it needs, the briefs that explain it, and the skill for the task at hand.
Learn more
