[SenolIsci]

Documents the new pass2.domain_range_policy key (off | warn | strict, default warn) in docs/architecture.md alongside Pass 2 validation. Docs-only; companion to the implementation PR. (CLAUDE.md D60 is gitignored, so it was updated locally only.)

🤖 Generated with Claude Code

Summary by Sourcery

Documentation:

• Document the optional Pass 2 domain/range validation policy, including its off, warn, and strict modes, configuration behavior, and validation outcomes.

[sourcery-ai]

Reviewer's guide (collapsed on small PRs)

Reviewer's Guide

Adds architecture documentation for the new pass2.domain_range_policy setting, covering configuration, inheritance-aware validation semantics, warning versus strict enforcement, retry/recovery behavior, and the rationale for the default.

Flow diagram for Pass 2 domain/range policy enforcement

flowchart TD
    A[_extract_chunk] --> B[_normalize_scalars]
    B --> C[_apply_domain_range_policy]
    C --> D{domain_range_policy}
    D -->|off| E[Continue existing validation]
    D -->|warn| F[Keep remaining violations and tag domain_range_violation]
    D -->|strict| G[validate_extraction]
    G -->|Violations remain| H[Single LLM retry]
    H --> I[_partial_recover drops violating edges]
    I --> J[Clean up hallucinated anchor nodes]
    G -->|No violations| K[Continue extraction]
    F --> L[Assemble output]
    E --> L
    J --> L
    K --> L

Loading

File-Level Changes

Change	Details	Files
Document the configurable Pass 2 domain/range validation policy and its runtime behavior.	Describe off, warn (default), and strict modes, including inheritance-aware endpoint matching. Document configuration lookup, profile-level opt-in, invalid-value fallback, and the intentional omission from mykg_config.yaml. Explain validation, retry, partial-recovery, metadata tagging, and edge cleanup behavior. Record rationale for the default and clarify that orphan_inferred edges are excluded.	docs/architecture.md

---

Tips and commands

Interacting with Sourcery

• Trigger a new review: Comment @sourcery-ai review on the pull request.
• Continue discussions: Reply directly to Sourcery's review comments.
• Generate a GitHub issue from a review comment: Ask Sourcery to create an
  issue from a review comment by replying to it. You can also reply to a
  review comment with @sourcery-ai issue to create an issue from it.
• Generate a pull request title: Write @sourcery-ai anywhere in the pull
  request title to generate a title at any time. You can also comment
  @sourcery-ai title on the pull request to (re-)generate the title at any time.
• Generate a pull request summary: Write @sourcery-ai summary anywhere in
  the pull request body to generate a PR summary at any time exactly where you
  want it. You can also comment @sourcery-ai summary on the pull request to
  (re-)generate the summary at any time.
• Generate reviewer's guide: Comment @sourcery-ai guide on the pull
  request to (re-)generate the reviewer's guide at any time.
• Resolve all Sourcery comments: Comment @sourcery-ai resolve on the
  pull request to resolve all Sourcery comments. Useful if you've already
  addressed all the comments and don't want to see them anymore.
• Dismiss all Sourcery reviews: Comment @sourcery-ai dismiss on the pull
  request to dismiss all existing Sourcery reviews. Especially useful if you
  want to start fresh with a new review - don't forget to comment
  @sourcery-ai review to trigger a new review!

Customizing Your Experience

Access your dashboard to:

• Enable or disable review features such as the Sourcery-generated pull request
  summary, the reviewer's guide, and others.
• Change the review language.
• Add, remove or edit custom review instructions.
• Adjust other review settings.

Getting Help

• Contact our support team for questions or feedback.
• Visit our documentation for detailed guides and information.
• Keep in touch with the Sourcery team by following us on X/Twitter, LinkedIn or GitHub.

