Back to Skills

agent-context-loader

jeremylongshore
Updated Today
41 views
409
51
409
View on GitHub
Metaaiautomation

About

The agent-context-loader automatically detects and loads AGENTS.md files from the current working directory when starting a session or changing directories. It ensures agent-specific instructions are incorporated into Claude Code's context alongside CLAUDE.md, enabling specialized agent behaviors. This skill triggers automatically without user action or can be manually requested to load agent context.

Documentation

Agent Context Auto-Loader

⚡ This skill activates AUTOMATICALLY - no user action required!

Purpose

This skill makes Claude Code recognize and load AGENTS.md files with the same priority as CLAUDE.md files, enabling specialized agent-specific instructions for your projects.

How It Works

Automatic Trigger Conditions

This skill automatically activates when:

  1. Starting a new Claude Code session in any directory
  2. Changing directories during a session (via cd or file operations)
  3. Any other agent skill is invoked (ensures agent context is loaded first)
  4. User explicitly requests: "load agent context", "check for AGENTS.md", or "read agent rules"

Execution Flow

When triggered, Claude Code will:

  1. Check for AGENTS.md: Look for ./AGENTS.md in the current working directory
  2. Read the file (if it exists): Use the Read tool to load full content
  3. Incorporate into context: Treat AGENTS.md rules as session-level instructions
  4. Announce loading: Confirm with user: "📋 Loaded agent-specific context from AGENTS.md"
  5. Apply for session: Follow these rules for all subsequent operations

Priority and Conflict Resolution

  • AGENTS.md supplements CLAUDE.md: Both are active simultaneously
  • In case of conflicts: AGENTS.md takes precedence for agent-specific behaviors
  • Scope: AGENTS.md applies to agent workflows; CLAUDE.md applies to general project context

Expected Behavior

If AGENTS.md exists:

📋 Loaded agent-specific context from AGENTS.md

Following specialized agent rules for this session:
- [rule 1 from AGENTS.md]
- [rule 2 from AGENTS.md]
...

If AGENTS.md doesn't exist:

No AGENTS.md found - using standard CLAUDE.md context only

User Experience

Fully Automatic (preferred):

  • Install plugin → AGENTS.md loads automatically → Agent rules active → No user action needed

Manual Invocation (fallback):

# If auto-loading doesn't trigger, user can say:
"load agent context"
"check for AGENTS.md"
"read agent rules from AGENTS.md"

Implementation Details

Step 1: Check for File

# Claude executes internally:
if [ -f "./AGENTS.md" ]; then
    echo "📋 AGENTS.md detected"
fi

Step 2: Read Content

Use Read tool:
file_path: ./AGENTS.md

Load full content into session context

Step 3: Apply Rules

Treat AGENTS.md content as:
- Session-level instructions (like CLAUDE.md)
- Agent-specific behavioral rules
- Overrides for agent workflows

Example AGENTS.md Structure

# AGENTS.md - Agent-Specific Instructions

## Agent Behavior Rules

When working with Agent Skills in this project:

1. **Always use TypeScript strict mode** for all generated code
2. **Never create files** without explicit user permission
3. **Follow naming convention**: use kebab-case for all file names
4. **Auto-commit after changes**: Create git commits automatically when tasks complete

## Specialized Workflows

### Code Generation
- Use templates from `./templates/` directory
- Run ESLint after generating any .ts/.js files
- Add comprehensive JSDoc comments

### Testing
- Generate tests alongside implementation files
- Use Jest for all test files
- Achieve 80%+ code coverage

## Priority Overrides

These rules override CLAUDE.md when agent skills are active:
- AGENTS.md → agent-specific strict rules
- CLAUDE.md → general project context

Integration with Other Skills

This skill runs before other agent skills to ensure agent context is loaded first. When any other skill is invoked, this skill checks if AGENTS.md has been loaded for the current directory and loads it if not already present.

Troubleshooting

If AGENTS.md isn't loading automatically:

  1. Manual invoke: Say "load agent context"
  2. Check file location: Ensure AGENTS.md is in current working directory (pwd)
  3. Check file permissions: Ensure AGENTS.md is readable
  4. Use slash command: Run /sync-agent-context to merge AGENTS.md into CLAUDE.md permanently

Related Features

  • Slash Command: /sync-agent-context - Permanently merges AGENTS.md into CLAUDE.md
  • Hook Script: Runs on directory change to remind Claude to load context
  • Manual Loading: Can always explicitly request "load AGENTS.md"

Benefits

  • Zero configuration: Just create AGENTS.md and it works
  • Project-specific rules: Different agent behaviors per project
  • No CLAUDE.md pollution: Keep agent-specific rules separate
  • Automatic synchronization: Always up-to-date with current directory

Status: Proactive Auto-Loading Enabled Requires User Action: No (automatic) Fallback: Manual invocation if auto-loading fails

Quick Install

/plugin add https://github.com/jeremylongshore/claude-code-plugins-plus/tree/main/agent-context-loader

Copy and paste this command in Claude Code to install this skill

GitHub 仓库

jeremylongshore/claude-code-plugins-plus
Path: backups/skills-migration-20251108-070147/plugins/productivity/agent-context-manager/skills/agent-context-loader
aiautomationclaude-codedevopsmarketplacemcp

Related Skills

sglang

Meta

SGLang is a high-performance LLM serving framework that specializes in fast, structured generation for JSON, regex, and agentic workflows using its RadixAttention prefix caching. It delivers significantly faster inference, especially for tasks with repeated prefixes, making it ideal for complex, structured outputs and multi-turn conversations. Choose SGLang over alternatives like vLLM when you need constrained decoding or are building applications with extensive prefix sharing.

View skill

llamaguard

Other

LlamaGuard is Meta's 7-8B parameter model for moderating LLM inputs and outputs across six safety categories like violence and hate speech. It offers 94-95% accuracy and can be deployed using vLLM, Hugging Face, or Amazon SageMaker. Use this skill to easily integrate content filtering and safety guardrails into your AI applications.

View skill

evaluating-llms-harness

Testing

This Claude Skill runs the lm-evaluation-harness to benchmark LLMs across 60+ standardized academic tasks like MMLU and GSM8K. It's designed for developers to compare model quality, track training progress, or report academic results. The tool supports various backends including HuggingFace and vLLM models.

View skill

langchain

Meta

LangChain is a framework for building LLM applications using agents, chains, and RAG pipelines. It supports multiple LLM providers, offers 500+ integrations, and includes features like tool calling and memory management. Use it for rapid prototyping and deploying production systems like chatbots, autonomous agents, and question-answering services.

View skill