open-code-review-toolkit 0.7.0__tar.gz → 0.8.0__tar.gz

{open_code_review_toolkit-0.7.0 → open_code_review_toolkit-0.8.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: open-code-review-toolkit
-Version: 0.7.0
+Version: 0.8.0
 Summary: Unofficial GitLab CI integration layer for Open Code Review
 Project-URL: Homepage, https://github.com/xeonvs/open-code-review-toolkit
 Project-URL: Repository, https://github.com/xeonvs/open-code-review-toolkit
@@ -254,17 +254,17 @@ Stable distributions are published to [PyPI](https://pypi.org/project/open-code-
 
 ## How reviews evolve
 
-On a successful rerun, the toolkit replaces untouched OCR-only notes instead of accumulating stale reviews. A human reply transfers that discussion to the team: the conversation is preserved and a matching finding is suppressed. Reply with `/ocr suppress` to keep a discussion open without future repeats, or `/ocr resolve` to suppress it and resolve the discussion after the next successful posting transaction.
+On a successful rerun, the toolkit replaces untouched OCR-only notes instead of accumulating stale reviews. A human reply transfers that discussion to the team: the conversation is preserved and a matching finding is suppressed. Reply with `/ocr suppress` or `@<live-bot-username> suppress` to keep a discussion open without future repeats; use the corresponding `resolve` command to resolve it after the next successful posting transaction. For example, a bot named `mr.bot` accepts the exact reply `@mr.bot resolve`.
 
 Suppression uses both the GitLab diff position and a stable finding fingerprint, so ordinary line shifts do not normally bring the same bug back. A materially changed finding can still receive a new discussion. See [GitLab review operations](docs/operations.md) for the complete lifecycle, posting modes, permissions, failure behavior, and Mermaid state diagram.
 
-After every current review note publishes, the GitLab adapter can add a conservative approval bound to receipt v4's exact reviewed source SHA and merge-request author. This write is enabled by default; set `OCR_AUTO_APPROVE=false` when the bot must remain comment-only. Complete optional MR metadata and complete zero-record enrichment remain eligible, while degraded metadata, required context degradation, admitted mutable context, legacy receipts, any direct external MCP, author movement, or bot self-authorship prevents an approval write. GitLab approval rules and protected-branch policy remain authoritative. The toolkit only adds an eligible approval; it never removes an existing approval when a later review is ineligible or disabled.
+After every current review note publishes, the GitLab adapter can add a conservative approval bound to receipt v5's exact reviewed source SHA and merge-request author. This write is enabled by default; set `OCR_AUTO_APPROVE=false` when the bot must remain comment-only. DLP-clean metadata, generic discussions, and adapter records do not independently block approval, while degraded metadata, DLP rejection, required context degradation, admitted remediation history, legacy receipts, publication filtering, any direct external MCP, author movement, or bot self-authorship prevents an approval write. GitLab approval rules and protected-branch policy remain authoritative. The toolkit only adds an eligible approval; it never removes an existing approval when a later review is ineligible or disabled.
 
 Accepted tradeoffs can be recorded in `.opencodereview/accepted-decisions.md`; the evidence collector supplies only applicable target-ref decisions and never lets a source change self-authorize its review. Root and nested target `AGENTS.md`/`CLAUDE.md` guidance is similarly exposed through the existing evidence MCP with deterministic scope and precedence, while any guidance touched by the merge request is excluded. See [Accepted project decisions](docs/configuration.md#accepted-project-decisions) and [Target project guidance](docs/configuration.md#target-project-guidance) for formats and trust boundaries.
 
