[eldonm]

Problem and change

Walker construction repeatedly introspected Pydantic private factories and instance descriptors, while large graph-backed reads lacked an adapter-level way to retrieve selected values or verify bounded one-hop binding facts without hydrating every entity.

• Discover decorated Walker methods from declared class functions, preserving overrides and per-instance state without evaluating arbitrary descriptors. Avoid repeated builtin-factory signature parsing.
• Add public selected-value reads with a portable default, native PostgreSQL projections, missing/null distinction and full filter/order semantics.
• Add immutable LiteralBindingFact and GraphBindingSummary contracts with optional bounded PostgreSQL aggregation. Unsupported adapters return None; capped results remain explicitly incomplete. These report storage facts, never authorization grants.
• Preserve transaction connection boundaries, cache bypass for partial records and upstream observability.
• Correct compound identity indexes to use root id/entity paths and document the additive public APIs.

Validation

• Full framework suite with coverage: 2,636 passed, 221 optional-service skips.
• Final framework Core/database regression suite: 1,306 passed, 100 optional-adapter skips, including actual PostgreSQL projection, malformed-binding and transaction tests.
• Pinned pre-commit formatting, lint, typing and secret checks pass. The final checks corrected capability-property typing and made transaction projection explicitly use its own find method.
• Downstream integration against a reviewed local wheel passed exact-record and authority checks, and all six 50,000-entry PostgreSQL regression budgets (79–424 ms p95). This is local protected-owner evidence, not hosted or mixed-role capacity qualification.

Review and release boundary

Companion Core PR: V75inc/integral-core#197

No package version change or publication is included. Core retains its current published dependency and compatible fallback; adopting these native optimizations in a release requires publishing this framework change and separately updating Core's pin. Keep both PRs open for review; do not merge during this handoff.

[github-actions]

Benchmark comparison

Threshold: ±25% (informational, does not block merge)

benchmark	baseline (s)	current (s)	delta	status
tests/benchmarks/test_deferred_save_benchmarks.py::test_bench_deferred_save_batched_100	0.027652	0.010997	-60.2%	IMPROVED (-60.2%)
tests/benchmarks/test_deferred_save_benchmarks.py::test_bench_immediate_save_100	0.027265	0.010770	-60.5%	IMPROVED (-60.5%)
tests/benchmarks/test_graph_projection_benchmarks.py::test_warm_projection_capture[hierarchy]	0.004792	0.002051	-57.2%	IMPROVED (-57.2%)
tests/benchmarks/test_graph_projection_benchmarks.py::test_warm_projection_capture[ontology]	0.004515	0.002040	-54.8%	IMPROVED (-54.8%)
tests/benchmarks/test_graph_projection_benchmarks.py::test_warm_projection_capture[relationships]	0.004537	0.001998	-56.0%	IMPROVED (-56.0%)
tests/benchmarks/test_graph_projection_benchmarks.py::test_warm_rich_projection_capture[hierarchy]	0.003863	0.001744	-54.9%	IMPROVED (-54.9%)
tests/benchmarks/test_graph_projection_benchmarks.py::test_warm_rich_projection_capture[ontology]	0.003808	0.001708	-55.1%	IMPROVED (-55.1%)
tests/benchmarks/test_graph_projection_benchmarks.py::test_warm_rich_projection_capture[relationships]	0.004045	0.001724	-57.4%	IMPROVED (-57.4%)
tests/benchmarks/test_jsondb_benchmarks.py::test_bench_jsondb_batched_saves_500	0.503749	0.375118	-25.5%	IMPROVED (-25.5%)
tests/benchmarks/test_jsondb_benchmarks.py::test_bench_jsondb_count_empty_query	0.002080	0.001109	-46.7%	IMPROVED (-46.7%)
tests/benchmarks/test_jsondb_benchmarks.py::test_bench_jsondb_count_filtered	0.099028	0.051101	-48.4%	IMPROVED (-48.4%)
tests/benchmarks/test_jsondb_benchmarks.py::test_bench_jsondb_find_filtered	0.102074	0.050745	-50.3%	IMPROVED (-50.3%)
tests/benchmarks/test_jsondb_benchmarks.py::test_bench_jsondb_save_throughput	0.001701	0.001162	-31.7%	IMPROVED (-31.7%)
tests/benchmarks/test_postgres_benchmarks.py::test_bench_postgres_find_many_bulk	0.001119	0.000491	-56.1%	IMPROVED (-56.1%)
tests/benchmarks/test_postgres_benchmarks.py::test_bench_postgres_traverse_depth	0.000658	0.000276	-58.0%	IMPROVED (-58.0%)
tests/benchmarks/test_sqlite_benchmarks.py::test_bench_sqlite_count_empty	0.000260	0.000113	-56.7%	IMPROVED (-56.7%)
tests/benchmarks/test_sqlite_benchmarks.py::test_bench_sqlite_count_fallback_via_regex	0.030648	0.015038	-50.9%	IMPROVED (-50.9%)
tests/benchmarks/test_sqlite_benchmarks.py::test_bench_sqlite_count_pushdown	0.001514	0.000701	-53.7%	IMPROVED (-53.7%)
tests/benchmarks/test_sqlite_benchmarks.py::test_bench_sqlite_find_fallback_via_regex	0.030852	0.015282	-50.5%	IMPROVED (-50.5%)
tests/benchmarks/test_sqlite_benchmarks.py::test_bench_sqlite_find_pushdown	0.000430	0.000201	-53.3%	IMPROVED (-53.3%)
tests/benchmarks/test_sqlite_benchmarks.py::test_bench_sqlite_sort_limit_pushdown	0.002620	0.001302	-50.3%	IMPROVED (-50.3%)