Skip to content
Foundry

Codebase Analysis

What it is

Section titled “What it is”

Codebase Analysis runs AI-powered scans of connected repositories against a program’s requirements. It determines which requirements are implemented, partially implemented, or missing — then presents results in a review queue where humans can approve or reject AI assessments.

Why it matters

Section titled “Why it matters”

Manually auditing a codebase against requirements is tedious and error-prone. Automated analysis catches implementation gaps, verifies coverage, and generates evidence that feeds into gate evaluations and client reporting.

Key concepts

Section titled “Key concepts”
  • Analysis runs: AnalysisRunTimeline shows the history of analysis runs with status and timing
  • Implementation status: fully_implemented, partially_implemented, not_found, needs_verification — per-requirement classification
  • Implementation badge: ImplementationBadge renders a color-coded status indicator
  • Status filters: Filter results by implementation status to focus on gaps
  • Review queue: ReviewQueue presents pending AI assessments for human review
  • Approve/reject: Reviewers can approve or reject each AI assessment
  • Analysis config panel: AnalysisConfigPanel configures what to analyze (repositories, directories, requirements scope)
  • Directory picker: DirectoryPicker selects specific directories within a repository to analyze
  • Analysis run detail: AnalysisRunDetail shows full results from a single run
  • Workstream analysis tab: WorkstreamAnalysisTab groups results by workstream
  • Task analysis panel: TaskAnalysisPanel shows per-task analysis details
  • Subtask proposal list: SubtaskProposalList suggests subtasks for requirements with incomplete implementation

How to use it

Section titled “How to use it”
  1. Navigate to [Program] > Analysis.
  2. Configure the analysis scope — select repositories and directories.
  3. Run an analysis — the AI scans code against all program requirements.
  4. Review the results grouped by implementation status.
  5. Filter by status to focus on not_found or partially_implemented requirements.
  6. Open the Review Queue to approve or reject AI assessments.
  7. View proposed subtasks for requirements that need additional implementation.

Data model

Section titled “Data model”

This feature uses the following tables:

Section titled “Related”
Requirements
Source Control
Repositories