The .evidence Format

Using Claude Code, Cursor, or another coding agent? Paste this into your prompt to run cross-browser and real-device tests, debug sessions, and wire up CI on the TestMu AI cloud:

Read https://www.testmuai.com/support/docs/SKILL.md to set up TestMu AI (formerly LambdaTest) cloud testing.

The .evidence pack is not a kane-cli-only file. It is an open, framework-agnostic format with its own specification, validator, and Apache-2.0 licensed tooling, published at github.com/LambdaTest/evidence-cli. One shape, whatever made it: a browser agent, a Playwright suite, a Jest run, or an API check. A pack is readable by a CI dashboard, an auditor, or a human without knowing the framework that wrote it.

Who writes what

This is the part worth understanding, because it explains what the format guarantees and what it does not.

Part of the pack	Written by
`run.yaml`, each `result.yaml`, the test definition, the per-step folders and screenshots, and the console, network and runner logs	kane-cli
The derived totals, each definition hash, the run-level failure index, and the seal	`evidence finalize`

The tooling in evidence-cli captures nothing. It drives no browser and executes no tests. It validates a pack, derives what can be derived, and seals it. A screenshot is in your pack because kane-cli put it there. That is the point of the split: the container is open, so anything can produce a pack, and any tool can read one.

Profiles

A profile is a rung on one contract. L1 adds requirements to L0 and never rewrites it.

Profile	Requires
L0	The minimal core: a top-level `run.yaml`, and per test a definition file plus a `result.yaml`.
L1	All of L0, plus, once the run is finalized, the captured artifact layer: declared logs, the `steps/` layer, a `coverage/` directory, and the run-level failure index. Video is optional. Before finalize, L1 checks only the shape of what is already there.

The definition file is opaque. The format references and hashes it, and never parses it. It can be a test.md from kane-cli, a login.spec.ts from Playwright, or anything else.

The contract version and the profile are different axes. The version, currently 0.1, is the meaning of the fields. Adding a profile is additive and never changes it. Only a breaking change to an existing meaning bumps the version.

What sealing does and does not give you

finalize rolls up the totals, writes each definition’s content hash, sets the run to finalized, and seals the directory into the zip. The seal replaces the live directory in place, atomically, so a complete copy exists at every instant. The recorded definition hash is an integrity check on the thing that was tested: validate fails a finalized pack whose definition no longer hashes to the value recorded at seal time. Packs are not signed, so this says nothing about who produced the pack.

Using the format directly

If you want to produce or read packs outside kane-cli, the tooling is on npm:

npm install -g @testmuai/evidence-cli

evidence validate my-run.evidence --profile L0
evidence finalize my-run.evidence/

The standalone CLI ships validate, finalize, and merge. Exit codes differ by command: validate returns 0 valid, 1 invalid, 2 usage error, and merge returns 0 merged, 1 policy abort, 2 usage error. It reads its config from ~/.testmuai/evidence/config.json by default, overridable with --config or the EVIDENCE_CONFIG environment variable, and the active profile resolves from the --profile flag, then the config, then the built-in default of L0. It is also a library, so validate and finalize can be called in process:

import { validate, finalize } from "@testmuai/evidence-cli";

const report = await validate("my-run.evidence", { profile: "L1" });
if (!report.valid) {
  for (const d of report.diagnostics) {
    console.error(`${d.severity} ${d.location}: ${d.message} [${d.code}]`);
  }
}

The standalone evidence CLI defaults to the L0 profile, while kane-cli evidence validate defaults to L1.

Next steps

Pack structure — the layout in full.
Validating packs — the checks the validator runs.

​Who writes what

​Profiles

​What sealing does and does not give you

​Using the format directly

​Next steps

Who writes what

Profiles

What sealing does and does not give you

Using the format directly

Next steps