agent-skills-state-2026
Monthly study of public agent skill, plugin and MCP repositories: dataset, method, hand-checked precision and scripts.
What it is
Muhammad Basit Ali, 13 October 2026. Dataset, method and scripts are in this repository. Data and text: CC BY 4.0. Code: MIT.
This is the first edition. The scan is re-run every month and each month's data is added alongside; see Updated monthly.
Agent skills, plugin marketplaces and MCP servers are now installed the way browser extensions once were: someone finds a repository, copies a line into a config file and the agent gains new instructions, new shell access or a new process that starts with every session. We wanted to know what that supply looks like on the ground, measured rather than guessed. So we took a random sample of 2,999 public GitHub repositories that ship a SKILL.md, a plugin or marketplace manifest, or an MCP server.json, scanned them with open-source tools, and hand-checked the scanners before trusting them.
Links
Releases
Posts about it
- Do MCP servers say which tools are safe?
I read the tool registrations of 304 public MCP servers. Six in ten mark none of their tools as read-only or destructive, and only one delete-or-send tool in five says it is destructive. - The state of public agent skills and MCP servers
A random sample of 2,999 public repositories that ship agent skills, plugins or MCP servers, scanned and hand-checked. Five findings, with denominators, and what to check before you install one.
More in Data and lists
- ai-agent-incidents: An open dataset of AI agent and LLM security incidents, with a browsable site.
- awesome-agent-security: Curated list of AI agent security tools, papers and datasets.
Topics: agent-security, agent-skills, ai-security, claude-code, claude-code-plugins, dataset, empirical-study, github-code-search, mcp, mcp-servers, model-context-protocol, monthly-index, open-data, reproducible-research, security-research, skill-md, supply-chain-security