agent-lifecycle-kit 2.36.0__tar.gz → 2.37.0__tar.gz

{agent_lifecycle_kit-2.36.0/src/agent_lifecycle_kit.egg-info → agent_lifecycle_kit-2.37.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: agent-lifecycle-kit
-Version: 2.36.0
+Version: 2.37.0
 Summary: Provider-neutral lifecycle control layer that helps coding agents plan, execute, validate, and prove software tasks to completion.
 Author: Agent Lifecycle Kit contributors
 License: Apache-2.0
@@ -33,7 +33,7 @@ Dynamic: license-file
 
 **Agent Lifecycle Kit (ALK)** coordinates coding-agent work through a verifiable finish. It keeps the requested outcome, reviewed plan, execution boundaries, evidence and acceptance decisions connected while the external agent changes the project.
 
-Release 2.36 secures imports. Release 2.35 documents the [external AI SDLC CLI contract](docs/guides/ai-sdlc-integration.md) and eight bounded consumer cases. Wheel/sdist include both guides and case JSON; the plugin ships the paired CLI contract. Native authority and release gates remain required.
+ALK secures imports and documents the [external AI SDLC CLI contract](docs/guides/ai-sdlc-integration.md) and eight bounded consumer cases. Wheel/sdist include both guides and case JSON; the plugin ships the paired CLI contract. Native authority and release gates remain required.
 
 The provider-neutral workflow works with Codex, Claude Code, Qwen Code, Goose, OpenInterpreter, Pi, Grok Build, or another CLI. Provider commands, model choice and secret handling stay in adapters or host-local profiles.
 
@@ -42,7 +42,7 @@ Release 2.28 adds [bounded lifecycle and accounting baselines](docs/reference/pe
 ## Quick start
 Start with [install ALK and make the first run](docs/guides/install-and-first-run.md). Release 2.23 adds [compact S0/S1 authoring](docs/reference/tier-proportional-plan-packages.md) with explicit checklist completion and unchanged approval gates. Release 2.22 adds [conservative plan-review reuse](docs/reference/conservative-delta-review.md) with independent current applicability and no automatic full reruns. Release 2.21 adds [command accounting, bounded preparation and response recovery](docs/reference/source-native-workflow-accounting.md). Release 2.20 adds [target-bound initialization](docs/reference/cli.md#release-220-composed-initialization-and-digest-display), [bounded local continuation](docs/reference/workflow-continuation.md) and batched Git snapshots. Release 2.19 adds [safe artifact authoring and planning preflight](docs/reference/safe-artifact-authoring.md), [bounded review handoff](docs/reference/review-mesh.md#bounded-review-handoff) and thin skills with complete references. Release 2.18 bounds process/Git diagnostics, hardens signing and distinguishes OpenCode 1.x/2.x compatibility ([adapter notes](docs/adapters/opencode.md)); 2.17 strengthens [workflow validation and recovery](docs/reference/workflow-completeness.md). Release 2.16 introduced cumulative accounting within verified host capabilities; see [retention maintenance](docs/operations/retention-maintenance.md).
 
-It includes macOS, Linux and Windows instructions, Python 3.11-3.15, [PyPI](https://pypi.org/project/agent-lifecycle-kit/) installation with `python -m pip install agent-lifecycle-kit==2.36.0`, and common `agent-lifecycle version` errors. See [2.15 migration notes](docs/reference/cli.md#migrating-authority-artifacts-to-215) and [authority boundaries](docs/reference/authority-artifact-boundaries.md).
+It includes macOS, Linux and Windows instructions, Python 3.11-3.15, [PyPI](https://pypi.org/project/agent-lifecycle-kit/) installation with `python -m pip install agent-lifecycle-kit==2.37.0`, and common `agent-lifecycle version` errors. See [2.15 migration notes](docs/reference/cli.md#migrating-authority-artifacts-to-215) and [authority boundaries](docs/reference/authority-artifact-boundaries.md).
 
 For a quick start, run:
 

{agent_lifecycle_kit-2.36.0 → agent_lifecycle_kit-2.37.0}/README.md

@@ -10,7 +10,7 @@
 
 **Agent Lifecycle Kit (ALK)** coordinates coding-agent work through a verifiable finish. It keeps the requested outcome, reviewed plan, execution boundaries, evidence and acceptance decisions connected while the external agent changes the project.
 
