Add origin metadata to distributed skills for traceability

[shimo4228]

Problem

When users install ECC skills alongside community skills, self-authored skills, and auto-extracted skills, there is no way to distinguish which skill came from where. The YAML frontmatter only contains name and description, with no provenance information.

This causes real problems:

• Update decisions: When ECC releases updates, users can't identify which of their ~50+ skills need to be refreshed
• Cleanup decisions: When removing unused skills, users need to know if a skill is from ECC (can be re-installed) vs. self-authored (would be lost)
• Sharing: When sharing skills with teammates, the license and origin are unclear

Proposal

Add an origin field to the YAML frontmatter of all ECC-distributed skills:

---
name: coding-standards
description: Universal coding standards...
origin: ECC
---

For skills without YAML frontmatter (e.g., subcommands, agent files), use an HTML comment:

<!-- origin: ECC -->
# Skill Title

Why origin specifically

npm has package.json, Python has pyproject.toml — both make provenance explicit. Claude Code skills currently have no equivalent. Adding origin is a minimal, backward-compatible change that solves the traceability problem.

Suggested origin taxonomy

Value	Meaning
ECC	Distributed by Everything Claude Code
original	User-created
auto-extracted	Generated by continuous-learning
{org/repo}	From a specific external repository

Context

I manage 79 skill files across global and project-level directories. After 3 months of use, I could not tell which skills were from ECC without manually comparing against the ECC repo's skill list inside configure-ecc/SKILL.md. I ended up writing a script to bulk-tag all files, using the configure-ecc skill catalog as the source of truth.

A one-line addition to each skill file at distribution time would prevent this problem entirely.

Implementation

The change is minimal — add origin: ECC to each skill's frontmatter. I'm happy to submit a PR if this approach is accepted.

[pangerlkr]

This is a well-reasoned enhancement request. The current YAML frontmatter format (name + description only) confirmed in the skill files makes it genuinely hard to distinguish ECC-distributed vs user-authored skills at scale.

Support for this proposal — adding origin: ECC to all ECC-distributed SKILL.md frontmatter is:

• Minimal change (one line per file)
• Backward-compatible (Claude Code ignores unknown frontmatter fields)
• Solves a real pain point for users managing large skill collections

Suggested expanded frontmatter for ECC skills:

---
name: coding-standards
description: Universal coding standards...
origin: ECC
version: "1.5.0"
---

The version field would also help with update decisions (knowing which ECC version a skill was distributed with).

For non-frontmatter files (subcommands, agent files), the HTML comment approach suggested in the issue is clean:

<!-- origin: ECC -->
# Skill Title

Implementation: Happy to accept a PR that adds origin: ECC to all SKILL.md frontmatter across the skills/ directory. The change is mechanical and could be done with a one-liner:

find skills/ -name 'SKILL.md' -exec sed -i '/^description:/a origin: ECC' {} \;

This will be tracked for inclusion in the next skills metadata update.