docs(claude): drop Skill-vs-Command split from testing-command table

[WilcoLouwerse]

Summary

docs/claude/commands.md had a "Type" column that labelled some testing entries as Skill and others as Command (1 agent), perpetuating a split that no longer exists in Claude Code.

• Anthropic merged custom commands into skills (docs): "Custom commands have been merged into skills. A file at .claude/commands/deploy.md and a skill at .claude/skills/deploy/SKILL.md both create /deploy and work the same way."
• This workspace was always skill-only — no .claude/commands/ directory has ever existed.

The column was actually carrying two facts: skill-vs-command (false split) and parallel-agent count (real info). Renamed the column to Agents so the agent-count info is preserved and added a one-line note above the table explaining the merge.

Test plan

• Skim docs/claude/commands.md §Testing Commands → table renders, intro reads cleanly, agent counts (1, 8, "1 or 6") are unchanged
• No other doc in .github/docs/ still draws a Skill-vs-Command type distinction

🤖 Generated with Claude Code

[MWest2020]

Code Review

Overview

Docs-only change to docs/claude/commands.md (+17 / -15). Replaces the "Type" column in the Testing Commands table (which mixed two facts: skill-vs-command + parallel-agent count) with a single numeric Agents column, and adds a one-line intro explaining that custom commands have been merged into skills upstream.

What's good

• Premise is correct. Anthropic's skills docs state: "Custom commands have been merged into skills. A file at .claude/commands/deploy.md and a skill at .claude/skills/deploy/SKILL.md both create /deploy and work the same way." The split this table was drawing is genuinely obsolete.
• Information preserved. The agent-count fact (1, 8, "1 or 6") survives the rename — nothing lost.
• Boring & auditable. Minimal diff, single file, no behavior change. Reads cleanly in the rendered markdown.
• Intro paragraph is well-scoped — explains why in one sentence with a parenthetical, no over-explanation.

Issues / risks

1. The same Skill-vs-Command split still exists in the file this section links to.
docs/claude/testing.md (linked on line 53: "see testing.md") has a near-identical table with the same false distinction, lines 9–26:

```
| Command / Skill | Type | Focus | Agents | Output |
| `/test-counsel` | Skill | ...
| `/test-functional` | Command | ...
```

The PR's own test plan item #2 ("No other doc in `.github/docs/` still draws a Skill-vs-Command type distinction") is not satisfied — a reader who follows the link from commands.md immediately encounters the exact split the PR claims to retire. The rationale applies identically to testing.md and ideally lands in the same PR (or an immediate follow-up).

2. `"1 or 6"` reads awkwardly in a numeric-looking column.
Minor. Now that the column header is `Agents` and every other cell is an integer, `"1 or 6"` for `/test-app` stands out. Consider `1–6` or `1 / 6` (range vs. mode-dependent). Not blocking.

3. Test plan checkboxes are unchecked.
Both items in the PR body are still `[ ]` — likely just a template artifact, but flagging since #2 (as above) is actually false against the current branch.

Style / conventions

• Header rename + alignment dashes are correct.
• No trailing-whitespace or CRLF noise (clean line endings).
• Matches the table style of the OpenSpec / App Management tables higher up in the same file (3-column `Command | | Description`).

Security / performance

N/A. Docs-only change.

Recommendation

Approve with one follow-up: apply the same column rewrite to `docs/claude/testing.md` (lines 9–26) — either in this PR or as an immediate follow-up — so the linked reference doesn't immediately contradict the cleanup.

[WilcoLouwerse]

Addressed in dbd824c:

• testing.md Quick Comparison — dropped the "Type" column (Skill / Command), renamed to Command | Focus | Agents | Output, and added the same one-line intro as commands.md (skills-only, custom commands merged upstream).
• "1 or 6" → "1 / 6" — applied in commands.md (Testing Commands table) and in the testing.md /test-counsel vs /test-app comparison table, so the Agents column reads cleanly as two modes rather than a range.
• Grepped the rest of docs/ for any residual Skill-vs-Command split — none left.

Test-plan checkbox #2 is now actually true on the branch.