 ## Project architecture
 
-The shipped Repository Evidence Engine reads immutable base/head Git objects, stores bounded typed facts and deltas, creates the compact bootstrap used by OCR, and exposes detailed facts, scoped completeness, and base/head changes through the mandatory built-in read-only MCP server. Version 0.7.0 establishes protected-policy enriched reviews: they can acquire stable GitLab discussions and authorize external issue/document records through a provider-neutral stdio or HTTPS proxy before OCR. The same built-in MCP exposes only opaque committed `context_list`/`context_get` handles; it has no provider network or arbitrary identifier path. Direct external MCP remains a separate privileged, comment-only operator boundary.
+The shipped Repository Evidence Engine reads immutable base/head Git objects, stores bounded typed facts and deltas, creates the compact bootstrap used by OCR, and exposes detailed facts, scoped completeness, and base/head changes through the mandatory built-in read-only MCP server. Protected-policy enriched reviews can acquire stable GitLab discussions, verified remediation history, and authorized external issue/document records before OCR. Forge-specific acquisition and posting stay at provider edges; the broker, DLP, store, MCP, receipts, and tests use common contracts so a future GitHub adapter can reuse them without inheriting GitLab API semantics. The same built-in MCP exposes only opaque committed `context_list`/`context_get` handles; it has no provider network or arbitrary identifier path. Direct external MCP remains a separate privileged, comment-only operator boundary.
 
