ffprobe for documents

Know what a file is before you open it.

The same Rust probe engine that powers the CLI, compiled to WebAssembly and running in this tab. Pick a document, choose the facts you want, and read back structured evidence with confidence and measured I/O cost.

Document bytes: never uploaded
Output contract: schema v2 JSON
Rendering, OCR, macros: never executed
Install: npm i @deckflow/deckprobe

Choose a document

01 / input

Sample documents

Open a file from this machine pdf · office · legacy office · modern iwork

— No document selected Pick a sample or open a local file

Samples ship with this repository. Either way the bytes are read locally and handed straight to WASM.

Shape the request

02 / request

Execution path

Target preset

Minimum confidence Raising this can force a costlier exact path.

Engine telemetry Adds elapsed_ms and I/O counters to the envelope.

Advanced request

Target selectors

Probe level

Allow piggyback Let the planner collect extra targets from paths it already reads.

Budget blank uses the engine default

max physical bytes max expanded bytes max archive entries timeout ms

sdk call

Waiting for a document.

Probe result

no run yet

03 / result

No probe yet

Pick a sample on the left, keep the default @summary preset, and run the probe. Identity comes back first, then the rest of the evidence.

Evidence appears after a probe.

filter

targetvaluestatus conf.evidence path

No rows.

The engine's own capability report.

format

Supported formats & report schema

The exact schema-v2 envelope the SDK returned.

Where it stops

This is not a viewer, converter, or editor.

01 No page or slide rendering, no visual previews, no text extraction for display, no conversion.
02 Evidence covers structure, metadata, security, and resource signals only. A signature being present is not a claim that it is valid.
03 PDF, Microsoft Office, Legacy Office, and modern IWA iWork are supported. .xlsb resolves identity only.