latex-paper-en

From the files and frontmatter at commit bfa0a6e. This is how much a skill can do, not whether it is risky.

• Downloads software

  It fetches programs or packages from the internet, or from a package source other than the default.

  references/CITATION_VERIFICATION.md:88 (opens in a new tab)

• Installs more skills

  It tells the agent to install other skills, plugins or tool servers.

  references/CITATION_VERIFICATION.md:88 (opens in a new tab)

• Runs shell commands

  Its instructions or files give the agent commands to run on your computer.

  references/CITATION_VERIFICATION.md:88 (opens in a new tab)

• Reaches the network

  It sends or fetches data over the internet, for example with curl, an HTTP library or a package install.

  references/CITATION_VERIFICATION.md:88 (opens in a new tab)

• Names web addresses

  It mentions outside web addresses the agent may visit. We list them but don't follow them.

  Outside links we did not follow

• Bundles scripts

  The folder holds code files the agent may run.

  scripts/check_references.py (opens in a new tab) · scripts/deai_batch.py (opens in a new tab) · scripts/extract_prose.py (opens in a new tab) and 1 more

• Pre-approves tools

  Its allowed-tools field lists tools the agent may use without asking you first, in agents that honour it.

  SKILL.md:8 (opens in a new tab)

Pattern check: 1 item to review · commit bfa0a6e · 3 Nov 2026

• R27 · HighInstalls more skills or tool servers without asking

  references/CITATION_VERIFICATION.md:88 (opens in a new tab)

  claude mcp add exa -- npx -y mcp-remote "https://mcp.exa.ai/mcp"

We checked every text file in this skill, at commit bfa0a6e on 3 Nov 2026, against 45 published patterns such as download-and-run commands, hidden text and credential access. We don't run the skill, we don't read the web pages it points to, and we can't judge everything its instructions might lead an agent to do. In published tests, pattern checks like this one missed most malicious skills. Read the skill before you install it — especially anything it asks you to download or run.

How the check worksReport an incorrect finding

Each agent reads SKILL.md a little differently. We check the file against each one's published rules; we don't run the skill in any agent.

• Agent Skills specPasses with warnings

  The open format from agentskills.io, held to the strictest version of each rule among the agents we check.

  • Warning

    The frontmatter uses [ ] or { } values. Most agents accept them, but the spec's validator (skills-ref) rejects the file. F4

    Fix: Write lists as "- item" lines and maps as indented key: value lines, without anchors or tags.

  • Warning

    Fields outside the spec: argument-hint (read by Claude Code and VS Code). Uploads to the Claude apps and API, and the spec's validator, reject the file; most agents ignore these fields. O1

    Fix: Delete them for a portable skill, or keep them if you only target the agents that read them.

  • Warning

    metadata values should be text; tags (a list) is not. The spec's validator turns them into text; other readers may not. O4

    Fix: Quote each value, for example version: "1.0".

  • Note

    Anthropic's packaging script leaves these out: evals/. F8

    Fix: Delete them, or list them in .gitignore.

  • Note

    There is no license field, so readers can't tell whether they may reuse the skill. O2

    Fix: Add license: with an SPDX name such as MIT or Apache-2.0, or the name of a bundled licence file.

  • Note

    Reference files that SKILL.md doesn't link: examples/compile-and-bibliography.md, examples/experiment-review.md, examples/figures-and-title.md, examples/grammar-and-logic.md, examples/translation-and-deai.md and 16 more. Anthropic suggests linking every reference file directly from SKILL.md. B4

    Fix: Link each one from SKILL.md and say when to read it.

  • Note

    Reference files over 100 lines with no table of contents at the top: references/CITATION_STYLES.md (114 lines), references/JOURNAL_ADAPTATION_WORKFLOW.md (109 lines), references/TABLE_GUIDE.md (116 lines), references/modules/LOGIC.md (142 lines) and references/modules/TITLE.md (209 lines). B5

    Fix: Add a short list of the sections at the top of each.

• Claude CodePasses

  Claude Code reads its own extra fields, ignores unknown ones, and still loads a skill whose frontmatter is broken, without its fields.

  • Note

    Anthropic's packaging script leaves these out: evals/. F8

    Fix: Delete them, or list them in .gitignore.

  • Note

    Reference files that SKILL.md doesn't link: examples/compile-and-bibliography.md, examples/experiment-review.md, examples/figures-and-title.md, examples/grammar-and-logic.md, examples/translation-and-deai.md and 16 more. Anthropic suggests linking every reference file directly from SKILL.md. B4

    Fix: Link each one from SKILL.md and say when to read it.

  • Note

    Reference files over 100 lines with no table of contents at the top: references/CITATION_STYLES.md (114 lines), references/JOURNAL_ADAPTATION_WORKFLOW.md (109 lines), references/TABLE_GUIDE.md (116 lines), references/modules/LOGIC.md (142 lines) and references/modules/TITLE.md (209 lines). B5

    Fix: Add a short list of the sections at the top of each.

• Claude apps & API uploadFails

  Uploading to claude.ai or the Claude API, or packaging with Anthropic's script, accepts only the six spec fields and stops on anything else.

  • Error

    Upload and packaging fail on fields outside the spec: argument-hint. O1

    Fix: Delete them, and keep your own data under metadata as text values.

  • Note

    Anthropic's packaging script leaves these out: evals/. F8

    Fix: Delete them, or list them in .gitignore.

  • Note

    The description is 980 characters. The claude.ai Help Center says 200 at most, though Anthropic's API docs and validator allow 1,024. D4

    Fix: If the claude.ai upload refuses it, trim the description to 200 characters.

  • Note

    metadata values should be text; tags (a list) is not. The spec defines metadata as text to text. O4

    Fix: Quote each value, for example version: "1.0".

  • Note

    Reference files that SKILL.md doesn't link: examples/compile-and-bibliography.md, examples/experiment-review.md, examples/figures-and-title.md, examples/grammar-and-logic.md, examples/translation-and-deai.md and 16 more. Anthropic suggests linking every reference file directly from SKILL.md. B4

    Fix: Link each one from SKILL.md and say when to read it.

  • Note

    Reference files over 100 lines with no table of contents at the top: references/CITATION_STYLES.md (114 lines), references/JOURNAL_ADAPTATION_WORKFLOW.md (109 lines), references/TABLE_GUIDE.md (116 lines), references/modules/LOGIC.md (142 lines) and references/modules/TITLE.md (209 lines). B5

    Fix: Add a short list of the sections at the top of each.

• OpenAI (Codex & API)Passes with warnings

  Codex finds SKILL.md by its exact name and needs a description; OpenAI's Skills API checks the frontmatter against the spec and caps uploads.

  • Warning

    Fields outside the spec: argument-hint. Codex ignores them; OpenAI doesn't say whether its upload accepts them. O1

    Fix: Delete them, or move your own data under metadata as text values.

  • Warning

    metadata values should be text; tags (a list) is not. OpenAI's upload follows the spec's text-to-text rule. O4

    Fix: Quote each value, for example version: "1.0".

  • Note

    Anthropic's packaging script leaves these out: evals/. F8

    Fix: Delete them, or list them in .gitignore.

• VS Code & CopilotPasses

  VS Code reads skills one folder below .github/skills, .claude/skills or .agents/skills and falls back to the folder name when the name is missing or differs; GitHub's Copilot docs require the exact name SKILL.md.

  • Note

    Anthropic's packaging script leaves these out: evals/. F8

    Fix: Delete them, or list them in .gitignore.

64 files