 - [Toolkit strategy](docs/engineering/toolkit_strategy.md) - durable product boundaries, architecture, invariants, and non-goals.
 - [Bounded review context](docs/review-context.md) - protected policy, adapter protocol, GitLab discussions, opaque handles, DLP, receipt, and cleanup contracts.
@@ -285,7 +285,7 @@ ocr-ci review --result /tmp/ocr-result.json --stderr /tmp/ocr-stderr.log -- ...
 ocr-ci post --result /tmp/ocr-result.json --stderr /tmp/ocr-stderr.log
 ```
 
-See the [`examples/gitlab/ocr-review.gitlab-ci.yml`](examples/gitlab/ocr-review.gitlab-ci.yml) example, the [GitLab setup guide](docs/gitlab.md), and [GitLab review operations](docs/operations.md).
+See the [GitLab mode matrix](examples/gitlab/README.md), the complete [`ocr-review.gitlab-ci.yml`](examples/gitlab/ocr-review.gitlab-ci.yml) pipeline, the [GitLab setup guide](docs/gitlab.md), and [GitLab review operations](docs/operations.md).
 
 ## Configuration and safety
 
@@ -295,6 +295,7 @@ Repository content, OCR output, and provider responses are untrusted inputs. The
 
 ## Development and release
 
+- [Documentation index](docs/README.md)
 - [Contributing](CONTRIBUTING.md)
 - [Code of Conduct](CODE_OF_CONDUCT.md)
 - [Development guide](docs/development.md)

{open_code_review_toolkit-0.7.0 → open_code_review_toolkit-0.8.0}/README.md

@@ -27,17 +27,17 @@ Stable distributions are published to [PyPI](https://pypi.org/project/open-code-
 
 ## How reviews evolve
 
-On a successful rerun, the toolkit replaces untouched OCR-only notes instead of accumulating stale reviews. A human reply transfers that discussion to the team: the conversation is preserved and a matching finding is suppressed. Reply with `/ocr suppress` to keep a discussion open without future repeats, or `/ocr resolve` to suppress it and resolve the discussion after the next successful posting transaction.
+On a successful rerun, the toolkit replaces untouched OCR-only notes instead of accumulating stale reviews. A human reply transfers that discussion to the team: the conversation is preserved and a matching finding is suppressed. Reply with `/ocr suppress` or `@<live-bot-username> suppress` to keep a discussion open without future repeats; use the corresponding `resolve` command to resolve it after the next successful posting transaction. For example, a bot named `mr.bot` accepts the exact reply `@mr.bot resolve`.
 
 Suppression uses both the GitLab diff position and a stable finding fingerprint, so ordinary line shifts do not normally bring the same bug back. A materially changed finding can still receive a new discussion. See [GitLab review operations](docs/operations.md) for the complete lifecycle, posting modes, permissions, failure behavior, and Mermaid state diagram.
 
-After every current review note publishes, the GitLab adapter can add a conservative approval bound to receipt v4's exact reviewed source SHA and merge-request author. This write is enabled by default; set `OCR_AUTO_APPROVE=false` when the bot must remain comment-only. Complete optional MR metadata and complete zero-record enrichment remain eligible, while degraded metadata, required context degradation, admitted mutable context, legacy receipts, any direct external MCP, author movement, or bot self-authorship prevents an approval write. GitLab approval rules and protected-branch policy remain authoritative. The toolkit only adds an eligible approval; it never removes an existing approval when a later review is ineligible or disabled.
+After every current review note publishes, the GitLab adapter can add a conservative approval bound to receipt v5's exact reviewed source SHA and merge-request author. This write is enabled by default; set `OCR_AUTO_APPROVE=false` when the bot must remain comment-only. DLP-clean metadata, generic discussions, and adapter records do not independently block approval, while degraded metadata, DLP rejection, required context degradation, admitted remediation history, legacy receipts, publication filtering, any direct external MCP, author movement, or bot self-authorship prevents an approval write. GitLab approval rules and protected-branch policy remain authoritative. The toolkit only adds an eligible approval; it never removes an existing approval when a later review is ineligible or disabled.
 
 Accepted tradeoffs can be recorded in `.opencodereview/accepted-decisions.md`; the evidence collector supplies only applicable target-ref decisions and never lets a source change self-authorize its review. Root and nested target `AGENTS.md`/`CLAUDE.md` guidance is similarly exposed through the existing evidence MCP with deterministic scope and precedence, while any guidance touched by the merge request is excluded. See [Accepted project decisions](docs/configuration.md#accepted-project-decisions) and [Target project guidance](docs/configuration.md#target-project-guidance) for formats and trust boundaries.
 
 ## Project architecture
 
-The shipped Repository Evidence Engine reads immutable base/head Git objects, stores bounded typed facts and deltas, creates the compact bootstrap used by OCR, and exposes detailed facts, scoped completeness, and base/head changes through the mandatory built-in read-only MCP server. Version 0.7.0 establishes protected-policy enriched reviews: they can acquire stable GitLab discussions and authorize external issue/document records through a provider-neutral stdio or HTTPS proxy before OCR. The same built-in MCP exposes only opaque committed `context_list`/`context_get` handles; it has no provider network or arbitrary identifier path. Direct external MCP remains a separate privileged, comment-only operator boundary.
+The shipped Repository Evidence Engine reads immutable base/head Git objects, stores bounded typed facts and deltas, creates the compact bootstrap used by OCR, and exposes detailed facts, scoped completeness, and base/head changes through the mandatory built-in read-only MCP server. Protected-policy enriched reviews can acquire stable GitLab discussions, verified remediation history, and authorized external issue/document records before OCR. Forge-specific acquisition and posting stay at provider edges; the broker, DLP, store, MCP, receipts, and tests use common contracts so a future GitHub adapter can reuse them without inheriting GitLab API semantics. The same built-in MCP exposes only opaque committed `context_list`/`context_get` handles; it has no provider network or arbitrary identifier path. Direct external MCP remains a separate privileged, comment-only operator boundary.
 
 - [Toolkit strategy](docs/engineering/toolkit_strategy.md) - durable product boundaries, architecture, invariants, and non-goals.
 - [Bounded review context](docs/review-context.md) - protected policy, adapter protocol, GitLab discussions, opaque handles, DLP, receipt, and cleanup contracts.
@@ -58,7 +58,7 @@ ocr-ci review --result /tmp/ocr-result.json --stderr /tmp/ocr-stderr.log -- ...
 ocr-ci post --result /tmp/ocr-result.json --stderr /tmp/ocr-stderr.log
 ```
 
-See the [`examples/gitlab/ocr-review.gitlab-ci.yml`](examples/gitlab/ocr-review.gitlab-ci.yml) example, the [GitLab setup guide](docs/gitlab.md), and [GitLab review operations](docs/operations.md).
+See the [GitLab mode matrix](examples/gitlab/README.md), the complete [`ocr-review.gitlab-ci.yml`](examples/gitlab/ocr-review.gitlab-ci.yml) pipeline, the [GitLab setup guide](docs/gitlab.md), and [GitLab review operations](docs/operations.md).
 
