Turn an agent you use into a template anyone can install. The process strips personal data and creates an example run with fabricated names. Your procedure stays, your secrets do not.
Navigate to the project where your gcontext agent lives. The agent must have at least one agent module with a completed run in runs/.
Copy the skill URL below and paste it into Claude Code. The skill walks you through packaging: it reads your agent, classifies every personal element, and asks you to confirm before it changes anything.
The skill replaces personal data with parameter slots, rewrites service-specific references to generic descriptions, and generates a setup command. Your steps stay structurally identical; only the specifics change.
A fabricated run with fake names and plausible data is generated from your most representative real run. This example is what visitors see on the marketplace page.
Run gcontext share <template-folder> to validate locally and print the PR steps. Then open a pull request against github.com/bleak-ai/agents that adds one folder at the repo root. A maintainer reviews and merges it. Your agent appears in the directory after the merge.
The share-agent skill handles the full process: eligibility checks, slot extraction, template generation, example fabrication, leak scanning, and submission. For the complete reference, read the full skill document on GitHub.