agentsclimarketplace

Compile cards

Skill chappyasel/meta-kb/.claude/skills/compile-cards

A self-improving knowledge base about LLM agent infrastructure

Install
npx -y skills add chappyasel/meta-kb --skill compile-cards

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
  • 23 stars23 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Writes reference cards for projects, concepts, or people in the wiki. Projects get 800-1500 words with architecture, benchmarks, limitations, and alternatives. Concepts get 1500-2500 words. People get 150-250 words. Use when writing wiki reference cards for specific entities.

SKILL.md

3.9 KB, as published. Nobody here has run it

Compile Reference Cards

Read config/domain.ts for domain context and taxonomy bucket definitions.

Write reference cards for projects, concepts, or people.

Card Types

TypeWordsOutput path
Project800-1500wiki/projects/<slug>.md
Concept1500-2500wiki/concepts/<slug>.md
Person150-250wiki/people/<slug>.md

Source Prioritization

  1. Deep sources first: raw/deep/repos/ and raw/deep/papers/
  2. Then shallow sources by relevance score
  3. Use top 5 sources per card

Abstract

Every card starts with <abstract></abstract> tags before any headings. One sentence, under 200 chars. State what it does plus the key differentiator. Not a generic description.

Project Card (800-1500 words)

Required Sections

  1. What It Does -- Concrete description. Not marketing copy. What problem it solves and the architectural approach that makes it different.

  2. Architecture -- How it works. Name specific files, modules, algorithms from deep sources. Enough detail that a reader understands HOW, not just WHAT.

  3. Core Mechanism -- The key algorithm or pattern with implementation detail from deep sources.

  4. Key Numbers -- Benchmarks with credibility assessment: self-reported, peer-reviewed, or verified in code. If README claims cannot be verified, say so.

  5. Strengths -- Each backed by evidence from a deep source.

  6. Critical Limitations -- At minimum:

    • One concrete failure mode (what breaks, not just "has limitations")
    • One unspoken infrastructure assumption
    • One gap in documentation or validation
  7. When NOT to Use It -- Operational conditions where this is the wrong choice.

  8. Unresolved Questions -- 3-5 specific unknowns: governance, cost at scale, conflict resolution, architectural growth limits. What the documentation does not explain.

  9. Alternatives -- Link to other project cards with guidance on when to pick each.

Concept Card (1500-2500 words)

Required Sections

  1. Definition -- What it is. Precise, no jargon layering.

  2. Why It Matters -- The specific problem it solves for practitioners.

  3. How It Works -- Implementation details from deep sources. Name projects that implement it and how their approaches differ.

  4. Who Implements It -- Link to project cards with each project's specific approach noted.

  5. Failure Modes -- When does this concept break down? Concrete scenarios.

  6. Open Questions -- What remains unresolved.

Person Card (150-250 words)

Brief profile: who they are, key contributions to the field, notable work. Link to their projects and publications in the wiki.

Citation Rules

  • Cite actual source paths: [Source](../raw/type/file.md) or [Source](../raw/deep/type/file.md)
  • Link to related wiki pages: [Mem0](../projects/mem0.md), [context graphs](../concepts/context-graphs.md)
  • Star counts on first mention for projects
  • Assess benchmark credibility when citing numbers
  • When sources disagree: "Source conflict: [Source A] claims X, while [Source B] claims Y."

Frontmatter

---
entity_id: <slug>
type: project | concept | person
bucket: <bucket-id>
abstract: "<1 sentence, under 200 chars>"
sources:
  - raw/type/file.md
  - raw/deep/type/file.md
related:
  - projects/slug.md
  - concepts/slug.md
last_compiled: "YYYY-MM-DD"
---

Writing Quality

Read and follow all rules in stop-slop/SKILL.md. No filler, no passive voice, no em dashes, no AI tells.

Templates

See references/template.md for the full structural templates for each card type.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.