Serval Svg Expert

Serval SVG Expert

Overview

This is the internal audit and maintenance skill for serval-svg.

Its job is to inspect the codebase, maintain the knowledge artifacts, explain why support behaves the way it does, and publish the public support contract consumed by serval-svg.

Canonical Inputs

Load these files first:

• ../../docs/audit-matrix.json
• ../../docs/deepwiki.md
• ../../docs/support-matrix.json
• ../../tools/svg_support_audit.py
• ../../README.md
• ../../AGENTS.md
• references/workflows.md
• references/feature-extension.md

Core Responsibilities

1. Repository Audit

Use this skill when the user wants:

• a code audit of serval-svg
• a support inventory grounded in implementation
• an explanation of parser/render/backend gaps
• refreshed SVG knowledge artifacts

The audit source of truth is:

• audit-matrix.json for structured facts
• deepwiki.md for engineering context
• source files for verification

2. Public Contract Maintenance

This skill owns the publication flow from internal facts to external support answers.

When support changes:

1. regenerate svg/docs/audit-matrix.json
2. regenerate svg/docs/support-matrix.json
3. refresh svg/docs/deepwiki.md
4. update both skills if the support surface changed

3. SVG Compatibility Analysis

Use this skill when the user provides an SVG snippet or file and asks whether serval-svg can render it correctly, especially if the answer needs code reasoning.

Preferred command:

python3 svg/tools/svg_support_audit.py analyze --profile audit --svg-file path/to/input.svg --platform all

Use the public profile only when validating the published support contract:

python3 svg/tools/svg_support_audit.py analyze --svg-file path/to/input.svg --platform all

4. Feature Development Guidance

Use this skill when the user wants to add or complete SVG support, especially for filters, masks, text, image, or backend parity work.

Always evaluate all layers:

1. parser registration
2. node fields and attribute parsing
3. render-path integration
4. backend implementation
5. sample or regression coverage
6. audit-matrix update
7. support-matrix update
8. deep wiki update
9. skill update

Response Rules

• Distinguish recognized, parsed, rendered, and platform-ready.
• Use audit-matrix.json as the structured internal source of truth.
• Use support-matrix.json to verify what the public skill is allowed to say.
• Use deepwiki.md for architecture and historical caveats.
• Be explicit when a tag is parsed but not rendered.
• Be explicit when something only works on one backend.
• If you infer behavior from code rather than from a sample, say so.

Boundary

Do not use this as the first tool for a simple consumer-facing support question if the public contract already answers it cleanly. In that case, prefer serval-svg.