
2 · configure
3 · deploy
ProAvailable product — files, demo URL and secure download get mapped in EDF 0.3.
Catch prompt regressions before your users do
Everything in PromptForge, plus the tooling for prompts that ship to production. Save a test run as a baseline and see exactly which cases regressed, got fixed or changed cost and speed after every edit. Import datasets from CSV or JSONL (expected_* columns become checks), validate answers against a JSON Schema, score them with an LLM rubric, and set limits for latency and cost. Insights charts pass rate, p50/p95 latency, cost and rubric scores per version and model. Reusable snippets, a live prompt linter, prompt chains, retries with back-off and a spending budget. Keep the whole library as Markdown files in git, and run the same tests in CI with a headless runner that writes HTML and JUnit reports and exits 1 on a regression. React 19 + TypeScript, no chart or UI library, optional Node proxy for keys.
Origin: EWDS Store (epicwebdesignstudio.com) · Roadmap window: EDF 0.2


Placed in the private vault layout (deploy/vault/products). Downloads stay disabled until checkout is connected.
Derived by the mapping engine (family · layer · tags · dependency rules). Stored in the catalog as compatibility, dependencies, relations[].
| VERSION | STATE | NOTES |
|---|---|---|
| v2.0.0 2026-10-03 | AVAILABLE | Current release. This package contains: Everything in PromptForge Classic: library, immutable versions with diff, streaming runs with metrics, model and version comparison, test sets, history, code export, backup; Regression check: save a run as the baseline, rerun after an edit, see regressed / fixed / still failing cases plus pass-rate, latency and cost change; download a self-contained HTML report or JUnit XML; Dataset import from CSV or JSONL; columns named like your variables fill the inputs and expected_contains, expected_equals, expected_regex, expected_json_keys or expected_schema columns turn into checks; New checks: JSON Schema (own validator, no dependency), rubric scoring with several criteria and a minimum score, answer within N ms, cost at most $N; Insights tab: pass-rate trend, median and p95 latency, cost, tokens and rubric scores per version and per model, drawn as plain SVG; Snippets with {{> name}}, with cycle detection; live prompt lint (unknown snippet, empty user prompt, unfenced input, absolute rules, no system prompt and more). |
| → EDF 0.2 | Roadmap | Target window for the next stage. |
Verified with 64 automated tests (libraries, the proxy, and the CI runner run as a real process), a clean-room install from the zip (npm ci, tests, build) and a headless-browser pass over every new screen on the offline Mock model. The Anthropic, OpenAI and Gemini calls follow their public API formats and are tested against fake servers, not run with real keys here, so try your own key first. Rubric scoring and the LLM judge depend on how well the model you choose follows the scoring format: read a few runs before you trust a gate. The app ships no price list (enter your own prices to see cost and to use the budget). Single user, no accounts, sync or collaboration; data lives in one browser profile (use the JSON backup or the Markdown export). The demo on this site runs the Mock model and keeps its data in your browser only. Requires Node 20.19+ for the tooling.
Standard (personal or commercial use in one end project or your own business) · Extended ×3 (SaaS, unlimited client projects, finished products only). Agency licence comes later.
Personal: use for yourself on your own devices and servers. Commercial: use for one client or company. Extended: unlimited clients and deployments, including your whole team. No resale, redistribution or hosting as a service for third parties.
Activated by 1.0 (download) and 2.0 (one-click deploy). Download will be a 10-minute signed URL from the R2 vault — never a public /files/product.zip.



