Provenance Record

Clean News-category piece at 862 words (well within 400-1200 range). Structured with Overview, What Are Agent Skills, What gh skill Does, Security Caveats, Why It Matters, and What We Don't Know sections. Every factual claim is attributed to a specific source with inline links. Technical detail (CLI version, subcommands, flags, installation syntax, provenance metadata) is accurate and well-organized.

{"https://github.blog/changelog/2026-04-16-manage-agent-skills-with-github-cli/":"VERIFIED — Primary source. Confirms April 16, 2026 release, GitHub CLI v2.90.0 requirement, public preview status, six supported agents (Copilot, Claude Code, Cursor, Codex, Gemini CLI, Antigravity), the five subcommands (install, search, preview, update, publish), the flags (--pin, --agent, --scope), installation syntax (OWNER/REPOSITORY SKILL with @tag and @commitSHA), tree SHA / git ref provenance metadata in SKILL.md frontmatter, the 'same guarantees you expect from package managers' framing, and the prompt-injection warning about skills containing 'prompt injections, hidden instructions, or malicious scripts'. All quoted and paraphrased content is verbatim or faithful.","https://venturebeat.com/technology/anthropic-launches-enterprise-agent-skills-and-opens-the-standard":"VERIFIED — Dec 18, 2025 by Michael Nuñez. Exact title 'Anthropic launches enterprise Agent Skills and opens the standard, challenging OpenAI in workplace AI' confirms the article's 'challenging OpenAI in enterprise workflow AI' framing. Partner skills list (Atlassian, Figma, Canva, Stripe, Notion, Zapier) matches exactly. Confirms 'independent open standard with a specification and reference SDK available at https://agentskills.io'.","https://www.infoq.com/news/2026/01/claude-cowork/":"VERIFIED — Published January 20, 2026. Confirms all ecosystem-adoption claims: Microsoft integrated Skills into VS Code and GitHub Copilot; OpenAI adopted the format for Codex CLI in December 2025; Cursor, Goose, Amp, OpenCode, and Letta all implemented compatible skill loading. No misattribution.","https://www.anthropic.com/engineering/equipping-agents-for-the-real-world-with-agent-skills":"VERIFIED — Anthropic's engineering post introducing Agent Skills on October 16, 2025. Confirms the 'directory containing a SKILL.md file' structure with YAML frontmatter, progressive-disclosure behavior (agents read only name and short description to decide applicability, loading full instructions only when a task triggers the skill), and the context-window-leaning rationale. Article's paraphrase is faithful."}

All claims traced to cited sources with no hallucinations. The direct quote ('may contain prompt injections, hidden instructions, or malicious scripts') matches the GitHub changelog verbatim. The 'same guarantees you expect from package managers' phrase is an accurate paraphrase of the changelog framing. Dates (Oct 2025 Anthropic intro, Dec 2025 open standard, April 16 2026 gh skill launch) all align with primary sources.

High-quality, well-sourced News piece on a legitimately significant developer-tools story. Every claim is anchored to a verified source. The analytical framing in 'Why It Matters' is supported and not speculative. Approve for publication.