-Release 2.36 secures imports. Release 2.35 documents the [external AI SDLC CLI contract](docs/guides/ai-sdlc-integration.md) and eight bounded consumer cases. Wheel/sdist include both guides and case JSON; the plugin ships the paired CLI contract. Native authority and release gates remain required.
+ALK secures imports and documents the [external AI SDLC CLI contract](docs/guides/ai-sdlc-integration.md) and eight bounded consumer cases. Wheel/sdist include both guides and case JSON; the plugin ships the paired CLI contract. Native authority and release gates remain required.
 
 The provider-neutral workflow works with Codex, Claude Code, Qwen Code, Goose, OpenInterpreter, Pi, Grok Build, or another CLI. Provider commands, model choice and secret handling stay in adapters or host-local profiles.
 
@@ -19,7 +19,7 @@ Release 2.28 adds [bounded lifecycle and accounting baselines](docs/reference/pe
 ## Quick start
 Start with [install ALK and make the first run](docs/guides/install-and-first-run.md). Release 2.23 adds [compact S0/S1 authoring](docs/reference/tier-proportional-plan-packages.md) with explicit checklist completion and unchanged approval gates. Release 2.22 adds [conservative plan-review reuse](docs/reference/conservative-delta-review.md) with independent current applicability and no automatic full reruns. Release 2.21 adds [command accounting, bounded preparation and response recovery](docs/reference/source-native-workflow-accounting.md). Release 2.20 adds [target-bound initialization](docs/reference/cli.md#release-220-composed-initialization-and-digest-display), [bounded local continuation](docs/reference/workflow-continuation.md) and batched Git snapshots. Release 2.19 adds [safe artifact authoring and planning preflight](docs/reference/safe-artifact-authoring.md), [bounded review handoff](docs/reference/review-mesh.md#bounded-review-handoff) and thin skills with complete references. Release 2.18 bounds process/Git diagnostics, hardens signing and distinguishes OpenCode 1.x/2.x compatibility ([adapter notes](docs/adapters/opencode.md)); 2.17 strengthens [workflow validation and recovery](docs/reference/workflow-completeness.md). Release 2.16 introduced cumulative accounting within verified host capabilities; see [retention maintenance](docs/operations/retention-maintenance.md).
 
-It includes macOS, Linux and Windows instructions, Python 3.11-3.15, [PyPI](https://pypi.org/project/agent-lifecycle-kit/) installation with `python -m pip install agent-lifecycle-kit==2.36.0`, and common `agent-lifecycle version` errors. See [2.15 migration notes](docs/reference/cli.md#migrating-authority-artifacts-to-215) and [authority boundaries](docs/reference/authority-artifact-boundaries.md).
+It includes macOS, Linux and Windows instructions, Python 3.11-3.15, [PyPI](https://pypi.org/project/agent-lifecycle-kit/) installation with `python -m pip install agent-lifecycle-kit==2.37.0`, and common `agent-lifecycle version` errors. See [2.15 migration notes](docs/reference/cli.md#migrating-authority-artifacts-to-215) and [authority boundaries](docs/reference/authority-artifact-boundaries.md).
 
 For a quick start, run:
 

{agent_lifecycle_kit-2.36.0 → agent_lifecycle_kit-2.37.0}/docs/guides/ai-sdlc-integration.md

@@ -7,7 +7,7 @@ required. ALK owns workflow authority. The client owns process execution,
 presentation and reconciliation. The host owns model execution and genuine usage
 attestation; an independent reviewer owns review conclusions.
 
-This guide defines a provider-neutral CLI consumer contract for release 2.35.0.
+This guide defines a provider-neutral CLI consumer contract.
 It adds no server, daemon, provider broker, runtime API or second state machine.
 The supported surface below is deliberately bounded. Other operations retain
 their [CLI contracts](../reference/cli.md).
@@ -16,7 +16,7 @@ their [CLI contracts](../reference/cli.md).
 
 The client MUST run `agent-lifecycle version` and require
 `agent-lifecycle-version.v1`. It MUST compare `version` with its explicitly tested
-version mapping. The 2.35.0 contract does not qualify an earlier or later release
+version mapping. A tested contract does not qualify another release
 merely because its version looks similar. The client MUST refuse mutations for
 an unknown version, unknown `schemaVersion`, unsupported action or incompatible
 required field. Read-only diagnostics can remain available, clearly labelled as
@@ -146,6 +146,34 @@ by writing a success event or replaying a host command. Native operation guards
 do not guarantee exactly-once execution of arbitrary host/model/external side
 effects; those require their own host-owned reconciliation evidence.
 
