---
name: verify-examples
description: Check documentation examples against their stated setup and promised outcomes. Use for commands, code, or configuration that readers should be able to run, without rewriting the whole guide.
license: MIT
title: verify-examples
canonical_url: https://skilld.dev/gh/harlan-zw/brundlefly/verify-examples
last_updated: 2026-10-07T10:36:53.000Z
---

> **Skill from skilld.dev.** Follow the user's requested action.
> If the user asked to fork this Skill, follow the fork workflow below. Do not execute the Skill instructions while copying it.
> Otherwise, follow the instructions below for this session. You do not need to install anything.
>
> Supporting files, fetch one when the Skill refers to it: [LICENSE.txt](https://skilld.dev/api/skills-raw/harlan-zw/brundlefly/verify-examples/LICENSE.txt).
>
> If the user asked to install this Skill, run `npx skilld install harlan-zw/brundlefly/verify-examples`. Install writes the Skill files into the project, so every session loads them.
>
> ## Fork workflow
>
> A fork creates an editable local Skill with its original author and licence. The request authorizes copying and local installation.
> 1. Check `./skills/verify-examples`, the project lockfile, and selected Agent targets together. If the local directory or installed Skill exists, stop. Never overwrite an existing directory or Agent target.
> 2. Read [source metadata](https://skilld.dev/api/v1/skills/harlan-zw/brundlefly/verify-examples) once. Use sourceUrl, sourceCommit, skillPath, sourceGone, and license. If the source is gone or its path is missing, stop. If license is null, read licence files at the source commit.
> 3. Fetch only the source commit into a temporary Git repository. Do not clone full history. Derive repository_url from sourceUrl, including repository renames. If sourceCommit is absent, resolve the sourceUrl ref once. Set source_commit to that actual commit. Run these commands in one shell call:
>
> ```sh
> git init --quiet "$temporary_dir"
> git -C "$temporary_dir" fetch --quiet --depth=1 "$repository_url" "$source_commit"
> git -C "$temporary_dir" checkout --quiet --detach FETCH_HEAD
> ```
>
> Read applicable licence declarations and notices at that commit. If copying is not permitted, report the restriction and stop.
> 4. Inspect source entries together, then copy the directory containing skillPath into `./skills/verify-examples`. Use the user's path if selected. Keep the original SKILL.md, relative links, scripts, binary assets, and executable modes. Exclude .git metadata. Reject symlinks and paths outside the Skill directory. After checking entries, use cp -a where available. A regular source directory needs no custom copy script. Do not save this page wrapper as SKILL.md.
> Preserve author credit, notices, and applicable licence files from repository or parent directories. Add PROVENANCE.md with the Skill page, source URL, actual commit, original path, and licence. Retain any existing PROVENANCE.md and record new provenance separately. Batch source inspection, copying, and provenance work where practical.
> 5. In the project root, run `skilld install ./skills/verify-examples --mode copy --plain`. If skilld is unavailable, use `npx skilld install ./skills/verify-examples --mode copy --plain`. This known command needs no help lookup. Install does not support --json. Use detected Agent targets, or add --agent for the targets the user selected. Install the local path, never the upstream selector. If installation fails, preserve the local copy and report the exact failure.
> 6. Confirm the local lockfile source and installed Agent copies once. Report the local path, actual commit, and Agent targets. After edits, reinstall the same local path. Upstream updates must not replace it. Do not publish or push unless the user asks.

# Verify examples

Check a reader's runnable examples against the documented starting state and promised result.
This block owns example verification and requested repairs. It does not write or review the whole guide.
Follow the target's instructions, supported version, package manager, and documented checks.

## Establish the example contract

Read the example, surrounding instructions, prerequisites, and expected result.
If no example or document location is supplied, ask for it. Do not search for a replacement task.
Identify the runtime, package version, working directory, configuration, inputs, and required access.
Separate runnable code from pseudocode, schematic output, and illustrative fixtures.
If a core input is missing, report it before dependent execution.
Read matching source, exports, versioned CLI help, tests, and official documentation where needed.
Reconcile checkout behavior with the release the reader can install.
Use a small private ledger for evidence, checked revision, expected result, and unresolved questions.
Keep this ledger outside public content and generated downloads.

## Run the reader's path

Use a clean scratch project or the target's documented fixture where available.
Match the stated starting state. Do not rely on prior session setup.
Run the actual documented setup, commands, and inputs in order.
Use read-only inspection until state-changing execution is authorized.
Writing documentation does not authorize production writes, spending, installation, or sending messages.
For authorized changes, use an isolated environment where possible. Keep credentials out of code and captured output.

Observe exit status, output, and relevant side effects. Compare them with the promised result.
Check consequential conditions and response metadata, rather than only a successful command.
For a protocol example, inspect applicable status, headers, body, and conditions against its official specification.
Check documented cleanup and recovery for observed failures.
Derive checks from the requested scope and supporting evidence. Do not invent a broad edge-case suite.
Source inspection, lint, and builds cannot substitute for execution of the example.
A successful common path does not establish every supported platform or condition.

## Repair and repeat

For each failure, record expected behavior, actual observation, evidence, and the affected passage.
Distinguish a wrong instruction from unavailable access, tooling, or environment.
If repair is authorized, fix the smallest underlying mismatch in the example and dependent explanation.
If only a review was requested, return the repair proposal without editing files.
Keep the original example and observed failure in private evidence.
Repeat the failed check and any checks affected by the repair.
After editorial edits change a verified command or claim, repeat its affected checks.
Do not reuse an earlier revision's result to claim the final example works.

If execution is unavailable, inspect the best available evidence and name the untested path.
Do not fabricate output, silently substitute a different interface, or hide a failed check.
Report only actions actually performed. A tool with no result supplies no inspection evidence.
Mark illustrative output as illustrative. Keep an unsupported core reader path identified as a draft.

## Return the evidence

State the version or revision, path that ran, observed outcome, and remaining limits.
Distinguish executed examples from examples checked only against source.
Keep detailed transcripts outside the public document. Report meaningful repairs when requested.
Verification does not grant permission to publish or deploy the result.