 ## Configuration and safety
 
@@ -68,6 +68,7 @@ Repository content, OCR output, and provider responses are untrusted inputs. The
 
 ## Development and release
 
+- [Documentation index](docs/README.md)
 - [Contributing](CONTRIBUTING.md)
 - [Code of Conduct](CODE_OF_CONDUCT.md)
 - [Development guide](docs/development.md)

{open_code_review_toolkit-0.7.0 → open_code_review_toolkit-0.8.0}/pyproject.toml

@@ -106,7 +106,7 @@ source = ["ocr_toolkit"]
 omit = ["src/ocr_toolkit/_version.py"]
 
 [tool.coverage.report]
-fail_under = 70
+fail_under = 85
 show_missing = true
 skip_covered = true
 exclude_also = [
@@ -133,6 +133,11 @@ directory = "bugfix"
 name = "🐛 Bug Fixes"
 showcontent = true
 
+[[tool.towncrier.type]]
+directory = "maintenance"
+name = "🛠 Maintenance"
+showcontent = true
+
 [[tool.towncrier.type]]
 directory = "refactor"
 name = "🔧 Refactoring"

{open_code_review_toolkit-0.7.0 → open_code_review_toolkit-0.8.0}/src/ocr_toolkit/_version.py

@@ -18,7 +18,7 @@ version_tuple: tuple[int | str, ...]
 commit_id: str | None
 __commit_id__: str | None
 
-__version__ = version = '0.7.0'
-__version_tuple__ = version_tuple = (0, 7, 0)
+__version__ = version = '0.8.0'
+__version_tuple__ = version_tuple = (0, 8, 0)
 
 __commit_id__ = commit_id = None

{open_code_review_toolkit-0.7.0 → open_code_review_toolkit-0.8.0}/src/ocr_toolkit/cli.py

@@ -32,6 +32,11 @@ def build_parser() -> argparse.ArgumentParser:
     )
     review_parser.add_argument("--result", required=True, help="OCR JSON output path.")
     review_parser.add_argument("--stderr", required=True, help="Full OCR stderr artifact path.")
+    review_parser.add_argument(
+        "--preserve-private-artifacts",
+        action="store_true",
+        help="Retain sensitive OCR session artifacts after a local diagnostic review.",
+    )
     review_parser.add_argument(
         "ocr_args", nargs=argparse.REMAINDER, help="OCR review arguments after --."
     )
@@ -65,7 +70,12 @@ def main(argv: Sequence[str] | None = None) -> int:
     if args.command == "review":
         ocr_args = args.ocr_args[1:] if args.ocr_args[:1] == ["--"] else args.ocr_args
         try:
-            return review_runner.run_evidence_review(Path(args.result), Path(args.stderr), ocr_args)
+            return review_runner.run_evidence_review(
+                Path(args.result),
+                Path(args.stderr),
+                ocr_args,
+                preserve_private_artifacts=args.preserve_private_artifacts,
+            )
         except review_runner.ReviewRunnerError as exc:
             print(f"Cannot run Open Code Review: {exc}", file=sys.stderr)
             return 2

{open_code_review_toolkit-0.7.0 → open_code_review_toolkit-0.8.0}/src/ocr_toolkit/configure.py

@@ -74,22 +74,16 @@ def _parse_extra_body(value: str) -> Any:
 
 
 def _llm_protocol() -> str:
-    """Resolve and validate the OCR 1.7.10 LLM wire protocol."""
-
-    protocol = _env("OCR_LLM_PROTOCOL")
-    legacy_mode = _env("OCR_USE_ANTHROPIC").lower()
-    if legacy_mode not in {"", "false", "true"}:
-        raise OCRRuntimeConfigError("OCR_USE_ANTHROPIC must be true or false when set")
-    if not protocol:
-        protocol = "anthropic" if legacy_mode == "true" else "openai"
+    """Resolve and validate the explicit OCR LLM wire protocol."""
+
+    if "OCR_USE_ANTHROPIC" in os.environ:
+        raise OCRRuntimeConfigError(
+            "OCR_USE_ANTHROPIC was removed; set OCR_LLM_PROTOCOL=anthropic explicitly"
+        )
+    protocol = _env("OCR_LLM_PROTOCOL", "openai") or "openai"
     if protocol not in LLM_PROTOCOLS:
         allowed = ", ".join(sorted(LLM_PROTOCOLS))
         raise OCRRuntimeConfigError(f"OCR_LLM_PROTOCOL must be one of: {allowed}")
-
-    if legacy_mode == "true" and protocol != "anthropic":
-        raise OCRRuntimeConfigError("OCR_LLM_PROTOCOL conflicts with OCR_USE_ANTHROPIC")
-    if legacy_mode == "false" and protocol == "anthropic":
-        raise OCRRuntimeConfigError("OCR_LLM_PROTOCOL conflicts with OCR_USE_ANTHROPIC")
     return protocol
 
