review

Review reads the work and looks things up. The verifier runs the work and measures it. This agent never touches the file system and never executes what it is reviewing, but it can consult the web to check a claim: whether an API is used correctly, whether a cited source says what the work says it does. For Agency source specifically, agencyReviewAgent adds parse and typecheck findings.

Functions

buildTools

buildTools(): any[]

Return the reviewer's lookup tools. It reads and checks; it has no tools that change anything.

Returns: any[]

(source)

reviewAgent

reviewAgent(
  work: string,
  task: string = "",
  context: string = "",
  maxCost: number = $10.00,
  maxTime: number = 10m,
  model: string = "",
  provider: string = "",
  session: string = "",
  extraTools: any[] = [],
): Result<Feedback[]>

Review a work product against a task and return findings. error=true marks a way the work fails the task; error=false marks a satisfied point.

@param work - The work product to review, as text @param task - What the work should accomplish @param context - Extra material for the judgment, or "" @param maxCost - Hard spend cap @param maxTime - Hard wall-clock cap @param model - Model override, or "" for the ambient model @param provider - Provider for the model override @param session - Session name to share a thread across calls, or "" for isolated @param extraTools - Extra tools to offer the LLM, appended to the built-in set

Parameters:

Name	Type	Default
work	string
task	string	""
context	string	""
maxCost	number	$10.00
maxTime	number	10m
model	string	""
provider	string	""
session	string	""
extraTools	any[]	[]

Returns: Result<Feedback[]>

Throws: std::guard

(source)