vzt-flow: Skill for Claude Code

.claude/skills/verify-dictation/SKILL.md

verify-dictation is a skill for Claude Code from vonzelle-vzt/vzt-flow. It costs 53 tokens per session (1,694 once invoked), scanned A, original, MIT.

A verification checklist for VZT Flow, a dictation app whose promise is that holding a key and speaking places a transcript at the cursor. It checks the running build and tests the complete audio-to-text path.

In plain words
What is it for?
Use it before releases, after dictation-pipeline changes, or when investigating hotkey and transcription problems.
Why use it?
It prevents declaring a dictation change successful just because the code compiles. The checklist confirms that the tested application is the intended version and that the real workflow works.

Skill for Claude Code

Written for Claude Code: installed under .claude/. Also seen: mentions CLAUDE.md.

This is vonzelle-vzt/vzt-flow's own configuration. It tells Claude Code how to work on vzt-flow itself, so it is not a mod to install elsewhere. Copy it as a starting point and replace the rules that are about this project. Everything vzt-flow configures →

Needs its repository: it runs a file that does not travel with it, so clone the repository first. The line is ./target/release/flow status | grep version # the RUNNING daemon's version.

Reuse

Borrowing it

Nothing to install: this file belongs to vonzelle-vzt/vzt-flow. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file
curl -O https://raw.githubusercontent.com/vonzelle-vzt/vzt-flow/main/.claude/skills/verify-dictation/SKILL.md
Clone the repo
git clone --depth 1 https://github.com/vonzelle-vzt/vzt-flow

Made for: Claude Code.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for verify-dictation

README.md
[![agentmods](https://agentmods.dev/badge/skills/vonzelle-vzt/vzt-flow/verify-dictation/github.svg)](https://agentmods.dev/skills/vonzelle-vzt/vzt-flow/verify-dictation)
Your own site
<a href="https://agentmods.dev/skills/vonzelle-vzt/vzt-flow/verify-dictation"><img src="https://agentmods.dev/badge/skills/vonzelle-vzt/vzt-flow/verify-dictation/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for verify-dictation

Your own site · 80×15
<a href="https://agentmods.dev/skills/vonzelle-vzt/vzt-flow/verify-dictation"><img src="https://agentmods.dev/badge/skills/vonzelle-vzt/vzt-flow/verify-dictation.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 53 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,694 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00053 $0.01694
Opus 5 $0.00026 $0.00847
Sonnet 5 $0.00011 $0.00339
Haiku 4.5 $0.00005 $0.00169

Measured 11d ago against content hash 7d36baf03ab4, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-11, from the pricing page.

Security

Grade A, and why

verify-dictation scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 11d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.claude/skills/verify-dictation/SKILL.md · 149 lines

How it starts

The opening of the file, as written. The whole thing — 149 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Verify dictation

VZT Flow's core promise is "hold a key, talk, transcript lands wherever your cursor is." This checklist proves that promise end to end, with real audio and real numbers — not "the code compiles so it probably works."

0. Prerequisites

First, confirm which build you are about to verify. Verifying the wrong binary is worse than not verifying — it produces a confident green result about an app nobody is running. This has happened: a "hotkey is broken" investigation ran for an hour against a healthy process before anyone checked that /Applications/VZT Flow.app was an ad-hoc v0.3.2 dev build, two versions behind a crash fix it was still carrying (CLAUDE.md gotcha (j)).

./target/release/flow status | grep version    # the RUNNING daemon's version
grep -m1 '^version' Cargo.toml                 # what this tree is
codesign -dv --verbose=2 "/Applications/VZT Flow.app" 2>&1 | grep -E "flags|TeamIdentifier"

Release = flags=0x10000(runtime) + TeamIdentifier=LKHKU5BW73. Local build = flags=0x2(adhoc) + TeamIdentifier=not set. State which one you tested; a report that doesn't name the build under test is not a verification.

source ~/.cargo/env
cargo build --release --workspace
./target/release/flow doctor

flow doctor must show both models present (Parakeet always; Qwen3 cleanup if you're testing clean/polish). If a model is missing:

./target/release/flow models download parakeet-v3
./target/release/flow models download cleanup   # only if testing clean/polish

1. The human hotkey test (do this yourself, not via script)

  1. Launch the desktop app (or relaunch a dev build — see CLAUDE.md gotcha (a): use nohup <dev-binary> &, never open, and never kill the user's running daily-driver instance without relaunching it for them).
  2. Hold Right Option, speak a full sentence, release. Confirm: overlay shows Recording → Transcribing (with a mode badge) → Done, and the transcript pastes into whatever's focused.
  3. Tap Right Option (press+release under 300ms). Confirm hands-free starts, then stops on its own after ~2.5s of silence (or tap again to stop manually).
  4. Hold, start speaking, hit Esc mid-recording. Confirm nothing is transcribed or pasted — the overlay just clears.
  5. Focus a password field, dictate. Confirm the overlay shows "Secure field — transcript on clipboard" and nothing gets typed into the field.

Read the full file on GitHub · 149 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 11d ago First seen · 149 lines · 53 tokens per session scan A 7d36baf03ab4

Subscribe to this mod's changes

verify-dictation is a skill published in the GitHub repository vonzelle-vzt/vzt-flow (5 stars, last pushed yesterday), licensed MIT. It adds 53 tokens to every session and 1,694 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.