.github/skills/indicator-stream/references/provider-selection.md (1)

7-17: Wrap decision tree in fenced code block for consistency.

The ASCII art decision tree (lines 7-17) currently lacks a fenced code block wrapper. Per markdown guidelines, text-based diagrams should be wrapped in ```text blocks for consistency with other code examples and to ensure proper rendering across platforms.

🔎 Proposed fix

  ## Provider decision tree
  
- ```text
+ ```
  Does indicator produce chainable output (single Value)?
  ├─ Yes → Does it need OHLCV data?
  │  ├─ Yes → Does it need TWO synchronized series?
  │  │  ├─ Yes → PairsProvider<IReusable, TResult>
  │  │  └─ No → ChainProvider<IQuote, TResult>
  │  └─ No → ChainProvider<IReusable, TResult>
  └─ No → Does output transform quote to quote?
     ├─ Yes → QuoteProvider<IQuote, TResult>
     └─ No → StreamHub<TIn, TResult> (ISeries output)

</details>

</blockquote></details>
<details>
<summary>.github/skills/indicator-series/references/decision-tree.md (1)</summary><blockquote>

`1-138`: **Catalog skill document is comprehensive and actionable.**

The CatalogListingBuilder pattern, method naming, and anti-patterns are well-explained. The critical rule on line 71 (`.WithMethodName()` placement) and line 84–85 (`isReusable` constraints) are highlighted appropriately and align with learnings.

One readability suggestion: The code block at lines 18–36 demonstrates indicator construction alongside catalog patterns. While not incorrect, clarifying that lines 20–36 show *how* an indicator uses the catalog (not the catalog file itself) would prevent potential confusion about which code belongs where.

</blockquote></details>
<details>
<summary>AGENTS.md (1)</summary><blockquote>

`132-181`: **Skills framework sections are well-organized; consider clarifying structure for usage examples.**

The new sections introducing Agent Skills (lines 132–153 "Skills for indicator development" and 154–167 "Skills for focused assistance") successfully replace references to old instruction files with the new skills-based framework. Content is clear and aligns with the PR objectives. The skill descriptions are actionable and include appropriate use cases.

Minor observation: Lines 168–181 provide usage examples but flow directly from the skills table without an explicit section header. Adding a header like "## Usage examples" would improve clarity and help readers distinguish between the reference table (lines 154–167) and practical examples (lines 168–181).

</blockquote></details>
<details>
<summary>.github/skills/performance-testing/references/benchmark-patterns.md (1)</summary><blockquote>

`137-143`: **Minor: Consider documenting BenchmarkConfig.cs as single source of configuration.**

The learnings indicate that `BenchmarkConfig.cs` should serve as the "single source of benchmark configuration" with exporters (GitHub markdown and JSON), columns (Mean, Error, StdDev), and logger settings. While your memory profiling section is solid, briefly mentioning this configuration file would help developers understand the centralized setup approach.

</blockquote></details>

</blockquote></details>

<!-- This is an auto-generated comment by CodeRabbit for review status -->