Documentation
CodeForge docs
A local-first workbench for static sites. Everything below runs on local source in your browser.
Getting started
Open a project
Load your static site folder into the workspace. Files are read locally and stay on your machine.
Make the change
Edit in place, or use project-wide search to find every hit and fix it once across files.
Checkpoint
Snapshot the project before risky moves. Diff against the checkpoint at any time.
Prove the release
Run quality gates, dependency review, and the route graph. Package the release from the evidence.
Concepts
Workspace - the open project: files, editor, and command palette. Quality gate - a named check with a pass/fail verdict and the failing detail attached. Checkpoint - a snapshot of project state you can diff against. Dependency review - a static scan listing every remote script, style, image, link, and module import your source references. Route graph - the crawl of your local HTML: internal links resolved, fragments verified, orphans found, titles checked, form labels flagged.
Search and fix
Search runs across the whole project and shows each hit in file context. A fix applies across files with the hit list still visible, so the blast radius of a rename is on screen while you make it.
Quality gates
Gates return verdicts with evidence: which check, which file, which line or rule. The release view assembles gate results, diffs, and the route crawl into the package you ship - a release is a claim with its proof attached.
Checkpoints and diffs
Checkpoint before a risky operation, then diff file by file. The history view shows exactly what changed between two points - no summarization, no lost files.
Dependency review
The scan reads your source and lists remote references: scripts, stylesheets, images, links, module imports. It performs no installs and issues no network requests - the output is what a visitor's browser would fetch, derived statically.
Route graph
The crawler resolves internal links across local HTML files, verifies fragment targets, finds orphan pages, checks that page titles are unique, and flags unlabeled form controls. Everything is computed from local files; nothing leaves the browser.
Data and export
Project files stay on your machine. Checkpoints and gate results live in browser storage for the session and can be exported with the release package.
Privacy
No analytics, no cookies, no uploads. Analysis is static and local. See the trust center.