Offline support?

[zhaob1n]

It looks like the skill content is based on web url, is it possible to make the skill completely offline?

[pavelzw]

agree with this. it would be really nice to be able to use this skill in locked-down environments with no internet access. this is where skills really shine imo; otherwise the agents can also just fetch the appropriate websites themselves

[pavelzw]

pinging @adtyavrdhn and @Kludex as it seems you are not watching your own repo

[Kludex]

What do you mean by this? Where is the url?

[pavelzw]

skills/skills/building-pydantic-ai-agents/SKILL.md

Lines 217 to 262 in cb098e8

	| Create or configure agents | [Agents](https://ai.pydantic.dev/agents/) |
	| Bundle reusable behavior (tools, hooks, instructions) | [Capabilities](https://ai.pydantic.dev/capabilities/) |
	| Intercept model requests, tool calls, or runs | [Hooks](https://ai.pydantic.dev/hooks/) |
	| Define agents in YAML/JSON without Python code | [Agent Specs](https://ai.pydantic.dev/agent-spec/) |
	| Use template strings in agent instructions | [Template Strings](https://ai.pydantic.dev/agent-spec/#template-strings) |
	| Let my agent call external APIs or functions | [Tools](https://ai.pydantic.dev/tools/) |
	| Organize or restrict which tools an agent can use | [Toolsets](https://ai.pydantic.dev/toolsets/) |
	| Give my agent web search with automatic provider fallback | [WebSearch Capability](https://ai.pydantic.dev/capabilities/#provider-adaptive-tools) |
	| Give my agent URL fetching with automatic provider fallback | [WebFetch Capability](https://ai.pydantic.dev/capabilities/#provider-adaptive-tools) |
	| Give my agent web search or code execution (builtin tools) | [Built-in Tools](https://ai.pydantic.dev/builtin-tools/) |
	| Search with DuckDuckGo/Tavily/Exa | [Common Tools](https://ai.pydantic.dev/common-tools/) |
	| Ensure my agent returns data in a specific format | [Structured Output](https://ai.pydantic.dev/output/#structured-output) |
	| Pass database connections, API clients, or config to tools | [Dependencies](https://ai.pydantic.dev/dependencies/) |
	| Access usage stats, message history, or retry count in tools | [RunContext](https://ai.pydantic.dev/tools/) |
	| Choose or configure models | [Models](https://ai.pydantic.dev/models/) |
	| Automatically switch to backup model when primary fails | [Fallback Model](https://ai.pydantic.dev/models/#fallback-model) |
	| Show real-time progress as my agent works | [Streaming Events and Final Output](https://ai.pydantic.dev/agents/#streaming-events-and-final-output) |
	| Work with messages and multimedia | [Message History](https://ai.pydantic.dev/message-history/) |
	| Reduce token costs by trimming or filtering conversation history | [Processing Message History](https://ai.pydantic.dev/message-history/#processing-message-history) |
	| Keep long conversations manageable without losing context | [Summarize Old Messages](https://ai.pydantic.dev/message-history/#summarize-old-messages) |
	| Use MCP servers | [MCP](https://ai.pydantic.dev/mcp/) |
	| Build multi-step graphs | [Graph](https://ai.pydantic.dev/graph/) |
	| Debug a failed agent run or see what went wrong | [Model Errors](https://ai.pydantic.dev/agents/#model-errors) |
	| Make my agent resilient to temporary failures | [Retries](https://ai.pydantic.dev/retries/) |
	| Understand why my agent made specific decisions | [Using Logfire](https://ai.pydantic.dev/logfire/#using-logfire) |
	| Write deterministic tests for my agent | [Unit testing with TestModel](https://ai.pydantic.dev/testing/#unit-testing-with-testmodel) |
	| Enable thinking/reasoning across any provider | [Thinking](https://ai.pydantic.dev/thinking/) · [Thinking Capability](https://ai.pydantic.dev/capabilities/#thinking) |
	| Systematically verify my agent works correctly | [Evals](https://ai.pydantic.dev/evals/) |
	| Use embeddings for RAG | [Embeddings](https://ai.pydantic.dev/embeddings/) |
	| Use durable execution | [Durable Execution](https://ai.pydantic.dev/durable_execution/overview/) |
	| Have one agent delegate tasks to another | [Agent Delegation](https://ai.pydantic.dev/multi-agent-applications/#agent-delegation) |
	| Route requests to different agents based on intent | [Programmatic Agent Hand-off](https://ai.pydantic.dev/multi-agent-applications/#programmatic-agent-hand-off) |
	| Require tool approval (human-in-the-loop) | [Deferred Tools](https://ai.pydantic.dev/deferred-tools/) |
	| Use images, audio, video, or documents | [Input](https://ai.pydantic.dev/input/) |
	| Use advanced tool features | [Advanced Tools](https://ai.pydantic.dev/tools-advanced/) |
	| Validate or require approval before tool execution | [Advanced Tools](https://ai.pydantic.dev/tools-advanced/) |
	| Call the model without using an agent | [Direct API](https://ai.pydantic.dev/direct/) |
	| Expose agents as HTTP servers (A2A) | [A2A](https://ai.pydantic.dev/a2a/) |
	| Handle network errors and rate limiting automatically | [Retries](https://ai.pydantic.dev/retries/) |
	| Use LangChain or ACI.dev tools | [Third-Party Tools](https://ai.pydantic.dev/third-party-tools/) |
	| Publish reusable agent extensions as packages | [Extensibility](https://ai.pydantic.dev/extensibility/) |
	| Build custom toolsets, models, or agents | [Extensibility](https://ai.pydantic.dev/extensibility/) |
	| Debug common issues | [Troubleshooting](https://ai.pydantic.dev/troubleshooting/) |
	| Migrate from deprecated APIs | [Changelog](https://ai.pydantic.dev/changelog/) |
	| See advanced real-world examples | [Examples](https://ai.pydantic.dev/examples/) |
	| Look up an import path | [API Reference](https://ai.pydantic.dev/api/) |

will lead to the agent making web requests. this is not possible in some scenarios

[Kludex]

In our case we reference the docs, is the point that you think we should not do that?

[pavelzw]

Yes.
One example of my use case:
Im bundling skills as conda packages in https://github.com/pavelzw/skill-forge so that I can then install them into my repository. Then, I add a symlink from .claude/skills/<skill> to the appropriate place in the Conda environment. As long as I have installed the environment correctly, the skills are available across all devices where I use them.
They are also available in locked-down corporate environments where I only have access to a internal Artifactory instance to download packages as all other traffic is blocked.

https://github.com/pavelzw/skill-forge/tree/main/recipes/typst for example contains additional markdown files in references/ as advised on https://agentskills.io

See my blog post about this for more details.

[adtyavrdhn]

@pavelzw How would you feel if we kept it simple and inlined these references like the other references in the skills with snippets for the most important bits of the table?

[pavelzw]

Yes sounds good