 

{open_code_review_toolkit-0.7.0 → open_code_review_toolkit-0.8.0}/src/ocr_toolkit/context/adapters.py

@@ -19,7 +19,11 @@ from pathlib import Path
 from typing import Any
 
 from ocr_toolkit.common.redaction import redact_sensitive
-from ocr_toolkit.context.contracts import REQUEST_SCHEMA, RESOURCE_CLASSES, RESPONSE_SCHEMA
+from ocr_toolkit.context.contracts import (
+    REFERENCE_RESOURCE_CLASSES,
+    REQUEST_SCHEMA,
+    RESPONSE_SCHEMA,
+)
 
 MAX_ADAPTER_CONFIG_BYTES = 64 * 1024
 MAX_ADAPTERS = 16
@@ -197,7 +201,7 @@ def parse_adapter_config(raw: str | None) -> tuple[AdapterConfig, ...]:
         resource_classes = _names(
             item.get("resource_classes"),
             label="adapter resource classes",
-            allowed=RESOURCE_CLASSES,
+            allowed=REFERENCE_RESOURCE_CLASSES,
         )
         if kind == "stdio":
             if set(item).difference(common | {"command", "args", "env_from"}):
@@ -590,7 +594,7 @@ def authorize_and_resolve(
         or request.adapter != config.name
         or request.tenant not in config.tenants
         or request.resource_class not in config.resource_classes
-        or request.resource_class not in RESOURCE_CLASSES
+        or request.resource_class not in REFERENCE_RESOURCE_CLASSES
         or request.requested_fields != tuple(sorted(set(request.requested_fields)))
         or not 1 <= request.max_chars <= 100_000
         or not 1 <= request.max_bytes <= 400_000

{open_code_review_toolkit-0.7.0 → open_code_review_toolkit-0.8.0}/src/ocr_toolkit/context/broker.py

@@ -21,8 +21,10 @@ from ocr_toolkit.context.adapters import (
 from ocr_toolkit.context.contracts import (
     ACCOUNT_CLASSES,
     ContextPolicy,
+    ContextProjections,
     DiscussionPolicy,
     ReferencePolicy,
+    RemediationThreadPolicy,
     TextBudgets,
 )
 from ocr_toolkit.context.dlp import check_text, normalize_text
@@ -50,6 +52,15 @@ class BrokerResult:
     required_degraded: bool
 
 
+@dataclass(frozen=True, slots=True)
+class ContextOrigin:
+    """Name one provider composition edge without coupling the broker to its API."""
+
+    source: str
+    adapter: str
+    tenant: str
+
+
 class DiscussionView(Protocol):
     """Expose only normalized discussion fields needed for store projection."""
 
@@ -67,6 +78,31 @@ class DiscussionView(Protocol):
     digest: str
 
 
+class RemediationReplyView(Protocol):
+    """Expose one provider-normalized reply in a verified remediation thread."""
+
+    order: int
+    author_class: str
+    author_pseudonym: str
+    body: str
+    created_at: int
+    updated_at: int
+
+
+class RemediationThreadView(Protocol):
+    """Expose only the common verified remediation bundle contract."""
+
+    root_author_pseudonym: str
+    root_body: str
+    anchor_state: str
+    replies: Sequence[RemediationReplyView]
+    completeness: str
+    resolved_count: int
+    outdated_count: int
+    version: str
+    digest: str
+
+
 STATE_RE = re.compile(r"[a-z][a-z0-9_-]{0,63}\Z")
 PSEUDONYM_RE = re.compile(r"actor-[0-9a-f]{16}\Z")
 SHA256_RE = re.compile(r"[0-9a-f]{64}\Z")
@@ -79,19 +115,44 @@ def _project(
     return {field: record[field] for field in fields if field in record}
 
