Find recent source reuse by author and creation date (no κ)

docs/research/agent-procedure.md

@@ -121,6 +121,24 @@ that a task is open. A deterministic seed is a reproducibility aid, not novelty.
 Each card supplies a first-step command, expected evidence, a next experiment,
 its connection to the bottleneck, allowed variations and scope-change triggers.
 
+Before taking a source's next question, query its issue/PR identifier directly
+in the dedicated PR body intake command. Author and creation-date filters can
+isolate an account's recent contributions without GitHub's indexed body search:
+
+```sh
+python3 -B scripts/find_source_overlap.py --reference 6537 --pr-author petalas \
+  --created-since 2026-10-27 --include-closed --cache /tmp/source-overlap.json
+```
+
+Use the actual source number, login and UTC cutoff. `--reference` is repeatable
+and excludes decimal/hash substring hits; ordinary positional terms retain
+literal substring semantics. Author/date filters search only PR titles and
+bodies. `--include-closed` adds a separately paginated closed/merged corpus.
+The existing `find_overlap.py` CLI is unchanged. Online intake queries refresh
+PR caches that lack author/creation metadata; offline queries with missing metadata fail
+explicitly rather than reporting no matches. Filtered results are leads, not
+evidence of semantic novelty, and do not replace live claim checks.
+
 **Resolve the generic orientation command before running a historical card.**
 If its `first_step.command` is `python3 -B scripts/research_baseline.py`, use the
 current orientation command in [current priorities](current-priorities.md).

