Golden PPTX Keyword Categorization
A focused reference skill for testing keyword categorization across structured presentation content.
No malware, credential theft, or external side effects are visible; the artifact is a static presentation extraction and is labeled as generated by Skill Seeker. Points are deducted because the skill does not disclose data-flow boundaries, sensitive-data handling, user confirmation, rollback, or precise source-attribution mechanisms, and its code examples cannot be independently verified from the skill itself.
SKILL.md, index.md, and the three section files broadly agree on slide counts, section counts, and extracted content, supporting basic structural consistency. Points are deducted because the Python example is truncated as “x3...”, quality ratings lack supporting criteria, and there is no diagnostic guidance for malformed input, missing references, or extraction failures; static calibration also limits the score.
The trigger, “testing keyword categorization,” is explicit, and the artifact is suitable for reviewing this presentation's sections, code, tables, and speaker notes. Points are deducted because it is a single PowerPoint test fixture rather than the described documentation/GitHub/PDF conversion skill, and it lacks non-fit boundaries, input/output contracts, semantic trigger limits, and declared Chinese-language support.
The documentation includes purpose, section overview, quick reference, examples, statistics, and reference navigation; repository context supplies an MIT license, version metadata, and test configuration. Points are deducted because the selected skill itself lacks installation notes, versioning, changelog, maintenance ownership, troubleshooting, known limitations, and stable parameter guidance; generation is labeled but provenance and update paths remain incomplete.
The artifact provides a usable navigation and summary layer for an eight-slide presentation, while the reference files preserve basic text, code, tables, and speaker notes. Points are deducted because the main code example is incomplete, several areas contain only headings or statistics, and there is limited evidence of complete coverage or directly usable output for a real user task.
The main document, index, and section references provide static cross-checks for slide counts, sections, and several examples. Points are deducted because there is no executable reproduction record, third-party validation, or dedicated evidence that this specific skill is correct; repository tests and CI mostly validate project-level workflows, not the factual completeness of this artifact.
- This file is a test presentation extraction and should not represent the repository's broader documentation, GitHub, or PDF conversion capabilities.
- The long Python example in SKILL.md is truncated and should not be treated as complete code.
- Source location, version pinning, maintenance ownership, and abnormal-input handling are not clearly specified; manually verify against the original PPTX before relying on it.
What it does & when to use it
golden-pptx-kw is a presentation reference skill designed for keyword-categorization tests. It covers an eight-slide presentation organized into Setup, Api, and Other sections, with material on getting started, architecture, and summary content. The skill exposes slide topics, Bash and Python examples, table summaries, presentation statistics, and reference-file navigation. It is useful for validating retrieval and categorization behavior, but it is not documented as a general PowerPoint authoring or editing tool.
It directs an agent to review presentation sections, key points, slide topics, code examples, tables, statistics, speaker notes, and additional context. The supplied content reports 8 slides, 3 sections, 3 code blocks, 2 images or diagrams, 2 tables, and 2 programming languages. It also points to section_s1-s1.md, section_s2-s2.md, section_s3-s3.md, and references/index.md for navigation.
- A developer testing whether keyword categorization correctly identifies Setup, Api, and Other sections.
- A reviewer checking whether the presentation's getting-started, architecture, and common-error topics are retrieved.
- An evaluator validating retrieval of Bash and Python code examples from presentation content.
- A tester verifying that table summaries such as debug=false and port=8080 are surfaced correctly.
- A team assessing whether an agent can combine slide content with speaker notes and reference-file context.
Pros & cons
- Its purpose is narrowly defined around presentation keyword-categorization testing.
- It combines section structure, slide topics, code, tables, and statistics.
- Reference-file paths make the extracted presentation structure easy to navigate.
- Its frontmatter uses the simple name-and-description form.
- The content is tied to one specific eight-slide presentation.
- No categorization algorithm, test command, or automation script is provided.
- Support for arbitrary PowerPoint files or generated outputs is not documented.
- The first Python example is truncated and cannot serve as a complete implementation reference.
How to install
The skill is located at tests/golden/phase2/pptx_kw/SKILL.md in the repository. Copy the directory containing this file as a golden-pptx-kw skill into the skills directory of an Agent Skills-compatible client; the source does not specify client-specific paths. The full Skill Seekers collection can be installed with pip install skill-seekers, but the README does not document a dedicated installation command for this individual test skill.
How to use
In an Agent Skills-compatible client, ask: "Use golden-pptx-kw to test keyword categorization for this presentation, then review its sections, slide topics, code examples, and table summaries." For deeper navigation, ask the agent to consult references/section_s1-s1.md, references/section_s2-s2.md, references/section_s3-s3.md, or references/index.md.