 
+def _publish_projection(
+    record: Mapping[str, object],
+    fields: Sequence[str],
+    *,
+    budgets: TextBudgets,
+    forbidden: tuple[str, ...],
+) -> dict[str, object] | None:
+    """Project publishable fields only after publication-specific text DLP."""
+
+    projected = _project(record, fields)
+    text = projected.get("text")
+    if isinstance(text, str):
+        checked = check_text(
+            text,
+            budgets=budgets,
+            publication=True,
+            forbidden=forbidden,
+        )
+        if not checked.admitted or checked.text != text:
+            return None
+    return projected
+
+
 def _normalized_record(
     record: Mapping[str, object],
     *,
-    policy: ReferencePolicy,
+    budgets: TextBudgets,
+    projections: ContextProjections,
+    resource_class: str,
     forbidden: tuple[str, ...],
 ) -> Mapping[str, object] | None:
     result: dict[str, object] = {}
-    for field in policy.projections.retrieve:
+    for field in projections.retrieve:
         if field not in record:
             continue
         value = record[field]
         if field == "text":
-            checked = check_text(value, budgets=policy.budgets, forbidden=forbidden)
+            checked = check_text(value, budgets=budgets, forbidden=forbidden)
             if not checked.admitted or checked.text is None:
                 return None
             result[field] = checked.text
@@ -106,7 +167,7 @@ def _normalized_record(
             )
             if not checked.admitted:
                 return None
-            if field == "descriptor" and normalized != policy.resource_class:
+            if field == "descriptor" and normalized != resource_class:
                 return None
             if field == "state" and STATE_RE.fullmatch(normalized) is None:
                 return None
@@ -160,7 +221,9 @@ def prepare_discussion_records(
     records: Sequence[DiscussionView],
     *,
     policy: DiscussionPolicy,
+    origin: ContextOrigin,
     expiry: int,
+    forbidden: tuple[str, ...] = (),
 ) -> tuple[PendingContextRecord, ...]:
     """Project provider-normalized discussions into the common private store contract."""
 
@@ -181,18 +244,41 @@ def prepare_discussion_records(
             "version": record.version,
             "expiry": expiry,
         }
-        retrieved = _project(value, policy.projections.retrieve)
+        retrieved = _normalized_record(
+            value,
+            budgets=policy.budgets,
+            projections=policy.projections,
+            resource_class="discussion",
+            forbidden=forbidden,
+        )
+        if (
+            retrieved is None
+            or not isinstance(record.digest, str)
+            or SHA256_RE.fullmatch(record.digest) is None
+            or normalize_text(record.version) != record.version
+            or not record.version
+            or len(record.version) > 512
+        ):
+            continue
+        publish = _publish_projection(
+            retrieved,
+            policy.projections.publish,
+            budgets=policy.budgets,
+            forbidden=forbidden,
+        )
+        if publish is None:
+            continue
         pending.append(
             PendingContextRecord(
-                source="forge:gitlab_discussions",
-                adapter="gitlab",
-                tenant="project",
+                source=origin.source,
+                adapter=origin.adapter,
+                tenant=origin.tenant,
                 canonical_object=hashlib.sha256(f"discussion:{record.digest}".encode()).hexdigest(),
                 resource_class="issue",
                 descriptor="discussion",
                 projections={
                     "model": _project(retrieved, policy.projections.model),
-                    "publish": _project(retrieved, policy.projections.publish),
+                    "publish": publish,
                     "retain": _project(retrieved, policy.projections.retain),
                 },
                 version=record.version,
@@ -204,6 +290,127 @@ def prepare_discussion_records(
     return tuple(pending)
 
