feat(skills): add screen-recording awareness skill

[Jasonnnz]

Summary

• Add a bundled screen-recording skill with SKILL.md that teaches the model about screen recording capabilities
• Awareness-only skill (no TOOLS.json) — prepares for future tool-bearing recording control
• Covers recording lifecycle, health checks, retention, limitations, and when to mention recording to users

Test plan

• Verify SKILL.md has valid YAML frontmatter
• Verify file is in the correct bundled-skills directory
• No build step needed for SKILL.md-only skills

🤖 Generated with Claude Code

---

[devin-ai-integration]

Devin Review found 1 potential issue.

View 3 additional findings in Devin Review.

[devin-ai-integration]

🔴 Multi-line YAML metadata not parsed by line-based frontmatter parser, losing os constraint and emoji

The metadata field in the new SKILL.md uses multi-line nested YAML syntax, but the frontmatter parser (assistant/src/skills/frontmatter.ts:28-62) is a simple line-by-line key-value parser that splits on :. It does not support multi-line YAML or nested structures.

Root Cause and Impact

When the parser encounters:

metadata:
  vellum:
    emoji: "🎥"
    os: ["darwin"]

It parses each line independently, producing:

• metadata → "" (empty string)
• vellum → "" (empty string)
• emoji → "🎥"
• os → ["darwin"]

Since metadata is an empty string, the check at assistant/src/config/skills.ts:309 (if (metadataRaw)) evaluates to false and skips JSON parsing entirely. The metadata field on the skill will be undefined.

This causes two problems:

1. The os: ["darwin"] constraint is lost — the skill will appear available on all platforms instead of only macOS. The OS eligibility check at assistant/src/config/skills.ts:149-158 will never fire because vellum.os is undefined.
2. The emoji 🎥 is lost — the skill won't display its intended icon.

Every other skill in the codebase uses single-line JSON for metadata (e.g., metadata: {"vellum": {"emoji": "🍎", "os": ["darwin"]}} in assistant/src/config/bundled-skills/macos-automation/SKILL.md).

Suggested change

	metadata:
	vellum:
	emoji: "🎥"
	os: ["darwin"]
	metadata: {"vellum": {"emoji": "🎥", "os": ["darwin"]}}

---

Was this helpful? React with 👍 or 👎 to provide feedback.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 30a6047ecf

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Use JSON metadata frontmatter for OS gating

The skill loader only parses metadata as a single-line JSON value (JSON.parse(fields.metadata) in assistant/src/config/skills.ts), but this frontmatter uses nested YAML under metadata:. That means metadata is effectively dropped, including os: ["darwin"], so this macOS-only skill becomes eligible on other platforms and can be surfaced/invoked where screen recording is unavailable.

Useful? React with 👍 / 👎.

[chatgpt-codex-connector]

Avoid claiming requiresRecording works outside QA

This instruction says requiresRecording: true enables recording outside QA mode, but the macOS session starts recording only in the qaMode path and app call sites only provide a ScreenRecorder when qaMode == true. In practice, a non-QA session with only requiresRecording cannot start recording and can hit recording-gate failure behavior, so this guidance teaches an unsupported flow.

Useful? React with 👍 / 👎.