Help us improve
Share bugs, ideas, or general feedback.
From evolve-lite
Captures successful session workflows from conversation history, generating reusable SKILL.md files with docs, parameters, examples, and Python helper scripts. Use after multi-step tasks for reuse.
npx claudepluginhub agenttoolkit/altk-evolve --plugin evolve-liteHow this skill is triggered — by the user, by Claude, or both
Slash command
/evolve-lite:saveThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
This skill analyzes your current successful session and generates a new reusable skill with:
Analyzes session work, detects duplicate patterns, and automatically converts reusable workflows into Claude Code skills via Agent Teams.
Capture this session's repeatable process into a reusable skill. Call at end of a process you want to automate.
Turns Entire-tracked sessions, checkpoints, or repeated workflows into reusable SKILL.md drafts by extracting durable patterns via entire search and session metadata. Presents drafts for approval before writing.
Share bugs, ideas, or general feedback.
This skill analyzes your current successful session and generates a new reusable skill with:
It extracts the workflow pattern from your conversation history (user requests, reasoning steps, tool calls, and responses) and creates parameterized files that can be invoked in future sessions.
Use this skill when you've completed a task successfully and want to save the workflow for future reuse.
Analyze the conversation history available in the current context, which includes:
Action: Review the entire conversation from start to current point
Extract the high-level workflow by:
Example Pattern Recognition:
User Goal: "Read a file and display its contents"
Workflow Pattern:
1. Attempt to read file at expected location
2. If access denied → check allowed directories
3. Search for file in allowed directories
4. Read file from correct location
5. Format and present results
Apply conservative parameterization - only parameterize obvious session-specific values:
Parameterize:
{file_path} or {directory}{filename}{data_value}{project_name}{workspace_dir}Keep Unchanged:
read_file, execute_command)Example:
Original: "Read /home/user/projects/myapp/config.json"
Parameterized: "Read {project_dir}/{config_file}"
Analyze the workflow to determine if helper scripts would be beneficial:
Generate scripts when the workflow includes:
Script Types to Consider:
Example:
Workflow includes: Reading JSON file, extracting specific fields, formatting output
→ Generate: parse_and_format.py script
Create a new SKILL.md file with the following structure:
---
name: {skill-name}
description: {one-line description of what this skill does}
---
# {Skill Title}
## Overview
{Brief description of the skill's purpose and when to use it}
## Parameters
{List parameters the user needs to provide}
- **{param_name}**: {description and example}
## Workflow
### Step 1: {Step Name}
{What this step does}
**Action**: {Tool or approach to use}
**Example**:
{Example tool call or command}
{If helper script exists, reference it}
**Helper Script**: Use `scripts/{script_name}.py` for this operation
{Repeat for each step}
## Helper Scripts
{If scripts were generated, document them}
### {script_name}.py
**Purpose**: {What the script does}
**Usage**:
```bash
python3 ${CLAUDE_PLUGIN_ROOT}/skills/{skill-name}/scripts/{script_name}.py [arguments]
Parameters:
{param}: {description}Example:
python3 ${CLAUDE_PLUGIN_ROOT}/skills/{skill-name}/scripts/parse_data.py input.json
{Common errors and how to handle them}
Input:
Expected Output: {What the user should see}
{Additional tips or context}
### Step 6: Generate Helper Scripts
For each identified script opportunity, create a Python script with:
**Script Template**:
```python
#!/usr/bin/env python3
"""
{Script description}
Usage:
python3 {script_name}.py [arguments]
Arguments:
{arg1}: {description}
{arg2}: {description}
"""
import sys
import json
import argparse
from pathlib import Path
def main():
"""Main function implementing the script logic."""
parser = argparse.ArgumentParser(description="{Script description}")
parser.add_argument("{arg1}", help="{description}")
parser.add_argument("{arg2}", help="{description}", nargs="?")
args = parser.parse_args()
# Implementation based on workflow pattern
try:
# Core logic here
result = process_data(args.{arg1})
print(json.dumps(result, indent=2))
except Exception as e:
print(f"Error: {e}", file=sys.stderr)
sys.exit(1)
def process_data(input_data):
"""Process the input data according to the workflow pattern."""
# Implementation extracted from session workflow
pass
if __name__ == "__main__":
main()
Script Guidelines:
chmod +x)Ask the user: "What would you like to name this skill?"
Naming Guidelines:
read-file-with-permissions, deploy-to-staging, analyze-logsSuggest a name based on the workflow if the user is unsure:
read-file-with-permission-checkBefore saving, check if a skill with this name already exists:
Action: Check if ~/.claude/skills/{skill-name}/SKILL.md exists
If exists:
Action: Create the skill directory structure and save all files
~/.claude/skills/{skill-name}/~/.claude/skills/{skill-name}/SKILL.md~/.claude/skills/{skill-name}/scripts/~/.claude/skills/{skill-name}/scripts/{script_name}.pychmod +x ~/.claude/skills/{skill-name}/scripts/*.pyDirectory Structure:
~/.claude/skills/{skill-name}/
├── SKILL.md
└── scripts/ (if applicable)
├── script1.py
└── script2.py
Note: The skill is saved to the user's home directory (~/.claude/skills/) making it available across all projects.
Present a clear summary to the user:
✅ Skill saved successfully!
**Skill Name**: {skill-name}
**Location**: ~/.claude/skills/{skill-name}/
**Files Created**:
- SKILL.md (workflow documentation)
{if scripts}
- scripts/{script1}.py (helper script for {purpose})
- scripts/{script2}.py (helper script for {purpose})
{endif}
**Summary**: {Brief description of what the skill does}
**Workflow Captured**:
1. {Step 1 summary}
2. {Step 2 summary}
3. {Step 3 summary}
...
**Parameters**:
- **{param1}**: {description}
- **{param2}**: {description}
**Helper Scripts**:
{if scripts}
- **{script1}.py**: {what it does}
- **{script2}.py**: {what it does}
{endif}
**To use this skill**: Simply reference it by name in future sessions: "{skill-name}"
Session Too Short:
No Clear Workflow:
Skill Name Conflicts:
Invalid Skill Name:
Script Generation Errors:
Session Context:
User: "Read the states.txt file and parse it into a JSON array"
Assistant: [Reads file, parses lines, converts to JSON, outputs result]
User: "Great! Save this as a skill"
Generated Skill Name: read-and-parse-file
Parameters Identified:
filename: The file to readoutput_format: Format for output (json, csv, etc.)Workflow Captured:
Scripts Generated:
parse_file.py: Reads a file and converts it to JSON formatFiles Created:
~/.claude/skills/read-and-parse-file/
├── SKILL.md
└── scripts/
└── parse_file.py
Session Context:
User: "Deploy the app to staging"
Assistant: [Runs tests, builds app, uploads to server, restarts service]
User: "Perfect! Save this workflow"
Generated Skill Name: deploy-to-staging
Parameters Identified:
app_name: Name of the applicationserver_address: Staging server addressWorkflow Captured:
Scripts Generated:
run_tests.py: Execute test suite and report resultsdeploy.py: Handle upload and service restartFiles Created:
~/.claude/skills/deploy-to-staging/
├── SKILL.md
└── scripts/
├── run_tests.py
└── deploy.py
Session Context:
User: "List all Python files in the project"
Assistant: [Uses glob tool to find *.py files, displays results]
User: "Save this"
Generated Skill Name: list-python-files
Workflow Captured:
Scripts Generated: None (simple tool call, no script needed)
Files Created:
~/.claude/skills/list-python-files/
└── SKILL.md
~/.claude/skills/ making them available in all your projects.~/.claude/skills/ directory to version control to track skill evolution.