review

an optional LLM judgment of whether the code accomplishes a task.

Review reads the work and looks things up. The verifier runs the work and measures it. This agent never executes the code it reviews, but it can consult the bundled Agency documentation and the web to check a claim.

Functions

reportFeedback

reportFeedback(report: TypeCheckReport): Feedback[]

Convert a typecheck report into findings: one error item per error, one advisory item per warning.

Parameters:

Name	Type	Default
report	TypeCheckReport

Returns: Feedback[]

(source)

buildTools

buildTools(): any[]

Return the Agency reviewer's lookup tools: the bundled language documentation, plus web lookups for anything outside it.

Returns: any[]

(source)

agencyReviewAgent

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

Review Agency source code and return findings. Always includes parse and typecheck findings. When a task is given, an LLM also judges whether the code accomplishes it.

@param source - Agency source code to review @param task - What the code should do, or "" for static checks only @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
source	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)