+## Reference caller resource boundaries
+
+The stdlib-only reference in `tests/cli/test_external_consumer_contract.py`
+drains two unbuffered pipes concurrently. Each stream reads at most 1 MiB plus
+one overflow-detection byte; it creates no output capture files. The runtime
+deadline is 20 seconds after process creation returns. A separate cleanup window
+is at most two seconds, covering direct-child stop/reaping and reader joins.
+OS process creation itself is not interruptible by this deadline.
+
+Timeout, overflow, invalid UTF-8/JSON, unknown output or unconfirmed cleanup
+records `UNKNOWN`, clears qualification and blocks all further calls. A unique
+bounded state/event match can resolve effect uncertainty; renegotiation is still
+required before mutation. It cannot clear unconfirmed resource cleanup: dispose
+that caller. Never retry a mutation to discover whether it succeeded.
+
+Executable hashing streams at most 32 MiB plus one detection byte before launch.
+This is an observed file identity, not attestation of every loaded runtime file.
+Reconciliation admits only regular non-symlink relative artifacts inside its
+working directory: state at most 1 MiB, event log at most 4 MiB, at most 1024 lines,
+and each line at most 65536 bytes including its newline. State is compared again
+after reading events; missing, changed, ambiguous, oversized or malformed input
+stays `UNKNOWN`. These checks are not an atomic filesystem snapshot or a sandbox
+for arbitrary executables, descendants or hostile concurrent filesystem changes.
+
+The eight disposable public cases also exercise a legacy v3 state fixture;
+they do not migrate native authority. Source/editable checks are intermediate
+proof. A final installed wheel needs its own exact-version qualification.
+
 ## Keep usage and evidence honest
 
 Retain exact source, executable/configuration/input digests, command/exit/time,

{agent_lifecycle_kit-2.36.0 → agent_lifecycle_kit-2.37.0}/docs/ru/guides/ai-sdlc-integration.md

@@ -7,7 +7,7 @@ ALK владеет workflow authority. Клиент отвечает за зап
 и сверку результата. Host отвечает за выполнение модели и подлинную usage
 attestation; независимый reviewer — за выводы review.
 
-Руководство задаёт provider-neutral контракт CLI-потребителя для release 2.35.0.
+Руководство задаёт provider-neutral контракт CLI-потребителя.
 Оно не добавляет server, daemon, provider broker, runtime API или вторую state
 machine. Описанная поверхность намеренно ограничена. Остальные операции сохраняют
 свои [CLI-контракты](../reference/cli.md).
@@ -16,7 +16,7 @@ machine. Описанная поверхность намеренно огран
 
 Клиент MUST выполнить `agent-lifecycle version` и потребовать
 `agent-lifecycle-version.v1`. Он MUST сравнить `version` с явно протестированной
-таблицей версий. Контракт 2.35.0 не квалифицирует более ранний или поздний release
+таблицей версий. Проверенный контракт не квалифицирует более ранний или поздний release
 только по сходству номера. Клиент MUST отказать в mutation при неизвестной версии,
 неизвестной `schemaVersion`, неподдерживаемом action или несовместимом обязательном
 поле. Read-only диагностика может оставаться доступной с явной пометкой об
@@ -152,6 +152,38 @@ revision не доказывает успех.
 exactly-once исполнение произвольных host/model/external side effects; для них
 нужны собственные host-owned evidence сверки.
 
+## Ограничения эталонного клиента
+
+Эталонный клиент `tests/cli/test_external_consumer_contract.py` использует только
+стандартную библиотеку. Он одновременно читает два небуферизованных канала:
+не более 1 МиБ и одного байта для обнаружения переполнения из каждого.
+Файлы захвата вывода не создаются. Срок исполнения — 20 секунд после возврата
+операции создания процесса. Отдельное окно очистки до двух секунд включает
+остановку и ожидание прямого дочернего процесса, а также завершение читателей.
+Само создание процесса средствами ОС этим сроком прервать нельзя.
+
+Таймаут, переполнение, неверный UTF-8/JSON, неизвестный ответ или неподтверждённая
+очистка записывают `UNKNOWN`, снимают квалификацию и запрещают следующие вызовы.
+Единственное согласованное событие вместе с state может разрешить неопределённость
+эффекта; перед изменением всё равно нужно повторное согласование версии.
+Совпадение эффекта не снимает запрет при неподтверждённой очистке ресурсов:
+прекратите использование такого клиента. Не повторяйте изменение, чтобы узнать
+результат первого вызова.
+
+До запуска отпечаток исполняемого файла считается потоково, не более 32 МиБ
+и одного байта обнаружения. Это наблюдаемая идентичность файла, а не подтверждение
+всех загруженных файлов среды. Сверка принимает только обычные файлы без
+символьных ссылок с относительными путями внутри рабочего каталога: state до
+1 МиБ, журнал до 4 МиБ, до 1024 строк и до 65536 байт на строку с переводом строки.
+После журнала state читается и сравнивается повторно. Недоступные, изменённые,
+неоднозначные, чрезмерные и повреждённые данные оставляют `UNKNOWN`. Это не
+атомарный снимок файловой системы и не изолированная среда для произвольных
+программ, потомков или враждебных одновременных изменений файлов.
+
+Восемь публичных сценариев на временных данных также используют старую fixture
+state v3; они не мигрируют нативные полномочия. Проверка исходного/editable-кода
+промежуточная: итоговый установленный wheel требует проверки своей точной версии.
+
 ## Сохранять подлинные usage и evidence
 
