95% of agent skills missing activation directives, GitHub scan reveals
rseroter · x · 2026-08-03
The author ran a static scan on 200 public SKILL.md files from GitHub using Tripwire's lint engine, revealing that 95% of valid skills are missing the crucial "Use when..." activation directive.
- Weak Routing Signals: A skill's description dictates when an agent loads it. Most authors describe what the skill is (e.g., "a helper for writing tests") but omit specific activation cases (e.g., "use when the user adds tests or views a failing test").
- Other Common Failures: 6 skills contained invalid YAML frontmatter; 20 incorrectly embedded code fences in the description; 13 left TODO placeholder text; and 11 failed to use kebab-case naming conventions.
- Context: The Agent Skills specification opened in December 2025 and is now supported by 30+ tools including Codex CLI and Cursor. Despite hosting around 90,000 skills, the routing field remains a freeform string with no strict enforcement.
More from coding & agent
- Mu: An Open-Source Toolkit for Building AI Agents — No-Cream3565 · 2026-08-04
- Self-Improving Agents Optimize vLLM, Boosting DeepSeek Throughput by 16% — yisongyue · 2026-08-04
- AI Agent Integration Woes: 'Native Integrations' Often Just DIY Webhooks — Hvan_7 · 2026-08-04
- Extreme Vibe Coding: Dev Ditches Databases Entirely for localStorage — jonathan_wilke · 2026-08-04
- MiniMax H3 Open-Weight Tested: 8s Video Gen in 15 Mins on RTX 5090 — DavidmComfort · 2026-08-04
- Hackathon Project: Porting go-diff to Rust for Better Performance — cneuralnetwork · 2026-08-04