research/source-reference-intake/README.md

@@ -0,0 +1,113 @@
+# Source-reference intake for research overlap
+
+A dedicated `scripts/find_source_overlap.py` CLI accepts reference queries
+and filters PR bodies by author and creation date. This makes an account's recent work
+on a proposed source visible directly in the fetched body corpus. Missing
+filter metadata produces an explicit error, so an old cache cannot silently
+hide a possible lead.
+
+Claim type: workflow tooling with behavioral regression evidence. This
+contribution was produced by Nick Petalas. No mathematical bound, kappa
+improvement or predecessor-result verification is claimed.
+
+## Use
+
+From the repository root:
+
+```sh
+~/git/integer-mult-bounds/.venv/bin/python3.12 -B scripts/find_source_overlap.py \
+  --reference 6537 --pr-author petalas --created-since 2026-10-27 \
+  --include-closed --cache /tmp/source-overlap.json
+```
+
+Use the actual source number, login and UTC cutoff. `--reference` is
+repeatable and applies identifier boundaries to numeric terms. Hashes and
+decimal substrings cannot displace a real reference. Positional terms retain
+their literal search semantics. The date filter uses creation, not update,
+and includes the cutoff day's midnight UTC.
+
+The dedicated intake CLI searches PR titles and bodies only. Closed/merged
+bodies require `--include-closed`, with separate coverage and timestamps.
+Local notes and Discussions are excluded. The existing `scripts/find_overlap.py`
+remains byte-identical to the PR base; its functions are imported without
+mutation. A filtered online query refreshes legacy PR metadata. An offline
+cache missing the requested fields fails with an explanation.
+`--pr-author` takes a login, rather than `@me` or a search expression.
+
+## Evidence and reproduction
+
+Eleven new regression groups cover identifier noise and spelling, literal
+queries, inclusive dates and author case, optional closed bodies, legacy
+offline errors, invalid metadata, creation/update separation, selective
+online refresh and invalid CLI input. The existing 37 overlap groups also
+pass. The registered replay runs all 48 groups in normal and optimized
+Python, without network access.
+
+```sh
+~/git/integer-mult-bounds/.venv/bin/python3.12 -B scripts/research_package.py --package research/source-reference-intake
+~/git/integer-mult-bounds/.venv/bin/python3.12 -I -W error -B research/source-reference-intake/verify.py
+~/git/integer-mult-bounds/.venv/bin/python3.12 -B scripts/check_research_commit.py --package research/source-reference-intake --check-source-history
+```
+
+Expected: `Ran 48 tests`, `OK`, `PASS_SOURCE_REFERENCE_INTAKE` and the
+registered package PASS. [VALIDATION.json](VALIDATION.json) is generated
+evidence bound to the tested script/test hashes. Default replay writes
+nothing persistent. `verify.py --write NEW_PATH` records a fresh execution
+and refuses to overwrite an existing path.
+
+The dedicated CLI's offline query over the retained 587-open-PR snapshot,
+fetched at 2026-11-03T14:10:56.715135+00:00, returned PR6571 as a source-6537 lead.
+This only establishes the tool's retrieval behavior on that snapshot.
+Neither its mathematics nor the source's result was evaluated. The eleven
+synthetic regression groups make the behavior reproducible offline.
+
+## Limits
+
+Reference text is an overlap lead, not evidence of semantic novelty or
+duplication. Work can share a model without citing its source. Current
+claims, package questions and merged work still need inspection. Queries
+do not provide an atomic server snapshot or a seven-day policy decision.
+No selected frontier or frozen research package changes.
+
+## Open next questions
+
+Could a separately scoped provenance index also surface uncited package
+reuse? First compare exact package-path references in PR bodies with a
+complete changed-path inventory. Missing citations, renamed packages and
+semantic equivalence remain blockers; they must not become automatic
+novelty claims.
+
+## Implementation and credit
+
+The new filters traverse finite cached PR lists and validate dates before
+comparing them. Unknown author/date values fail explicitly. The dedicated
+CLI reuses the existing normalizer, identifier matcher, atomic cache writer
+and PR deduplication functions. Its own PR-only cache refresh retains author
+and creation metadata and preserves older evidence on failed publication.
+The existing CLI and every other research package remain untouched.
+
+Power of Ten rule 3 deviation: the Python cache/search command
+allocates lists and dictionaries at runtime. The new pass adds linear
+storage in the existing validated PR corpus size. A fixed-capacity service
+port would require a separately declared corpus-size contract. Rule 9
+deviation: unittest dispatch and mocks use their fixed test callbacks.
+The replay runs one fixed subprocess with a 90-second timeout. Warning-strict
+normal/optimized execution, AST compilation and the full affected suite
+provide diagnostics; Ruff is unavailable in the prescribed environment.
+
+Douglas Colkitt's overlap helper and its credited repository contributors
+retain their attribution. Existing identifier matching and cache/pagination
+machinery are reused. The new dedicated query and metadata filters, regression
+fixtures and integration notes were produced by Nick Petalas. Apache-2.0.
+Independent human review is unclaimed.
+
+## CI compatibility repair
+
+At head `e55541991331d578d88660dcbb34af60506a534d`, the initial hosted run
+passed this package and four other jobs, but the historical overlap-path-index
+replay failed with `source blob provenance mismatch: scripts/find_overlap.py`.
+The shared-file edit in this PR caused that job to be selected. The correction
+restores the exact PR-base helper bytes and moves the feature to the dedicated
+CLI. The historical package and its source pins are unchanged. A regression
+checks that the new CLI does not mutate imported legacy functions or record
+shapes. New hosted confirmation remains pending.

research/source-reference-intake/SOURCES.json

@@ -0,0 +1,18 @@
+{
+  "schema_version": 1,
+  "base_commit": "7c49fe32c55de02c2c873655c804b16e4c3fef0a",
+  "files": [
+    {
+      "path": "certificates/selected-result.json",
+      "usage": "provenance",
+      "git_blob": "d59b5035e7b4cbaa9d688a1d32dd0d6eb79c466a",
+      "sha256": "9b8d0eea95b3bf0dfcb7938a732a6439b87a2685dec2a5ab9e525b3b34debbfa"
+    },
+    {
+      "path": "docs/research/agent-task-catalog.json",
+      "usage": "provenance",
+      "git_blob": "3404fbdd5e7529a0674eef26114b5071e4586cec",
+      "sha256": "ead6cdc9d0af6d58e8a49744d8ec8c7c15e13d2ba93c276c434bcf9b51b93a81"
+    }
+  ]
+}