 Сохраняйте exact source, digests executable/configuration/inputs, command/exit/time,

{agent_lifecycle_kit-2.36.0 → agent_lifecycle_kit-2.37.0}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "agent-lifecycle-kit"
-version = "2.36.0"
+version = "2.37.0"
 description = "Provider-neutral lifecycle control layer that helps coding agents plan, execute, validate, and prove software tasks to completion."
 readme = "README.md"
 requires-python = ">=3.11,<3.16"

{agent_lifecycle_kit-2.36.0 → agent_lifecycle_kit-2.37.0}/src/agent_lifecycle/_version.py

@@ -1,3 +1,3 @@
 """Package version."""
 
-__version__ = "2.36.0"
+__version__ = "2.37.0"

{agent_lifecycle_kit-2.36.0 → agent_lifecycle_kit-2.37.0}/src/agent_lifecycle/imports/external_dialects.py

@@ -20,15 +20,15 @@ from agent_lifecycle.imports.dialect_profiles import (
 from agent_lifecycle.imports.planning import (
     DEFAULT_MAX_INPUT_BYTES,
     DEFAULT_TARGET_TOKENS,
+    MAX_REQUIREMENT_CHARS,
     PLANNING_IMPORT_RESULT_SCHEMA,
+    import_content_blockers,
     require_no_import_secrets,
     validate_import_result,
 )
 from agent_lifecycle.planning import validate_plan_manifest
 
 EXTERNAL_IMPORT_VALIDATION_SCHEMA = "agent-external-dialect-import-validation.v1"
-MAX_REQUIREMENTS = 8
-MAX_REQUIREMENT_CHARS = 180
 
 _POSIX_USER_PREFIX = "/" + "Users/"
 _POSIX_VOLUME_PREFIX = "/" + "Volumes/"
@@ -254,19 +254,23 @@ def _parse_scalar(value: str) -> Any:
 
 def _map_family(family: str, parsed: dict[str, Any], text: str, blockers: list[dict[str, Any]]) -> dict[str, Any]:
     if family == "workflow":
-        return _map_workflow(parsed, text)
-    if family == "agent":
-        return _map_agent(parsed, text)
-    blockers.append({"code": "external-dialect-family-invalid", "family": family})
-    return {}
+        mapping = _map_workflow(parsed, text)
+    elif family == "agent":
+        mapping = _map_agent(parsed, text)
+    else:
+        blockers.append({"code": "external-dialect-family-invalid", "family": family})
+        return {}
+    for field in ("requirements", "validationHints"):
+        blockers.extend(import_content_blockers(mapping[field], prefix="external-dialect", field=field))
+    return mapping
 