 
+def prepare_remediation_records(
+    records: Sequence[RemediationThreadView],
+    *,
+    policy: RemediationThreadPolicy,
+    origin: ContextOrigin,
+    expiry: int,
+    forbidden: tuple[str, ...] = (),
+) -> tuple[PendingContextRecord, ...]:
+    """DLP-check normalized remediation views and build the fixed private projection."""
+
+    pending: list[PendingContextRecord] = []
+    for record in records:
+        root = check_text(record.root_body, budgets=policy.budgets, forbidden=forbidden)
+        if (
+            not root.admitted
+            or root.text != record.root_body
+            or not isinstance(record.root_author_pseudonym, str)
+            or PSEUDONYM_RE.fullmatch(record.root_author_pseudonym) is None
+            or record.anchor_state not in {"current", "outdated", "unpositioned"}
+            or record.completeness not in {"complete", "partial"}
+            or not isinstance(record.digest, str)
+            or SHA256_RE.fullmatch(record.digest) is None
+            or normalize_text(record.version) != record.version
+            or not record.version
+            or len(record.version) > 512
+            or not isinstance(record.resolved_count, int)
+            or isinstance(record.resolved_count, bool)
+            or not isinstance(record.outdated_count, int)
+            or isinstance(record.outdated_count, bool)
+            or record.resolved_count < 0
+            or record.outdated_count < 0
+        ):
+            continue
+        replies: list[dict[str, object]] = []
+        valid = True
+        for expected_order, reply in enumerate(record.replies):
+            checked = check_text(reply.body, budgets=policy.budgets, forbidden=forbidden)
+            if (
+                reply.order != expected_order
+                or reply.author_class not in policy.account_classes
+                or reply.author_class == "toolkit_bot"
+                or not isinstance(reply.author_pseudonym, str)
+                or PSEUDONYM_RE.fullmatch(reply.author_pseudonym) is None
+                or not checked.admitted
+                or checked.text != reply.body
+                or not isinstance(reply.created_at, int)
+                or isinstance(reply.created_at, bool)
+                or not isinstance(reply.updated_at, int)
+                or isinstance(reply.updated_at, bool)
+                or reply.created_at < 0
+                or reply.updated_at < reply.created_at
+                or reply.updated_at > expiry
+            ):
+                valid = False
+                break
+            replies.append(
+                {
+                    "order": reply.order,
+                    "author_class": reply.author_class,
+                    "author_pseudonym": reply.author_pseudonym,
+                    "text": checked.text,
+                    "created_at": reply.created_at,
+                    "updated_at": reply.updated_at,
+                }
+            )
+        if (
+            not valid
+            or not replies
+            or len(replies) > policy.max_replies_per_thread
+            or record.resolved_count > len(replies) + 1
+            or record.outdated_count > len(replies) + 1
+            or (record.anchor_state == "outdated" and record.outdated_count < 1)
+        ):
+            continue
+        remediation = {
+            "root": {
+                "text": root.text,
+                "author_pseudonym": record.root_author_pseudonym,
+            },
+            "anchor_state": record.anchor_state,
+            "replies": replies,
+            "completeness": record.completeness,
+            "counts": {
+                "replies": len(replies),
+                "resolved": record.resolved_count,
+                "outdated": record.outdated_count,
+            },
+        }
+        pending.append(
+            PendingContextRecord(
+                source=origin.source,
+                adapter=origin.adapter,
+                tenant=origin.tenant,
+                canonical_object=hashlib.sha256(
+                    f"remediation:{record.digest}".encode()
+                ).hexdigest(),
+                resource_class="remediation_thread",
+                descriptor="remediation_thread",
+                projections={
+                    "model": {
+                        "descriptor": "remediation_thread",
+                        "remediation_thread": remediation,
+                    },
+                    "publish": {"descriptor": "remediation_thread"},
+                    "retain": {
+                        "state": record.completeness,
+                        "count": len(replies),
+                        "digest": record.digest,
+                        "version": record.version,
+                        "expiry": expiry,
+                    },
+                },
+                version=record.version,
+                digest=record.digest,
+                mutable=True,
+                expiry=expiry,
+            )
+        )
+    return tuple(pending)
+
+
 def acquire_external_records(
     *,
     policy: ContextPolicy,
@@ -316,7 +523,13 @@ def acquire_external_records(
             degradation_counts["invalid"] += 1
             required_degraded = required_degraded or reference.required
             continue
-        normalized = _normalized_record(response.record, policy=reference, forbidden=forbidden)
+        normalized = _normalized_record(
+            response.record,
+            budgets=reference.budgets,
+            projections=reference.projections,
+            resource_class=reference.resource_class,
+            forbidden=forbidden,
+        )
         if normalized is None:
             completeness[source] = "unavailable"
             degradation_counts["invalid"] += 1
@@ -377,6 +590,17 @@ def acquire_external_records(
         total_chars += chars
         total_bytes += byte_count
         total_lines += lines
+        publish = _publish_projection(
+            normalized,
+            reference.projections.publish,
+            budgets=reference.budgets,
+            forbidden=forbidden,
+        )
+        if publish is None:
+            completeness[source] = "unavailable"
+            degradation_counts["invalid"] += 1
+            required_degraded = required_degraded or reference.required
+            continue
         admitted = PendingContextRecord(
             source=source,
             adapter=reference.adapter,
@@ -386,7 +610,7 @@ def acquire_external_records(
             descriptor=str(normalized.get("descriptor", reference.resource_class)),
             projections={
                 "model": _project(normalized, reference.projections.model),
-                "publish": _project(normalized, reference.projections.publish),
+                "publish": publish,
                 "retain": _project(normalized, reference.projections.retain),
             },
             version=response.version,

{open_code_review_toolkit-0.7.0 → open_code_review_toolkit-0.8.0}/src/ocr_toolkit/context/contracts.py

@@ -4,12 +4,16 @@ from __future__ import annotations
 
 from dataclasses import dataclass
 
-POLICY_SCHEMA = "ocr.review-context-policy/v1"
-STORE_SCHEMA = "ocr.context-store/v1"
+POLICY_SCHEMA_V1 = "ocr.review-context-policy/v1"
+POLICY_SCHEMA_V2 = "ocr.review-context-policy/v2"
+POLICY_SCHEMAS = frozenset({POLICY_SCHEMA_V1, POLICY_SCHEMA_V2})
+POLICY_SCHEMA = POLICY_SCHEMA_V2
+STORE_SCHEMA = "ocr.context-store/v2"
 REQUEST_SCHEMA = "ocr.context-adapter-request/v1"
 RESPONSE_SCHEMA = "ocr.context-adapter-response/v1"
 ACCOUNT_CLASSES = frozenset({"user", "automation", "system", "toolkit_bot"})
-RESOURCE_CLASSES = frozenset({"issue", "document"})
+REFERENCE_RESOURCE_CLASSES = frozenset({"issue", "document"})
+STORE_RESOURCE_CLASSES = frozenset({*REFERENCE_RESOURCE_CLASSES, "remediation_thread"})
 PROJECTION_FIELDS = frozenset(
     {
         "descriptor",
@@ -28,6 +32,8 @@ PROJECTION_FIELDS = frozenset(
         "expiry",
     }
 )
+REMEDIATION_MODEL_FIELD = "remediation_thread"
+STORE_PROJECTION_FIELDS = frozenset({*PROJECTION_FIELDS, REMEDIATION_MODEL_FIELD})
 RETENTION_FIELDS = frozenset({"state", "count", "digest", "version", "expiry"})
 
 
@@ -78,6 +84,21 @@ class DiscussionPolicy:
     projections: ContextProjections
 
 
+@dataclass(frozen=True, slots=True)
+class RemediationThreadPolicy:
+    """Select verified toolkit-owned GitLab remediation threads."""
+
+    required: bool
+    account_classes: tuple[str, ...]
+    include_resolved: bool
+    include_outdated: bool
+    max_age_seconds: int
+    max_threads: int
+    max_replies_per_thread: int
+    max_items: int
+    budgets: TextBudgets
+
+
 @dataclass(frozen=True, slots=True)
 class RecognizerPolicy:
     """Hold one fixed toolkit-authored candidate grammar."""
@@ -110,5 +131,6 @@ class ContextPolicy:
     schema_version: str
     budgets: AggregateBudgets
     forge_discussions: DiscussionPolicy | None
+    remediation_threads: RemediationThreadPolicy | None
     references: tuple[ReferencePolicy, ...]
     digest: str