docs: refresh skills tables and drop stale "Optional Files" section

Audit revealed three stale sections — only the obvious ones, not
cosmetic drift:

- CLAUDE.md + README.md: skills tables listed only `challenge-me`
  but skills/ now contains 5 skills (challenge-me, eli5, hot-seat,
  pair-debug, test-me). Expanded both tables.
- docs/creating-skills.md: "Optional Files" section documented
  examples/, scoring.md, context.md conventions that no skill
  actually uses (grep/find: zero hits). Removed the section — can
  re-add when a skill actually needs the convention.

All file paths, symbol refs, env vars, make targets, and module
claims in the other docs verified clean.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Author: mariuspruvot

CLAUDE.md

@@ -68,7 +68,11 @@ Skills are pluggable Claude Code agent definitions in `skills/`. See `skills/SKI
 
 | Skill | Purpose | PR fetch strategy |
 |-------|---------|-------------------|
-| `challenge-me` | Socratic quiz on PR changes | Shallow clone |
+| `challenge-me` | Socratic comprehension quiz on PR changes | `shallow_clone` |
+| `eli5` | Explain Like I'm 5 — vulgarize your own code | `shallow_clone` |
+| `hot-seat` | Architecture Hot Seat — defend design choices under pressure | `shallow_clone` |
+| `pair-debug` | Find the subtle bug Claude injected into your code | `shallow_clone` |
+| `test-me` | Predict whether test cases pass or fail on your code | `shallow_clone` |
 
 ## Code Style
 

README.md

@@ -58,6 +58,10 @@ Skills are pluggable Claude Code agent definitions. Each skill is a self-contain
 | Skill | Description | Duration |
 |-------|-------------|----------|
 | **[challenge-me](skills/challenge-me/)** | Socratic comprehension quiz -- probes whether the PR author truly understands their own changes. Generates 3-5 targeted questions, evaluates answers, and produces a score card. | 5-10 min |
+| **[eli5](skills/eli5/)** | Explain Like I'm 5 -- can you vulgarize your own code? | 5-8 min |
+| **[hot-seat](skills/hot-seat/)** | Architecture Hot Seat -- defend your design choices under pressure. | 5-10 min |
+| **[pair-debug](skills/pair-debug/)** | Pair Debug -- find the subtle bug Claude injected into your code. | 5-12 min |
+| **[test-me](skills/test-me/)** | Test Me -- predict whether test cases pass or fail on your code. | 5-8 min |
 
 Want to create your own? See [Creating Skills](docs/creating-skills.md).
 

docs/creating-skills.md

@@ -228,20 +228,6 @@ Or test via the full helPRs stack:
 
 ---
 
-## Optional Files
-
-Skills can include additional files referenced from `CLAUDE.md`:
-
-| File | Purpose |
-|------|---------|
-| `examples/` | Example inputs/outputs for testing |
-| `scoring.md` | Detailed scoring rubric |
-| `context.md` | Additional domain context |
-
-These must be explicitly referenced in `CLAUDE.md` -- Claude Code will not discover them automatically.
-
----
-
 ## Output Conventions
 
 For `sse_stream` skills, the frontend parses known patterns from the streamed output: