Use a structured starting point to write new context engineering skills. Organize activation conditions, responsibilities, practical steps, examples, and common mistakes, while keeping detailed references separate and clarifying boundaries with related skills.
Template for creating new Agent Skills for context engineering. Use this template when adding new skills to the collection.
Review the instructions the agent will receive before you install.
Provide a clear, concise description of what this skill covers and when to use it. This description appears in skill discovery and should help agents (and humans) determine when this skill is relevant.
Important: Keep the total SKILL.md body under 500 lines for optimal performance. Move detailed reference material to separate files in the references/ directory.
Every skill body must make its ownership boundary explicit. The description and When to Activate section should say what the skill owns and which adjacent skills own nearby work. This prevents broad skills from stealing activation from narrower skills.
Describe specific situations, tasks, or contexts where this skill should be activated. Include both direct triggers (specific keywords or task types) and indirect signals (broader patterns that indicate skill relevance).
Write in third person. The description is injected into the system prompt, and inconsistent point-of-view can cause discovery problems.
Include a short "Do not activate" block for adjacent skills. Example:
project-development.tool-design.Explain the fundamental concepts covered by this skill. These are the mental models, principles, or frameworks that the skill teaches.
Default assumption: Claude is already very smart. Only add context Claude does not already have. Challenge each piece of information:
Prefer behavior-changing mechanisms over general background. If a concept should be reusable across the corpus, add or update a record in researcher/mechanisms/registry.jsonl.
Provide detailed explanation of the first major topic. Include specific techniques, patterns, or approaches. Use examples to illustrate concepts.
Provide detailed explanation of the second major topic. Continue with additional topics as needed.
For longer topics, consider moving content to references/ and linking:
Provide actionable guidance for applying the skill. Include common patterns, anti-patterns to avoid, and decision frameworks for choosing between approaches.
Match the level of specificity to the task's fragility:
Practical guidance should be executable by an agent: a workflow, checklist, decision table, or concrete operating rule. If a section only explains history or motivation, move it to references/.
Provide concrete examples that illustrate skill application. Examples should show before/after comparisons, demonstrate correct usage, or show how to handle edge cases.
Use input/output pairs for clarity:
Example:
Input: [describe input]
Output: [show expected output]
List specific guidelines to follow when applying this skill. These should be actionable rules that can be checked or verified.
List experience-derived failure modes, common mistakes, and counterintuitive behaviors. These are the highest-signal content in any skill. Each gotcha should be specific, actionable, and non-overlapping with guidance already in the skill body. Use numbered format:
Explain how this skill integrates with other skills in the collection. List related skills as plain text (not links) to avoid cross-directory reference issues:
Internal reference (use relative path to skill's own reference files):
Related skills in this collection:
External resources:
Numeric, benchmark, volatile, or vendor-performance claims need an inline claim-* ID backed by researcher/claims/index.jsonl, or they should be softened and moved to dated reference material.
Created: [Date] Last Updated: [Date] Author: [Author or Attribution] Version: [Version number]
No reviews yet. Tell others what you thought.
No login needed. Share how it worked for you, ask a question, or note where it did not run.
No comments yet. Be the first to write one.
Design systems that use language models to judge AI responses. Covers scoring criteria, comparisons between two answers, bias reduction, confidence calibration, and checks against human judgments to make automated quality assessments more consistent and traceable.
Model what an agent believes, wants, and commits to doing. Connect external facts to goals and plans using structured relationships, with evidence and validity periods that help trace why the agent chose an action.
Turn ePub books into training examples for models that reproduce an author's writing style. Covers text extraction, coherent segmentation, varied instructions, model training, and checks for copied phrases and style transfer to new scenarios.
Guide multi-source research through source checks, comparison of key claims, and recovery from failed tool calls. Produce reports that distinguish retrieved evidence from unresolved questions and clearly document remaining gaps.
Design structured summaries for long agent sessions that preserve goals, file changes, decisions, and next actions. Choose compression methods and timing, then test whether the agent can continue accurately without repeatedly retrieving information it already processed.
Diagnose why an AI agent becomes less reliable as conversations and documents accumulate. Distinguish missed information, persistent false claims, irrelevant material, mixed tasks, and conflicting sources, then choose ways to clean, reorganize, or separate the context.