# Agent Skills (SKILL.md)

98 Hacker News commenters recommended Agent Skills (SKILL.md) and 57 criticized it, November 2025 to November 2026. It was recommended in 11 of 13 months.

**Verdict:** A simple Markdown-based way to give AI agents extra knowledge and tool instructions, loaded only when needed to save context; agents often skip or ignore skills.

## Why people recommend it

- Progressive disclosure saves context (8 comments): "Skills also have a nicer way of working with the context, by default (and in the main web uis), with their overview-driven lazy loading." ([source](https://news.ycombinator.com/item?id=46255619)) "they're certainly superior to MCP in ways, being much easier to implement and reducing context usage by being discoverable" ([source](https://news.ycombinator.com/item?id=46643546)) "OCI packaged, verifiable skills for Flux that should progressively discovery instead of loading everything into context" ([source](https://news.ycombinator.com/item?id=47686932))
- Teaches agents tools and knowledge (6 comments): "The reason SKILLs work so well is because there's 50 years of accumulated knowledge of how to run rudimentary Unix tools." ([source](https://news.ycombinator.com/item?id=47291782)) "Skills are better than MCP because they basically encode the API docs/context in an LLM friendly manner." ([source](https://news.ycombinator.com/item?id=47715672)) "With these Skills in place, the agent is able to do things it would really struggle with otherwise" ([source](https://news.ycombinator.com/item?id=47040706))
- Lighter alternative to MCP (5 comments): "they're certainly superior to MCP in ways, being much easier to implement and reducing context usage by being discoverable" ([source](https://news.ycombinator.com/item?id=46643546)) "You've effectively flattened that MCP server into some Markdown and Bash, only the thing you have now is more flexible" ([source](https://news.ycombinator.com/item?id=46454358)) "cli tool+skill can perfectly replace it with less token overhead and better matching due to skill's front matter" ([source](https://news.ycombinator.com/item?id=47714383))
- Simple, low barrier to write (4 comments): "Skills seem to me like the simplest possible design for a mechanism for adding extra capabilities to an existing coding agent." ([source](https://news.ycombinator.com/item?id=46251478)) "they're certainly superior to MCP in ways, being much easier to implement and reducing context usage by being discoverable" ([source](https://news.ycombinator.com/item?id=46643546)) "SKILL.md or AGENT.md is genuinely easier to use, with a low barrier to entry for building custom workflows" ([source](https://news.ycombinator.com/item?id=49333021))

## What people warn about

- Agents often skip or ignore them (4 comments): "I experimented with skills, but they were skipped too often for the tool to work as reliably as it now does" ([source](https://news.ycombinator.com/item?id=48137164)) "I still get a lot of results where they are ignored even after spending time trying to tighten them up." ([source](https://news.ycombinator.com/item?id=48052189)) "Despite my previous efforts with skills, I've found agents will still do things like call help on CLIs" ([source](https://news.ycombinator.com/item?id=47726381))
- Overhyped or unnecessary standardization (5 comments): "Is any of this standardization really needed? Who does it benefit, except the people who enjoy writing specs and establishing standards like this?" ([source](https://news.ycombinator.com/item?id=46872295)) "I do see other comments speculating these skills work more efficiently but I'm not sure I have seen any evidence for that" ([source](https://news.ycombinator.com/item?id=46876389)) "The real issue is that prompts are not tightly coupled with capabilities. If we admit that, then skills are over hyped." ([source](https://news.ycombinator.com/item?id=48051949))
- Security risks in shared skills (2 comments): "I found a bunch of potential vulnerabilities in the example Skills .py files provided by Anthropic." ([source](https://news.ycombinator.com/item?id=46627484)) "most of the more than 100,000 skills on GitHub don't have any secure verification processes" ([source](https://news.ycombinator.com/item?id=46908031))
- Can still add token overhead (2 comments): "It feels a bit like the most likely outcomes are increased token burn, review surface, and time per task" ([source](https://news.ycombinator.com/item?id=49169979)) "You've just reversed the context benefits because the content of the skill...goes into context." ([source](https://news.ycombinator.com/item?id=47388368))

## Cite as

Hacker Approved, "Agent Skills (SKILL.md)", https://hackerapproved.com/p/agent-skills/ (counts from 1,795,640 Hacker News comments, November 2025 to November 2026).