707
GitHub Stars
1
Bundled Files
2 months ago
Catalog Refreshed
4 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstrat where the catalogue uses aiagentskills.
npx veilstrat add skill massgen/massgen --skill evolving-skill-creator- SKILL.md6.3 KB
Overview
This skill guides creation of evolving skills — reusable workflow plans that get better with each execution. It defines how to write a short skill file with required YAML frontmatter, plan the Python tools you'll implement, and capture concrete learnings after running the workflow. The output is a repeatable, discoverable workflow plus reusable scripts.
How this skill works
Define a descriptive name and clear description in a small YAML-fronted skill file, then write a step-by-step workflow that specifies commands, checks, and artifacts. Predeclare the Python scripts you will implement (purpose, inputs, outputs, dependencies) so tools are planned before coding. Execute the steps, produce the scripts, and update the skill file with what worked, what failed, and tips for future runs.
When to use it
- Starting a new multi-step task that should be repeatable or handed off
- When you expect to write Python utilities as part of the task
- Building workflows that will be iterated and improved over time
- When you want discoverable, named workflows for future automation
- Preparing tasks for multi-agent or CLI-driven orchestration
Best practices
- Pick a name that describes the type of task, not a specific instance
- Include YAML frontmatter with name and description for discoverability
- Document tools to create before writing them; include purpose, inputs, outputs, dependencies
- Write concrete, numbered workflow steps with commands and expected outputs
- Run the workflow as a user and add a Learnings section after execution
Example use cases
- Create a data-scraper workflow that outputs cleaned JSON and a report generator script
- Build a static-site generator workflow that scrapes content and produces themed HTML
- Assemble a PDF-report-generator workflow that collects data, formats templates, and exports PDFs
- Plan a multi-agent orchestration where agents call planned Python tools for subtasks
- Iterate a machine-learning data-prep workflow that produces reusable preprocessing scripts
FAQ
Always include YAML frontmatter with name and description; these fields are used to identify and find the workflow later.
When should I update the Learnings section?
Update Learnings immediately after running the workflow — record what worked, what failed, and practical tips so future runs improve quickly.