[sourcery-ai]

Hey - I've found 2 issues

Prompt for AI Agents

Please address the comments from this code review:

## Individual Comments

### Comment 1
<location path="docs/architecture.md" line_range="344" />
<code_context>
+**Optional domain/range check (`pass2.domain_range_policy`).** Historically Pass 2 validated only edge type, `from`/`to` node IDs and confidence; a property's `domain`/`range` were prompt hints. The `pass2.domain_range_policy` key (`off` | `warn` | `strict`, default `warn`, exposed as `config.PASS2_DOMAIN_RANGE_POLICY`) adds an inheritance-aware check: endpoint types come from the response's nodes plus `prior_nodes`, and a subclass (walking each concept's `parent` chain) satisfies a domain or range. Edges whose property or endpoint type is unknown are left to the existing checks. The key is read but deliberately not surfaced in either `mykg_config.yaml` (same precedent as `llm.temperature_unsupported_prefixes`); opt in by adding it under a profile's `pass2:` block. An invalid value logs a warning and falls back to `warn`.
</code_context>
<issue_to_address>
**issue:** The documentation claims that `pass2.domain_range_policy` is read and exposed as `config.PASS2_DOMAIN_RANGE_POLICY`, but the repository has no such configuration constant or policy implementation, so adding the documented key currently has no effect.

**Triggers:** When this documentation is merged without the companion implementation being available in the same released codebase.

**Suggested fix:** Merge the implementation that reads and applies the setting before documenting it, or describe the key as unreleased and point to the required implementation version.
</issue_to_address>

### Comment 2
<location path="docs/architecture.md" line_range="349" />
<code_context>
+`_apply_domain_range_policy` runs in `_extract_chunk` after `_normalize_scalars`, on both the first attempt and the retry:
+
+- **`off`** — no-op.
+- **`warn`** (default) — an edge whose endpoint types match `(range, domain)` exactly is swapped (logged). Remaining violators are kept and tagged `domain_range_violation: true`, which reaches `edge_metadata.json` / `edges.jsonl` via the assembler's deepcopy of the first occurrence.
+- **`strict`** — same swap, then `validate_extraction` reports remaining violations as errors, triggering the existing single LLM retry with those errors as feedback. If the retry still violates, `_partial_recover` drops the violating edges *before* its hallucinated-anchor pass, so their anchor nodes are cleaned up too. Both pieces are required: `validate_extraction` only reports; `_partial_recover` is the only place that enforces.
+
</code_context>
<issue_to_address>
**issue (broader_impact):** A domain/range violation flag from a later occurrence of a duplicate edge is discarded when the assembler's first occurrence is clean, because edge deduplication deep-copies only the first occurrence and does not merge this top-level flag; the violation therefore does not reliably reach `edge_metadata.json` or `edges.jsonl`.

**Triggers:** When the same `(type, from, to)` edge is extracted in multiple files or chunks and the first occurrence does not violate domain/range while a later occurrence does.

**Suggested fix:** Merge `domain_range_violation` across all edge occurrences with an OR operation, or store it in a field that the deduplication merge logic explicitly preserves.
</issue_to_address>

Sourcery assessment

Approval pending. 2 findings to address first.

Blocking findings: docs/architecture.md:344, docs/architecture.md:349

---

Sourcery is free for open source - if you like our reviews please consider sharing them ✨

• X
• Mastodon
• LinkedIn
• Facebook

[sourcery-ai]

issue: The documentation claims that pass2.domain_range_policy is read and exposed as config.PASS2_DOMAIN_RANGE_POLICY, but the repository has no such configuration constant or policy implementation, so adding the documented key currently has no effect.

Triggers: When this documentation is merged without the companion implementation being available in the same released codebase.

Suggested fix: Merge the implementation that reads and applies the setting before documenting it, or describe the key as unreleased and point to the required implementation version.

[sourcery-ai]

issue (broader_impact): A domain/range violation flag from a later occurrence of a duplicate edge is discarded when the assembler's first occurrence is clean, because edge deduplication deep-copies only the first occurrence and does not merge this top-level flag; the violation therefore does not reliably reach edge_metadata.json or edges.jsonl.

Triggers: When the same (type, from, to) edge is extracted in multiple files or chunks and the first occurrence does not violate domain/range while a later occurrence does.

Suggested fix: Merge domain_range_violation across all edge occurrences with an OR operation, or store it in a field that the deduplication merge logic explicitly preserves.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[SenolIsci]

Abandoned: domain/range policy feature dropped.