Skip to main content

​
Overview

Use Agent Skills to extend EnConvo with task-specific capabilities. A skill packages instructions, resources, and optional scripts so EnConvo can follow a workflow reliably. You can share skills across teams or with the community.
Skills use progressive disclosure to manage context efficiently: EnConvo starts with each skill’s metadata (name, description, file path). Full instructions are loaded only when a skill is activated.

​
Skill Structure

A skill is a directory with a SKILL.md file plus optional scripts and references. 
my-skill/
├── SKILL.md          # Required: instructions + metadata
├── scripts/          # Optional: executable code
├── references/       # Optional: documentation
└── assets/           # Optional: templates, resources

​
SKILL.md Format

The SKILL.md file must include a frontmatter block with name and description: 
---
name: my-skill
description: Explain exactly when this skill should and should not trigger.
---

Step-by-step instructions for EnConvo to follow.
Because implicit matching depends on the description field, write descriptions with clear scope and boundaries. Be specific about when and when not to use the skill.

​
How Skills Are Activated

EnConvo can activate skills in two ways:

Explicit Invocation

Mention the skill directly in your prompt using the / prefix. Use this when you know exactly which skill you need.

Implicit Invocation

EnConvo automatically selects a skill when your task matches its description. No manual selection needed.

​
Skill Locations

EnConvo reads skills from multiple locations, with different scopes:
ScopeLocationUse Case
Project$PROJECT/.agents/skills/Skills specific to a project or workspace
User~/.agents/skills/Personal skills that apply across all projects
SystemBundled with EnConvoBuilt-in skills available to everyone
If two skills share the same name in different locations, both remain available — EnConvo does not merge them.

​
Creating a Skill

​
Using the Skill Creator

The fastest way to create a skill is with the built-in creator: 
/skill-creator
The creator walks you through:
  1. What the skill does
  2. When it should trigger
  3. Whether it should be instruction-only or include scripts
Instruction-only is the default and recommended approach. Use scripts only when you need deterministic behavior or external tooling.

​
Manual Creation

Create a folder with a SKILL.md file:
1

Create the skill directory

mkdir -p ~/.agents/skills/my-skill
2

Write SKILL.md

cat > ~/.agents/skills/my-skill/SKILL.md << 'EOF'
---
name: my-skill
description: Summarize selected text into bullet points. Use when the user asks for a summary or key takeaways. Do not use for translation or rewriting tasks.
---

## Instructions

1. Read the selected text from the user's input.
2. Identify the 3-5 most important points.
3. Return a bullet-point summary.
4. Keep each bullet under 20 words.
EOF
3

Verify

EnConvo detects skill changes automatically. If the new skill doesn’t appear, restart EnConvo.

​
Adding Scripts

For skills that need executable logic, add a scripts/ directory: 
my-skill/
├── SKILL.md
└── scripts/
    └── fetch_data.sh
Reference scripts in your SKILL.md: 
---
name: api-checker
description: Check API endpoint health and response times.
---

## Instructions

1. Run the health check script:
   ```bash
   bash scripts/fetch_data.sh $URL
  1. Parse the JSON output.
  2. Report status, response time, and any errors.

## Adding References

Include documentation or context files that the skill can consult:
my-skill/ ├── SKILL.md ├── references/ │ ├── api-docs.md │ └── examples.md └── scripts/ └── run.sh 

Reference them in instructions:

```markdown
Consult `references/api-docs.md` for the API schema before making requests.

​
Installing Skills

EnConvo supports two methods to install skills using /skill-installer:

​
From EnConvo Skills Store

Search and install skills by name from the built-in skills store: 
/skill-installer pdf
This searches the EnConvo skills store and installs the matching skill. EnConvo detects newly installed skills automatically.

​
From GitHub

Install skills directly from a GitHub repository URL — including subdirectories: 
/skill-installer https://github.com/anthropics/skills/tree/main/skills/pptx
GitHub installation supports subdirectory paths, so you can point to a specific skill folder within a larger repository. No need to clone the entire repo.

​
Best Practices

One skill = one job. A skill that does too many things is harder to trigger correctly and harder to maintain.
Use plain-language instructions unless you need deterministic behavior or external tooling. Instructions are easier to write, debug, and iterate on.
Be explicit about inputs, steps, and expected outputs. Vague instructions lead to inconsistent results.
Try various prompts against your skill’s description field to confirm it triggers (and doesn’t trigger) correctly.
In the description, state both what the skill should handle and what it should not handle.

​
Example Skills

​
Code Reviewer

---
name: code-reviewer
description: Review code for bugs, security issues, and best practices. Use when the user asks for a code review or wants to improve code quality. Do not use for code generation or writing new code.
---

## Instructions

1. Read the provided code carefully.
2. Check for:
   - Logic errors and edge cases
   - Security vulnerabilities (injection, XSS, etc.)
   - Performance issues
   - Code style and readability
3. Provide feedback as a numbered list.
4. For each issue, include:
   - Severity (🔴 Critical / 🟡 Warning / 🔵 Suggestion)
   - Line reference
   - Explanation and fix

​
Meeting Notes

---
name: meeting-notes
description: Format raw meeting transcripts into structured notes with action items. Use when the user provides meeting notes, transcripts, or asks to organize meeting content. Do not use for scheduling or calendar tasks.
---

## Instructions

1. Parse the meeting transcript or raw notes.
2. Extract and organize into sections:
   - **Attendees**: List all participants
   - **Key Decisions**: Bullet-point decisions made
   - **Discussion Summary**: 3-5 sentence overview
   - **Action Items**: Table with Owner, Task, Deadline
3. Use markdown formatting for readability.

​
Resources

EnConvo GitHub

Browse open-source extensions and skills

Community Discord

Share skills and get help from the community