CtrlK
BlogDocsLog inGet started
Tessl Logo

local-ai-agents

Build local-first AI agents that run entirely on a developer workstation with Microsoft Foundry Local and Qwen function-calling models. Covers Small Language Models (SLMs), the OpenAI-compatible local endpoint, sandboxed local tools, local RAG with Chroma, local MCP servers, hybrid cloud/local routing, and the privacy/cost/offline trade-offs. Based on Lesson 17 of AI Agents for Beginners. USE FOR: run an agent locally, offline agent, on-device agent, Foundry Local, Qwen function calling, local tool calling, local RAG, Chroma vector database, local MCP server, privacy-preserving agent, hybrid local and cloud agent, small language model agent, engineering assistant on my machine. DO NOT USE FOR: deploying agents to the cloud at scale (use deploying-scalable-agents / Lesson 16), building your first agent concept (Lesson 01), Foundry (cloud) hosted agents, GPU cluster / server-side inference provisioning.

75

Quality

93%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

SecuritybySnyk

Passed

No findings from the security scan

SKILL.md
Quality
Evals
Security

Quality

Content

87%

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

A well-structured, lean skill body with executable setup code and appropriate deferral to a notebook for full reproduction. Its main weakness is workflow clarity: the build process lacks explicit validation checkpoints and error-recovery feedback loops.

Suggestions

Add an explicit build sequence with validation checkpoints — e.g., after registering tools, verify the model emits a well-formed tool call before proceeding to the RAG/MCP steps, and include a validate→fix→retry loop for sandboxed file operations.

Inline a minimal end-to-end tool-calling loop snippet (or a clear step checklist) so the core 'build a local agent' task is actionable without opening the notebook.

Add a brief local smoke-test/verification step (e.g., expected `foundry service status` output or a one-line agent self-check) so learners can confirm the setup worked before building on it.

DimensionReasoningScore

Conciseness

The body is lean and skill-specific — rationale like 'let the SLM orchestrate and let tools do the heavy lifting' and 'MCP is a transport, not a cloud service' supports specific recommendations without generic concept padding Claude already knows.

3 / 3

Actionability

Provides executable, copy-paste-ready setup code (foundry CLI commands and the FoundryLocalManager/OpenAI client snippet) with concrete RAM guidance, and explicitly signals deferral to the notebook for the full loop.

3 / 3

Workflow Clarity

Patterns are listed and the tool-calling loop gives a sequence ('register tools... execute... feed results back, repeat'), but the build workflow lacks explicit validate→fix→retry checkpoints, which caps it below 3.

2 / 3

Progressive Disclosure

Well-organized sections (Triggers, Core mental model, Why these pieces, Setup, Key patterns, Hybrid routing, Guardrails) with a single clearly-signaled one-level reference to the notebook; no bundle files present.

3 / 3

Total

11

/

12

Passed

Description

100%

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A strong, specific description with concrete capabilities, a rich natural trigger list, and explicit use/don't-use boundaries in third-person voice. It fully satisfies all four description dimensions.

DimensionReasoningScore

Specificity

Lists multiple concrete capabilities (Foundry Local, Qwen function-calling, Chroma RAG, local MCP servers, hybrid cloud/local routing) rather than vague language, matching the 'Lists multiple specific concrete actions' anchor.

3 / 3

Completeness

Explicitly answers both 'what' ('Build local-first AI agents... Covers SLMs...') and 'when' via the 'USE FOR:' triggers, plus a 'DO NOT USE FOR:' boundary — clearly both what AND when.

3 / 3

Trigger Term Quality

The 'USE FOR:' clause provides strong natural-language coverage ('run an agent locally', 'offline agent', 'on-device agent', 'privacy-preserving agent', 'engineering assistant on my machine') that a user would actually say.

3 / 3

Distinctiveness Conflict Risk

A clear niche (on-device/local-only agents) with explicit 'DO NOT USE FOR:' redirects to deploying-scalable-agents/Lesson 16, making conflict with adjacent skills unlikely.

3 / 3

Total

12

/

12

Passed

Validation

93%

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation15 / 16 Passed

Validation for skill structure

CriteriaDescriptionResult

relative_links

Relative link issues: 2 suspicious

Warning

Total

15

/

16

Passed

Repository
microsoft/ai-agents-for-beginners
Reviewed

Table of Contents

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.