research/source-reference-intake/TASK.json

@@ -0,0 +1,90 @@
+{
+  "schema_version": 1,
+  "task_id": "replay-runtime",
+  "priority_revision": 8,
+  "base_commit": "7c49fe32c55de02c2c873655c804b16e4c3fef0a",
+  "original_card": {
+    "id": "replay-runtime",
+    "question": "Can one measured replay bottleneck be sped up without dropping coverage?",
+    "starting_points": [
+      "scripts/run_isolated_tests.py",
+      "scripts/verify_replay_coverage.py",
+      "certificates/selected-result.json"
+    ],
+    "methods": [
+      "Profile one isolated replay and optimize the measured bottleneck.",
+      "Design source-bound reuse with explicit dependency invalidation and negative controls."
+    ],
+    "acceptance_gate": "Report before/after wall time on the same fixture, retained checks and invalidation behavior.",
+    "first_step": {
+      "command": [
+        "python3",
+        "-B",
+        "scripts/research_selected_baseline.py"
+      ],
+      "expected": "PASS_RESEARCH_BASELINE_BINDINGS_ONLY",
+      "inspect": "Resolve the current selected review, separate complex/ordinary BIT/admitted BIT/outer budgets, and compare the card against live handoffs before starting. Historical ceilings require exact source/model equivalence.",
+      "scope": "Source-binding orientation only; not a scientific replay or all-size proof. Materialize selected inputs if this checkout is sparse."
+    },
+    "first_experiment": "Measure one unchanged fixture, preserve check coverage, and test one optimization with invalidation controls.",
+    "bottleneck_link": "Faster trustworthy replays increase review throughput and let mechanism candidates be checked sooner.",
+    "allowed_variations": [
+      "Profile one stage or design one source-bound cache boundary."
+    ],
+    "scope_change_triggers": [
+      "Skipping coverage or comparing different fixtures to claim speedup."
+    ],
+    "overlap_terms": [
+      "replay runtime",
+      "cache invalidation",
+      "isolated tests"
+    ]
+  },
+  "scope_status": "changed",
+  "actual_scope": "Add a dedicated PR-source intake CLI with reference, author and creation-date queries while leaving scripts/find_overlap.py byte-identical to the PR base.",
+  "scope_change_reason": "A source-reuse collision was already represented in the complete body cache but was missed by a targeted indexed search and a broad task query. This is an original coordination-tooling improvement, not another predecessor verification or a scientific replay speedup.",
+  "bottleneck_link": "Coordination side result: expose recent source reuse before spending work on a target. It changes no multiplication bound or supplier obligation.",
+  "budget_minutes": 60,
+  "stop_rule": "Stop after one working source-specific query with passing regression coverage and explicit cache-coverage limits. Do not extend to semantic novelty detection.",
+  "outcome": "completed",
+  "evidence": "Eleven new behavioral regression groups and the 37 existing overlap-named groups pass. The standalone CLI retrieves PR6571 from the retained body corpus. VALIDATION.json records generated hash-bound local execution.",
+  "author": "Nick Petalas",
+  "ai_assistance": "Independent human and formal proof-kernel review are unclaimed.",
+  "open_questions": [
+    {
+      "question": "Can a separately scoped package-provenance index surface uncited reuse without treating missing references as novelty?",
+      "why_it_matters": "Reference matching cannot detect a shared model whose source is not cited.",
+      "first_step": "Compare exact package-path references in PR bodies with a complete changed-file inventory and report coverage explicitly.",
+      "blocked_by": "Missing citations, renames and semantic equivalence require separate evidence; no automatic novelty claim is justified."
+    }
+  ],
+  "claim_type": "workflow tooling with behavioral regression evidence",
+  "claim_url": "https://github.com/CrocSwap/integer-mult-bounds/discussions/1166#discussioncomment-19375429",
+  "scope_update_url": "https://github.com/CrocSwap/integer-mult-bounds/discussions/1166#discussioncomment-19375643",
+  "originality": "The CLI integration, author/creation metadata handling and regression fixtures are new. Existing numeric-reference matching, cache paging and source-history infrastructure retain their predecessor credit.",
+  "unrun_obligations": [
+    "Independent human review",
+    "Semantic novelty or duplication decisions",
+    "All mathematical predecessor replays and selected cold replay"
+  ],
+  "orientation": {
+    "command": "/Users/nick/git/integer-mult-bounds/.venv/bin/python3.12 -B scripts/research_selected_baseline.py",
+    "output": "PASS_RESEARCH_BASELINE_BINDINGS_ONLY",
+    "canonical_main": "7c49fe32c55de02c2c873655c804b16e4c3fef0a",
+    "priority_revision": 8,
+    "selected_pr": 6030
+  },
+  "overlap": {
+    "coverage": "Live handoffs, active Discussions, open/account PR lists, complete cached open-PR/Discussion pagination, main-tree distinctive and general-form searches, and merged author/filter searches.",
+    "limits": "Arbitrary diffs, uncited semantic counterparts and unpublished work are not fully covered."
+  },
+  "ci_repair": {
+    "failed_head": "e55541991331d578d88660dcbb34af60506a534d",
+    "run": "https://github.com/CrocSwap/integer-mult-bounds/actions/runs/41652301463",
+    "job": "replay (research/overlap-path-index)",
+    "error": "source blob provenance mismatch: scripts/find_overlap.py",
+    "cause": "The shared-file edit in this PR invalidated the historical consumer.",
+    "correction": "Restore the shared helper to the exact PR-base bytes and move the feature into scripts/find_source_overlap.py. No other research package or source pin is edited.",
+    "hosted_status": "pending"
+  }
+}

