Code-Graph-RAG Files

Highlights

• CI Improvements: Automated linting and formatting on pull requests via pre-commit.ci, and a Sonar zero-issues merge gate have been implemented.
• Graph Indexing: The name property is now indexed for improved graph read-path lookups. SQL routines are also indexed to resolve calls that name their target in a string.
• Duplicate Code Detection: An AST-based duplicate code detection system has been introduced, alongside fixes for duplicate detection across multiple repositories.
• Agentic QA: A new agentic QA benchmark harness and indexing-time benchmark have been added.
• Enhanced Documentation: Several documentation updates clarify various aspects including piwheels failures, relationship columns, macOS OpenSSL requirements, and the use of --clean with backfilling.

What's Changed

• ci: configure pre-commit.ci to auto-fix lint and format on PRs by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1389
• fix: index the name property for graph read-path lookups by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1384
• feat: return relationship context from generated caller queries by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1386
• fix(tests): make the well-known-symbols dead-code e2e observable and passing by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1393
• ci: promote every release highlight to a news entry by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1391
• feat: agentic QA benchmark harness and indexing-time bench by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1388
• feat: AST-based duplicate code detection (cgr duplicates + agent tool) by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1392
• fix(readme): render every latest-release highlight in Latest News by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1395
• feat: index SQL routines and resolve calls that name their target in a string by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1394
• ci: gate merges on a 5/5 Greptile review of the head commit by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1396
• feat: duplicates eval plus fixes for five bugs found dogfooding clone detection across 14 repos by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1404
• refactor: reduce cognitive complexity of _similar_groups and _emit_duplicates below Sonar threshold by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1406
• ci: add Sonar zero-issues merge gate by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1407
• fix: warn about public compose ports when the stack is already running (#1380) by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1408
• test: skip grammar-dependent tests when optional tree-sitter grammars are absent (#1371) by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1411
• fix: make cgr help work with typer's vendored click (#1409) by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1412
• test: skip non-grammar optional-dependency tests on a base install and add a base-install CI job (#1410) by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1413
• fix: C# ref write-back sees through interface and abstract dispatch with a sole implementer (#1356) by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1414
• fix: C# internal namespace imports target the referenced Module nodes (#1347) by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1416
• docs: explain the piwheels Bookworm failure and document the Python 3.12 floor by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1417
• docs: clarify the relationship-column warning in the query_graph description by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1418
• docs: document macOS pymgclient source-build OpenSSL requirements by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1419
• feat: fingerprint macro_rules definitions from their token trees by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1420
• docs: accept and document the same-qn bodied module node merge with a pin test by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1422
• docs: accept and pin the shadow-unaware nested factory identity resolution as a documented limitation by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1423
• feat: clickable editor links in the duplicates report plus --open side-by-side diff by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1421
• docs: exclude globs must be quoted and cover the whole path by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1424
• feat: parse markdown headings into a nested Section graph by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1428
• fix: re-parse ast-grep and document files on watched edits by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1430
• docs: add the claimed label for coordinating concurrent agent sessions by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1431
• feat: agent-level trust boundary for web_search via a sandboxed research sub-agent by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1425
• fix: keep ast-grep modules distinct when file stems collide across extensions by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1435
• fix: keep uv.lock's project version in step with releases by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1433
• fix: validate labels.yml on the PR that changes it rather than after merge by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1440
• test: assert the suffix-distinct module names, not merely that they differ by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1436
• docs: note that document module names carry their extension by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1437
• docs: state what --clean costs when backfilling fingerprints by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1439
• feat: LINKS_TO edges from markdown links to repository files by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1445
• feat: Ruby structure oracle graded against Prism by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1450
• feat: build orchestrator prompt search sections conditionally on registered tools by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1446
• fix: recommend 'cgr start --clean --update-graph' instead of bare --clean by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1444
• fix: skip the live dotnet-trace test when the tool cannot run by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1451
• fix: finding nodes were dropped from the protobuf index by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1453
• test: pin python frontend column conversions to absolute values by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1458
• feat: C++ libclang frontend behind the EmittingFrontend protocol by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1459
• feat: record per-module frontend degradation instead of hiding it by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1464
• docs: how to add a language frontend by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1457
• fix: dispatch EMITTING_FRONTENDS generically by phase by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1461
• fix: bound frontend helper-tool builds so a wedged toolchain degrades by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1463
• fix: fingerprint external tool versions, not just the frontend mode by @vitali87 in https://github.com/vitali87/code-graph-rag/pull/1466

Full Changelog: https://github.com/vitali87/code-graph-rag/compare/v0.0.720...v0.0.770