### expert_build/prompts.py
VERDICT: PASS
CORRECTNESS: VALID
SPEC_COMPLIANCE: N/A
ISSUE_COMPLIANCE: N/A
BELIEF_COMPLIANCE: CONSISTENT
TEST_COVERAGE: N/A
INTEGRATION: WIRED
REASONING: The `SUMMARIZE_CODE` prompt template is well-structured and provides clear, professional instructions for an LLM to generate structured notes from source code. It follows the established pattern of other prompts in the file.
---

### expert_build/summarize.py:cmd_summarize
VERDICT: CONCERN
CORRECTNESS: QUESTIONABLE
SPEC_COMPLIANCE: N/A
ISSUE_COMPLIANCE: N/A
BELIEF_COMPLIANCE: CONSISTENT
TEST_COVERAGE: UNTESTED
INTEGRATION: WIRED
REASONING: 
1. **Unreachable PDF Logic**: The truncation warning (lines 77-80) includes a specific suggestion for `.pdf` files: `Consider: code-expert chunk-pdf {source_path}`. However, the `sources` glob (line 27) only includes `*.md` and `*.py`. This means `source_path.suffix == ".pdf"` will never be true in the current implementation, making this a piece of dead/misleading code.
2. **Generic Entry Titles**: The title extraction logic at line 96 (`re.search(r"^#+ (.+)$", summary, ... )`) picks up the first Markdown header. Since `SUMMARIZE_CODE` explicitly requests a `## Overview` header as the first section, every Python file summarized will likely be titled "Overview" in the entry registry. This results in poor UX compared to using the filename or a more specific title.
3. **CLI Name Inconsistency**: The truncation warning refers to the tool as `code-expert`, while existing error messages (line 19) refer to it as `expert-build`. This inconsistency should be resolved to avoid user confusion.
4. **Lack of Tests**: The observation results indicate zero tests for this module. Given the addition of logic for handling different file types and truncation, unit tests should be added to verify that the correct template is selected and that truncation behaves as expected.
---

### SELF_REVIEW
LIMITATIONS: I do not have the content of the `SUMMARIZE` prompt, so I cannot compare it with `SUMMARIZE_CODE` to see if it also suffers from the generic title extraction issue. I also cannot verify the canonical name of the CLI tool (e.g., in `pyproject.toml`) to determine which of `code-expert` or `expert-build` is correct.

### FEATURE_REQUESTS
- Include the content of all referenced prompt templates in observations, even if they aren't modified in the diff.
- Provide the `scripts` or `project.scripts` section of `pyproject.toml` when reviewing CLI-related changes to verify tool names.