research/source-reference-intake/VALIDATION.json

@@ -0,0 +1,35 @@
+{
+  "claim_type": "workflow tooling with behavioral regression evidence",
+  "command": [
+    "/Users/nick/git/integer-mult-bounds/.venv/bin/python3.12",
+    "-I",
+    "-W",
+    "error",
+    "-B",
+    "-m",
+    "unittest",
+    "discover",
+    "-s",
+    "tests",
+    "-p",
+    "test_research_overlap_*.py",
+    "-q"
+  ],
+  "created_utc": "2026-11-03T14:39:20.279188+00:00",
+  "exit_code": 0,
+  "output": "----------------------------------------------------------------------\nRan 48 tests in 1.817s\n\nOK\n",
+  "platform": "macOS-27.0.1-arm64-arm-64bit",
+  "python": "3.12.15 (main, Oct 14 2026, 00:55:37) [Clang 23.1.3 ]",
+  "seconds": 1.900019499938935,
+  "sha256": {
+    "scripts/find_overlap.py": "c48d121d1fe4996a4c232b0d713d9163a1e47777ea3efe01e39bc36cec9aa053",
+    "scripts/find_source_overlap.py": "81788b200f4e2a195420016c289cc8387d3d7b531afdb60c2cb2f8aff3d489ef",
+    "scripts/research_package.py": "e1256016e7f51827d04fa2179255989f639ae1d97afaa9c16265f85a0c6abd70",
+    "tests/test_research_overlap_cache.py": "e49cf1c1707a16c0181f642ab8d969eb8e8689262277b42dba343d0a13295bad",
+    "tests/test_research_overlap_closed.py": "2a5d28e13fe57191702d86630763fdfd33842696077837d225c5677aea819b03",
+    "tests/test_research_overlap_intake.py": "152138d48ef155d33c4cb375a34200a98200b1ac5e1f073cf848096e1b631f7a",
+    "tests/test_research_overlap_local.py": "e7deb16c3f8d14523d31b568a7ea58f683cc56489af2009b2f5cada58a94c2da",
+    "tests/test_research_overlap_pagination.py": "b83e490c45f76b70eb394dd9706ff0b2f627211efe975909f9e1b40d8995f99c",
+    "tests/test_research_overlap_terms.py": "7dd29a1b44ab0c80b8b58a6a1eb7cff810dda942019754029b3bc38c9ff3e601"
+  }
+}

