← Explore all skills

EVIDENCE AND EVALUATION / Undominated.ai

Undominated · Keep every ladder rung

Refuse a published context-tier ladder whose rung count differs from the source. Dropping a rung flattens the ladder. Adding a rung invents one. Prices and boundaries are not read.

“Refuse a published context-tier ladder whose rung count differs from the source. Dropping a rung flattens the ladder. Adding a rung invents one. Prices and boundaries are not read.”

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: Compare the two counts. Do not read a price, a token boundary, or a multiple.

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-ladder-not-flattened/SKILL.md
  • skills/undominated-ladder-not-flattened/scripts/check.py
  • skills/undominated-ladder-not-flattened/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.
  • One input rule the file states is: `sourceRungs` and `publishedRungs` are integers of at least 1. A boolean is not an integer.
  • One condition that blocks a pass is: a published context-tier ladder has the same rung count as its source; prices and boundaries are not read.

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-ladder-not-flattened/bundle.zip and extract it into a new directory.
  2. Place the extracted directory at .agents/skills/undominated-ladder-not-flattened/ 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-ladder-not-flattened
description: Refuse a published context-tier ladder whose rung count differs from the source. Dropping a rung flattens the ladder. Adding a rung invents one. Prices and boundaries are not read.
license: MIT
metadata:
  author: Undominated.ai
  version: "1.0.0"
---

# Keep every ladder rung

Use when a context-tier ladder is about to be published. The source count and the published count have to be the same integer. A shorter published ladder is the failure that flattened a tier boundary. A longer one invents a rung.

1. Compare the two counts. Do not read a price, a token boundary, or a multiple.
2. A smaller published count is a review. A larger published count is a different review.
3. Context-tier still owns the multiple. Rung-select still owns which rung contains a length. This checker does neither.

## Input contract

`sourceRungs` and `publishedRungs` are integers of at least 1. A boolean is not an integer.

## Deliverable and limits

Return both counts and whether a rung was dropped or added. A pass means the counts match. It does not say the boundaries are right, and it does not price a request.

`examples/synthetic.json` passes on three source rungs published as three. The other examples are a dropped rung, an added rung, and a zero count.

## Run the local check

Resolve paths relative to this skill directory:

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

The examples are **synthetic**, not a vendor quote or a live page. Python 3.10+; standard library only. The script reads one explicit UTF-8 JSON file, prints JSON and makes no network requests or file writes.

Exit codes: `0` consistency check passed, `1` review required, `2` invalid input. CLI argument and malformed-input errors return structured JSON; `--help` displays usage text. Unknown or duplicate JSON fields, non-JSON constants and invalid types are rejected. Files are limited to 1 MiB.

This source-only skill is **not included in undominated-check@0.4.0**. Keep `SKILL.md`, the checker, examples and MIT licence together. A pass validates the bounded supplied-input contract, not production suitability or permission to publish. The skill does not authorize external actions.

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 f23e98818b4f7d0afbeb73c155b876279a72640ec2b700a452862b9cf8621eff

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-ladder-not-flattened/definition.md

    Supports: Workflow and input requirements, Stated limitations

  2. MIT licence ↗Checked

    https://undominated.ai/resources/skills/undominated-ladder-not-flattened/LICENSE.txt

    Supports: Redistribution terms and attribution

  3. Complete source bundle ↗Checked

    https://undominated.ai/resources/skills/undominated-ladder-not-flattened/bundle.zip

    Supports: Full local source, supporting files and integrity manifest

Evidence & Ask