◆ Kiroween 2026 / Frankenstein / MIT

The agent writes.
The evidence decides.

PRAMAN runs authored browser and command checks against real software, applies strict proof gates, and returns honest, evidence-backed verdicts. It is not another coding agent.

12 / 12 documented outcomes · CI green on Ubuntu + macOS · MIT · zero known vulnerabilities
◆ Proof pipeline · declared checks Evidence, not assertion
01 — DECLARE authored scenario · command 02 — EXECUTE real browser · trusted process 03 — PROVE checks · evidence · identity PASSED 12 / 12 outcomes REPLAY MATCHED PROOF NO MUTATION
◆ declarescenario.json · commands.json
◆ executeChromium · trusted process · 2–10 repeats
◆ provestrict gate · indexed evidence · replay
proof.jsonREPORT.mdfeedback.jsonreplay matched
The problem

A green build is not evidence. Evidence is evidence.

AI agents now write code faster than anyone can inspect it. A passing test suite proves the tests pass; it does not prove the feature works when a browser, a process, or an API is actually used. PRAMAN closes that gap.

What a passing run proves

  • Declared behavior was observed in a real browser or process
  • The same requirements passed in fresh repeated sessions
  • Checks, screenshots and outputs were collected and retained
  • The execution identity was stable across the proof

What it does not prove

  • Whole-application correctness
  • That unstated requirements are met
  • Permission to skip human review
How it works

Three steps. No self-reported success.

01 — AUTHOR

Declare the checks

The agent or author writes a scenario or command declaration: steps, checks, expected values, repeats and limits.

02 — EXECUTE

Run the real software

PRAMAN drives headless Chromium or trusted local processes, 2–10 fresh repeats, and records what actually happened.

03 — PROVE

Return an honest verdict

Passed requires every declared check, complete evidence and a stable identity. Missing evidence never turns green.

The loop closes when the next run disagrees less.

Evidence

Claimed, then checked.

376
tests / 43 files
12/12
judge outcomes
14
MCP tools
16
CLI commands
6
walkthroughs archived
Headlamp #8175A real Kubernetes UI crash was reproduced, repaired and re-verified. Five checks across three fresh sessions plus a recording replay. Open the upstream issue.REPRODUCED / FIXED
CIUbuntu and macOS green on the public repository; the Linux job also runs the documented first-use route. View runs.GREEN
Judge routeCodespaces or local: TESTING.md. Expected exit 0 with 12 of 12 CLI and MCP outcomes.RUNNABLE
HonestyUnavailable execution and incomplete evidence return insufficient_evidence. Missing evidence never becomes a pass.ENFORCED
One real repair

The first fix removed the crash. It did not make the feature work.

Headlamp called the Kubernetes object getter on the wrong receiver. Correcting that stopped the exception; the tooltip still did not appear. The link wrappers also had to forward the DOM reference. PRAMAN exposed the difference between “no error” and “works”.

Before — throws
tooltipText = (props as LinkObjectProps).getName()

Real Headlamp UI, live Kubernetes namespace: props.getName is not a function.

After — proves
const kubeObject = props.kubeObject; tooltipText = kubeObject.getName()

Same action, real tooltip, five acceptance checks across three fresh sessions and a replay.

Local checkout repair, not an upstream submission. Attribution and limits.

Judge access

Run it before you trust it.

Test build

Public source, real execution, no credentials.

Open the repository in Codespaces, or clone it and run the documented route. The expected result is exit 0 with twelve CLI and MCP outcomes across pass, mismatch and unavailable.

npm ci && npm run build && node examples/first-use/run.mjs

No hosted execution endpoint. PRAMAN runs trusted commands with your authority and has no sandbox, so exposing it publicly would be unsafe.

Limits

What this is. What it is not.

Is

  • Authored browser and command verification with evidence
  • Three explicit modes: acceptance, repair, comparison
  • CLI and stdio MCP for existing coding agents
  • Validated on macOS plus Ubuntu and macOS CI

Is not

  • A sandbox for hostile code
  • Autonomous test design or whole-app correctness
  • A hosted service or published package
  • Independently human-validated yet

Trust the run, not the claim. Current status · Roadmap

Kiro usage

Kiro was used. It did not build the core.

An AI coding assistant built PRAMAN’s original core before 5 November 2026. From 7 November, Kiro CLI agents built PRAMAN’s Kiro integration (a custom agent, a finish gate on the Stop hook, steering and an installable Power) and then fixed and extended PRAMAN itself through thirteen specs in recorded sessions: Ctrl-C cleanup, failure diagnosis, new browser actions, saved starting worlds, network faults, many users at once, recorded outside calls and a one-page report. Each change was shown first on a real failing run and proven with PRAMAN. The write-up gives a session ID for every claim, including what Kiro did not do.

Read the truthful Kiro usage account

Judge it

Evidence before you ship.

The agent writes the code. PRAMAN proves what actually happens when it runs.