research/source-reference-intake/VERIFY.json

@@ -0,0 +1,44 @@
+{
+  "schema_version": 1,
+  "status": "ready",
+  "python": "3.12",
+  "requirements": [],
+  "inputs": [
+    "scripts/find_source_overlap.py",
+    "scripts/find_overlap.py",
+    "scripts/research_package.py",
+    "tests/test_research_overlap_cache.py",
+    "tests/test_research_overlap_closed.py",
+    "tests/test_research_overlap_intake.py",
+    "tests/test_research_overlap_local.py",
+    "tests/test_research_overlap_pagination.py",
+    "tests/test_research_overlap_terms.py"
+  ],
+  "commands": [
+    {
+      "argv": [
+        "{python}",
+        "-I",
+        "-W",
+        "error",
+        "-B",
+        "{package}/verify.py"
+      ],
+      "timeout_seconds": 120,
+      "checks": "The focused suite: eleven dedicated intake groups and 37 legacy overlap groups, including legacy-module isolation and failed-refresh evidence preservation."
+    },
+    {
+      "argv": [
+        "{python}",
+        "-I",
+        "-O",
+        "-W",
+        "error",
+        "-B",
+        "{package}/verify.py"
+      ],
+      "timeout_seconds": 120,
+      "checks": "The same complete suite with optimized Python, propagating optimization into the test subprocess."
+    }
+  ]
+}

research/source-reference-intake/verify.py

@@ -0,0 +1,48 @@
+"""Replay source-intake behavior and the focused overlap regression suite.
+
+Produced by Nick Petalas. Apache-2.0.
+"""
+
+import argparse
+from datetime import datetime, timezone
+from hashlib import sha256
+import json
+from pathlib import Path
+import platform
+import subprocess
+import sys
+import time
+
+
+def main():
+    parser = argparse.ArgumentParser(description=__doc__)
+    parser.add_argument("--write", type=Path, help="Write fresh evidence; refuse overwrite")
+    args = parser.parse_args()
+    root = Path(__file__).resolve().parents[2]
+    command = [sys.executable, "-I", "-W", "error", "-B"]
+    if sys.flags.optimize:
+        command.append("-O")
+    command.extend(["-m", "unittest", "discover", "-s", "tests",
+                    "-p", "test_research_overlap_*.py", "-q"])
+    started = time.perf_counter()
+    result = subprocess.run(command, cwd=root, text=True, capture_output=True,
+                            timeout=90, check=True)
+    elapsed = time.perf_counter() - started
+    print(result.stdout + result.stderr, end="")
+    if args.write is not None:
+        inputs = ["scripts/find_source_overlap.py", "scripts/find_overlap.py", "scripts/research_package.py",
+                  *sorted(str(path.relative_to(root)) for path in
+                          (root / "tests").glob("test_research_overlap_*.py"))]
+        evidence = {"claim_type": "workflow tooling with behavioral regression evidence",
+                    "created_utc": datetime.now(timezone.utc).isoformat(),
+                    "python": sys.version, "platform": platform.platform(),
+                    "command": command, "seconds": elapsed, "exit_code": result.returncode,
+                    "output": result.stdout + result.stderr,
+                    "sha256": {name: sha256((root / name).read_bytes()).hexdigest() for name in inputs}}
+        with args.write.open("x", encoding="utf-8") as stream:
+            stream.write(json.dumps(evidence, indent=2, sort_keys=True) + "\n")
+    print("PASS_SOURCE_REFERENCE_INTAKE")
+
+
+if __name__ == "__main__":
+    main()