Data tools run before an agent and inject real-world information directly into its context. Without data tools, agents work only from the run brief and your company profile. With them, agents can pull live search results, fetch documents, or call internal APIs.
How PRE-phase tools work
When a run reaches an agent with a data tool, Forge executes each tool in sequence before the agent starts. The results are formatted and inserted into the agent's context window. The agent then processes both the run brief and the fetched data together to produce its output.
Adding web search
In the team editor, click the tools icon on an agent, then Add Tool → Web Search. Configure it:
Name
Give it a descriptive name, e.g. "Competitor Research" or "Market Data".
Query
Use template variables to make the query dynamic:
{{agent.title}} competitor analysis UK 2025
{{agent.title}} market size TAM SaaS growth rateResult count
5–10 results is usually sufficient. More results increase context but also increase run time.
Adding a PDF reader
Add Tool → PDF Reader. Set the URL to a static document (e.g. a tender specification, annual report, or technical spec). The full text is extracted and injected into the agent's context. The URL can also use template variables if a previous tool returns a document URL.
# Static URL
https://example.com/technical-specification.pdf
# Dynamic URL from previous tool output
{{agent.output}}Adding an HTTP fetch
Add Tool → HTTP GET. Useful for internal APIs that return JSON data — for example, a product catalogue, a pricing API, or a CRM endpoint. Set the URL and any required headers using {{secret.KEY_NAME}} for authentication tokens. The JSON response is formatted and injected into the agent's context.