Agent Skills

** PARITY GAP**: Grok One-Shot does not currently implement the Agent Skills system described in this document. This is a Claude Code feature planned for future implementation.

What Are Agent Skills?

Agent Skills are modular capabilities that extend an AI agent's functionality through organized folders containing instructions, scripts, and resources. Each Skill consists of a SKILL.md file with instructions that the AI reads when relevant, plus optional supporting files.

Status in Grok One-Shot

Current Status: Not Implemented

What Skills Enable (in Claude Code):

Extend AI capabilities: Package expertise into discoverable capabilities
Model-invoked: AI autonomously decides when to use them based on your request
Share expertise: Distribute Skills across teams via git or plugins
Reduce repetitive prompting: Encode common workflows as reusable Skills
Compose functionality: Combine multiple Skills for complex tasks

Example Use Cases:

PDF processing (extract text, fill forms, merge documents)
Code review checklists and standards
Project-specific workflows (deployment, testing, documentation)
Data analysis pipelines
Custom linting and validation

Alternative Approaches in Grok One-Shot

Since Skills aren't available yet, you can achieve similar outcomes through:

1. Project Context in GROK.md

Use the project's GROK.md file to encode domain knowledge and workflows:

# GROK.md

## PDF Processing

When working with PDF files:

1. Use `pypdf` for text extraction
2. Use `pdfplumber` for table extraction
3. Use `reportlab` for PDF generation

Example commands:
\`\`\`python
import pdfplumber
with pdfplumber.open("doc.pdf") as pdf:
text = pdf.pages[0].extract_text()
\`\`\`

## Code Review Standards

Always check:

- Error handling in all async functions
- Input validation for public APIs
- Test coverage for business logic
- Documentation for exported functions

2. Documentation in .agent/docs/

Create detailed guides in your project's .agent/docs/ directory:

# Project structure
.agent/
├── docs/
│ ├── workflows/
│ │ ├── pdf-processing.md
│ │ ├── deployment.md
│ │ └── code-review.md
│ └── standards/
│ ├── testing.md
│ └── security.md
└── docs-index.md

Grok One-Shot will load these on-demand when relevant.

3. Custom Slash Commands (Future)

Once custom slash commands are supported, you could create reusable prompts:

# .grok/commands/review-code.md (future)
---
description: Review code for best practices and potential issues
---

Review the specified code for:
1. Code organization and structure
2. Error handling
3. Performance considerations
4. Security concerns
5. Test coverage

4. Shell Scripts with MCP

Create helper scripts and expose them via MCP servers:

# scripts/pdf-extract.sh
#!/bin/bash
python3 -c "
import pdfplumber
import sys
with pdfplumber.open(sys.argv[1]) as pdf:
for page in pdf.pages:
print(page.extract_text())
" "$1"

Then reference these scripts in your GROK.md.

5. Session-Level Instructions

Provide detailed instructions at the start of each session:

> I'm working on PDF processing. When I ask you to extract PDF data, use pdfplumber for tables and pypdf for text. Always validate the file exists first and handle errors gracefully.

When Skills Are Implemented

Future Skills structure in Grok One-Shot would likely be:

~/.grok/skills/ # Personal Skills
.grok/skills/ # Project Skills
└── pdf-processing/
├── SKILL.md # Main skill definition
├── reference.md # Additional documentation
└── scripts/
└── extract.py # Helper scripts

Example SKILL.md (future):

---
name: pdf-processing
description: Extract text, fill forms, merge PDFs. Use when working with PDF files or when the user mentions PDFs, forms, or document extraction.
allowed-tools: Read, Write, Bash
---

# PDF Processing

## Instructions
1. Use pdfplumber for text extraction
2. Use pypdf for form filling
3. Always validate input files exist
4. Handle errors gracefully

## Examples
[Examples would go here]

Comparison: Skills vs Alternatives

Feature	Skills (Claude Code)	GROK.md	.agent/docs/
Auto-discovery	Yes (model-invoked)	No (always loaded)	Yes (on-demand)
Shareable	Via plugins	Via git	Via git
Tool restrictions	Yes (`allowed-tools`)	No	No
Multiple files	Yes	Single file	Yes
Progressive loading	Yes	No	Yes

What Are Agent Skills?​

Status in Grok One-Shot​

Alternative Approaches in Grok One-Shot​

1. Project Context in GROK.md​

2. Documentation in .agent/docs/​

3. Custom Slash Commands (Future)​

4. Shell Scripts with MCP​

5. Session-Level Instructions​

When Skills Are Implemented​

Comparison: Skills vs Alternatives​

See Also​