How the Shipping-Artifacts Skill Documents AI-Built Applications

The Shipping-Artifacts skill generates a standardized Markdown artifact that captures application architecture, version history, and security audit results to create a comprehensive review package for AI-generated code.

The Shipping-Artifacts skill in the phuryn/pm-skills repository automates documentation generation for applications built by large language models. As part of the pm-ai-shipping collection, this skill ensures that AI-built applications include structured, review-ready artifacts that human reviewers can evaluate efficiently. This article explains how the skill captures critical application details and formats them for integration into code review pipelines.

Overview of the Shipping-Artifacts Skill

Purpose and Command Invocation

The skill is triggered through the document-app command defined in pm-ai-shipping/commands/document-app.md. When executed, this command invokes the Shipping-Artifacts skill to begin collecting structured data about the AI-built application.

Integration with the pm-ai-shipping Collection

The skill operates alongside sibling skills including security-audit-static and performance-audit-static. According to the source code in pm-ai-shipping/skills/shipping-artifacts/SKILL.md, the skill references audit results from these complementary tools to include security and compliance notes in the final documentation.

How the Skill Documents AI-Built Applications

Prompt Flow and Data Collection

The skill employs a structured prompt flow that queries the LLM for four critical categories of information:

• Application overview: A concise description of the app's purpose and target users
• Version and changelog: The current version number and recent changes
• Key components: Architecture diagrams, data sources, and external services
• Security and compliance notes: References to static security or performance audits performed by sibling skills

Artifact Structure and Format

The skill assembles collected data into a Markdown document following the standard layout used for AI-generated artifacts in the repository. As implemented in pm-ai-shipping/skills/shipping-artifacts/SKILL.md, the artifact includes dedicated sections for:

• Summary
• Version
• Architecture
• Audit Results
• Next Steps

This standardized format ensures that reviewers receive a consistent, complete picture of what the application does and how it was built.

Output Handling and File Generation

Once generated, the document is written to the project's docs/ folder or a user-specified output path. The document-app command prints a direct link to the created file, enabling reviewers to open the artifact immediately from the CLI.

Integration with Review Pipelines

Because the generated artifact includes structured audit results and changelog data, downstream review tools can automatically ingest the Markdown. Code-review bots and PR templates parse these sections to surface information in pull request descriptions or review checklists, streamlining the validation process for AI-generated applications.

Practical Usage Examples

Generating an Artifact from the Terminal

Invoke the skill using the document-app command with the application name and desired output path:

pm-ai-shipping document-app \
  --app-name "SmartTicket" \
  --output docs/smartticket-artifact.md

This command launches the Shipping-Artifacts skill, gathers the required details, and writes a ready-to-review Markdown file to the specified location.

Referencing the Artifact in a PR Template

Link the generated documentation in your pull request template to provide reviewers with immediate access:

## AI-Built Application Review

**Artifact:** [SmartTicket Artifact](docs/smartticket-artifact.md)

- [ ] Architecture verified
- [ ] Security audit passed
- [ ] Performance benchmarks reviewed

The Markdown link points to the artifact created by the skill, establishing a single source of truth for the review process.

Key Source Files

The implementation of the Shipping-Artifacts skill spans several files in the phuryn/pm-skills repository:

• pm-ai-shipping/skills/shipping-artifacts/SKILL.md: Contains the detailed skill description, prompt definitions, expected inputs, and output format specifications.

• pm-ai-shipping/commands/document-app.md: Implements the CLI wrapper that invokes the skill and handles file writing operations.

• pm-ai-shipping/README.md: Provides high-level context for all shipping-related skills, including documentation standards and integration patterns.

Summary

• The Shipping-Artifacts skill standardizes documentation for AI-built applications through the document-app command.
• The skill collects application overview, version history, architecture details, and audit results via a structured prompt flow.
• Generated artifacts follow a consistent Markdown format with sections for Summary, Version, Architecture, Audit Results, and Next Steps.
• Output files are written to the docs/ directory and linked directly from the CLI for immediate reviewer access.
• The standardized format enables automatic ingestion by code-review bots and PR templates.

Frequently Asked Questions

What triggers the Shipping-Artifacts skill to document an AI-built application?

The skill is triggered through the document-app command located in pm-ai-shipping/commands/document-app.md. When executed from the CLI, this command initiates the prompt flow that gathers application details and generates the review artifact.

Where does the skill save the generated documentation?

According to the source code in pm-ai-shipping/skills/shipping-artifacts/SKILL.md, the skill writes the Markdown file to the project's docs/ folder by default, or to a user-specified path provided via the --output parameter.

How does the artifact integrate with security and performance audits?

The skill references results from sibling skills security-audit-static and performance-audit-static. It includes these audit results and compliance notes in the generated artifact's dedicated Audit Results section, ensuring reviewers have visibility into all validation checks performed on the AI-built application.

Can the output format be customized for specific review workflows?

While the skill adheres to the standard Markdown layout defined in the pm-ai-shipping collection, the output path is fully configurable via command-line arguments. Teams can specify custom directories or filenames using the --output flag to align with their specific review pipeline requirements.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}