← Explore all skills

EVIDENCE AND EVALUATION / Undominated.ai

Undominated · Install receipt

Compare an expected file manifest with a sha256sum receipt and, when asked, the bytes in a directory next to the input.

“Compare an expected file manifest with a sha256sum receipt and, when asked, the bytes in a directory next to the input.”

01 / THE REASONING

Why this made the selection.

  • Pairs a bounded workflow with an offline Python checker and explicitly synthetic example inputs.
  • Created for recurring evidence and comparison failures: missing facts remain unknown and every conclusion keeps its scope.
  • It names a concrete failure mode: It does not mean `npx`, the Skills CLI, or a registry published that directory.

02 / THE REVIEW RECORD

What we actually inspected.

Source review has boundaries.
A clear record is more useful than a “safe” badge.

Material inspected

  • skills/undominated-install-receipt/SKILL.md
  • skills/undominated-install-receipt/scripts/check.py
  • skills/undominated-install-receipt/examples/synthetic.json

Our findings

  • First-party resource authored by Undominated.ai; this is our own verification record, not an independent endorsement.
  • The download preserves the complete source and supporting files. Its manifest hashes identify the exact bytes.
  • The checker names its required fields as: Required: `resourceId` (slug), `sourceUrl` (HTTPS), `observedAt` (ISO date), `bindFiles` (boolean), `expectedFile`.
  • One condition that blocks a pass is: Manifest, receipt, and optional local bytes. This does not prove an npm or Skills CLI install occurred.

Not established by this review

  • Model compliance with these instructions on arbitrary tasks
  • Native integration in every agent host
  • Public GitHub source identity
  • Remote installation through the standard Skills CLI
  • The new npm resource CLI installation

The review applies to the material and revision named here. A newer upstream release can change its behavior.

03 / PUT IT TO WORK

Add a skill to your workflow.

Upstream setup instructions ↗
  1. Download the complete bundle from https://undominated.ai/resources/skills/undominated-install-receipt/bundle.zip and extract it into a new directory.
  2. Place the extracted directory at .agents/skills/undominated-install-receipt/ or your host's documented skills directory. Preserve SKILL.md, scripts, examples and licence files together.
  3. Run the documented synthetic example from the skill directory with Python 3.10 or later before using your own evidence.

Before you start

  • An Agent Skills compatible host
  • Python 3.10 or later for the optional checker

THE COMPLETE REVIEWED DEFINITION

Read it before you reuse it.

Original source bytes, with attribution.
Review the host-specific setup notes above.

---
name: undominated-install-receipt
description: Compare an expected file manifest with a sha256sum receipt and, when asked, the bytes in a directory next to the input.
license: MIT
metadata:
  author: Undominated.ai
  version: "1.0.0"
---

# Install receipt

Use after copying a skill, agent profile, or MCP bundle into a directory you control, before treating that copy as the reviewed revision.

1. Write the expected manifest yourself from the reviewed hashes: each relative path and its SHA-256. Save a `sha256sum` receipt (`64 lowercase hex`, two spaces, relative path) for the copy you are checking.
2. Put both documents next to the input JSON and record their SHA-256. Set `sourceUrl` to the page or commit you believe the bytes came from, and `observedAt` to the day you hashed them.
3. Set `bindFiles` to true and `root` to a relative directory when the bytes themselves are next to the input. The checker hashes only those files. It does not search a repository, follow symlinks, or accept absolute paths.
4. A pass means the manifest, the receipt, and the bound bytes agree. It does not mean `npx`, the Skills CLI, or a registry published that directory.
5. Missing files, extra files, and hash mismatches are review results. Do not delete the extra file from inside this checker. Decide that outside it.

## Run the local check

```sh
python3 scripts/check.py examples/synthetic.json
python3 scripts/check.py /absolute/path/to/your-input.json
```

The bundled fixture is **synthetic**. The script reads the input directory only, makes no network requests, and writes no files. Python 3.10+; no dependencies.

Exit codes: `0` the documents and bound bytes agree; `1` review required; `2` invalid input. Passing is not an installation certificate.

## Input contract

Required: `resourceId` (slug), `sourceUrl` (HTTPS), `observedAt` (ISO date), `bindFiles` (boolean), `expectedFile`, `expectedSha256`, `observedFile`, `observedSha256`. When `bindFiles` is true, `root` is a relative directory. The expected file is JSON `{"files":[{"path","sha256"}]}`. The observed file uses `sha256sum` lines. Paths are portable relative segments: letters, digits, `.`, `_`, `-`.

## Deliverable and limits

Return the resource id, file count, which paths were hashed from disk, and every mismatch. Keep the JSON with the explanation.

The user retains control over what is deleted or published. This skill does not install packages or contact a registry.

The download contains SKILL.md. Extract the whole bundle; the supporting files are required. Inspect the included MANIFEST.json for file hashes.

By Undominated.ai. Exact upstream source ↗ · Licence · Attribution

SHA-256 42d159cb2147561be58157a1ba318f911d57f265fa3f1d24f6c5b5dfdb708186

Read the applicable licence
MIT License

Copyright (c) 2026 Undominated.ai

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.

04 / FOLLOW THE EVIDENCE

The source trail.

Our notes are separate from the original resource.
Check upstream before adopting a new version.

  1. https://undominated.ai/resources/skills/undominated-install-receipt/definition.md

    Supports: Workflow and input requirements, Stated limitations

  2. MIT licence ↗Checked

    https://undominated.ai/resources/skills/undominated-install-receipt/LICENSE.txt

    Supports: Redistribution terms and attribution

  3. Complete source bundle ↗Checked

    https://undominated.ai/resources/skills/undominated-install-receipt/bundle.zip

    Supports: Full local source, supporting files and integrity manifest

Evidence & Ask