presonus-studiolive-mcp: Skill for Claude Code

Stub and Placeholder Elimination

Purpose

Use this skill to find, classify, and eliminate unfinished code, placeholder logic, fake success responses, empty implementations, TODO-only paths, and user-visible stubs.

The goal is not to delete all TODO comments. The goal is to prevent unfinished behavior from being exposed as real MCP capability.

Applies to

• TypeScript source files
• MCP tool handlers
• adapter methods
• domain conversion functions
• generated capability documentation
• tests and fixtures
• README/examples
• probe scripts

Search targets

Scan for these literal terms case-insensitively:

TODO
FIXME
HACK
XXX
stub
placeholder
not implemented
not_implemented
unimplemented
dummy
fake
temporary
probe_required
not_verifiable
inferred
return {}
return []
return null
return undefined
throw new Error("Not implemented")
throw new Error('Not implemented')

Also inspect suspicious patterns:

return { success: true };
return { ok: true };
return { status: "ok", data: {} };
return { status: "ok", data: [] };
return { capability: "observed" };

Any such response must be backed by real adapter interaction and tests.

Severity classification

Severity	Meaning	Required action
P0	User-visible MCP tool is exposed but not truly implemented.	Implement, hide, or mark as stub/probe_required in all docs and responses.
P1	Incomplete internal logic affects routing, AVB, writes, Fat Channel, state, or safety.	Fix before release or gate behind explicit experimental flag.
P2	Dev-only TODO or incomplete docs/tests that do not mislead users.	Track in issue; not release-blocking unless claim depends on it.
P3	Cosmetic comment or harmless future note.	Leave or clean opportunistically.

Required workflow

1. Run textual searches across the repo.
2. For each hit, inspect the surrounding implementation.
3. Determine whether the hit affects runtime behavior or only documentation.
4. Identify the user-visible surface, if any:
   • MCP tool,
   • CLI command,
   • README claim,
   • generated matrix,
   • release checklist.
5. Classify severity.
6. For P0 and P1, choose exactly one resolution:
   • implement behavior,
   • remove/hide tool from normal inventory,
   • keep exposed but explicitly return stub/probe_required with actionable probe instructions,
   • mark as not_supported.
7. Add or update tests.
8. Update capability and release docs.