 
 def _map_workflow(parsed: dict[str, Any], text: str) -> dict[str, Any]:
     steps = _list_from(parsed, "steps", "jobs", "tasks")
     validations = _list_from(parsed, "validation", "validations", "checks")
     requirements = [
-        f"Review imported workflow step {index + 1}: {_safe_summary(step)}."
-        for index, step in enumerate(steps[:MAX_REQUIREMENTS])
+        f"Review imported workflow step {index + 1}: {_safe_summary(step, limit=None)}."
+        for index, step in enumerate(steps)
     ]
     if not requirements:
         requirements = _requirements_from_lines(text, prefix="Review imported workflow instruction")
@@ -274,8 +278,8 @@ def _map_workflow(parsed: dict[str, Any], text: str) -> dict[str, Any]:
         "family": "workflow",
         "title": _title(parsed, "Imported workflow draft"),
         "requirements": requirements,
-        "workHints": [{"kind": "workflow-step", "summary": _safe_summary(step)} for step in steps[:MAX_REQUIREMENTS]],
-        "validationHints": [_safe_summary(item) for item in validations[:MAX_REQUIREMENTS]],
+        "workHints": [{"kind": "workflow-step", "summary": _safe_summary(step, limit=None)} for step in steps],
+        "validationHints": [_safe_summary(item, limit=None) for item in validations],
         "hostLocalHints": {},
         "redactions": _redaction_summary(parsed),
     }
@@ -287,18 +291,19 @@ def _map_agent(parsed: dict[str, Any], text: str) -> dict[str, Any]:
     role = _first_string(parsed, "role", "description", "purpose")
     if role:
         requirements.append(
-            f"Review imported agent role hint before converting it into ALK task scope: {_safe_summary(role)}."
+            "Review imported agent role hint before converting it into ALK task scope: "
+            f"{_safe_summary(role, limit=None)}."
         )
     requirements.extend(
-        f"Review imported agent policy hint {index + 1}: {_safe_summary(item)}."
-        for index, item in enumerate(policies[:MAX_REQUIREMENTS])
+        f"Review imported agent policy hint {index + 1}: {_safe_summary(item, limit=None)}."
+        for index, item in enumerate(policies)
     )
     if not requirements:
         requirements = _requirements_from_lines(text, prefix="Review imported agent instruction")
     return {
         "family": "agent",
         "title": _title(parsed, "Imported agent draft"),
-        "requirements": requirements[:MAX_REQUIREMENTS],
+        "requirements": requirements,
         "workHints": [],
         "validationHints": [],
         "hostLocalHints": _host_local_hints(parsed),
@@ -315,8 +320,8 @@ def _candidate_plan(
     profile: dict[str, Any],
 ) -> dict[str, Any] | None:
     requirements = [
-        {"id": f"R-EXT-{index + 1}", "description": _truncate(item, MAX_REQUIREMENT_CHARS)}
-        for index, item in enumerate(mapping.get("requirements", [])[:MAX_REQUIREMENTS])
+        {"id": f"R-EXT-{index + 1}", "description": item}
+        for index, item in enumerate(mapping.get("requirements", []))
         if isinstance(item, str) and item.strip()
     ]
     if not requirements:
@@ -462,11 +467,11 @@ def _requirements_from_lines(text: str, *, prefix: str) -> list[str]:
     for line in text.splitlines():
         stripped = line.strip()
         if stripped.startswith(("- ", "* ")):
-            requirements.append(f"{prefix}: {_safe_summary(stripped[2:])}.")
+            requirements.append(f"{prefix}: {_safe_summary(stripped[2:], limit=None)}.")
         elif re.match(r"^\d+\.\s+", stripped):
             numbered_text = re.sub(r"^\d+\.\s+", "", stripped)
-            requirements.append(f"{prefix}: {_safe_summary(numbered_text)}.")
-    return requirements[:MAX_REQUIREMENTS]
+            requirements.append(f"{prefix}: {_safe_summary(numbered_text, limit=None)}.")
+    return requirements
 
 
 def _list_from(parsed: dict[str, Any], *keys: str) -> list[Any]:
@@ -494,15 +499,17 @@ def _title(parsed: dict[str, Any], fallback: str) -> str:
     return _safe_summary(value) if value else fallback
 
 
-def _safe_summary(value: Any) -> str:
+def _safe_summary(value: Any, *, limit: int | None = MAX_REQUIREMENT_CHARS) -> str:
     if isinstance(value, dict):
-        text = " ".join(f"{_clean_key(str(key))} {_safe_summary(item)}" for key, item in sorted(value.items()))
+        text = " ".join(
+            f"{_clean_key(str(key))} {_safe_summary(item, limit=limit)}" for key, item in sorted(value.items())
+        )
     else:
         text = str(value)
     text = _LOCAL_PATH_RE.sub("[redacted-path]", text)
     text = _SECRET_ASSIGNMENT_RE.sub(r"\1=[redacted]", text)
     text = re.sub(r"\s+", " ", text).strip()
-    return _truncate(text, MAX_REQUIREMENT_CHARS)
+    return text if limit is None else _truncate(text, limit)
 
 
 def _safe_identifier(value: str) -> str:

{agent_lifecycle_kit-2.36.0 → agent_lifecycle_kit-2.37.0}/src/agent_lifecycle/imports/planning.py

@@ -23,6 +23,12 @@ DEFAULT_TARGET_TOKENS = 4096
 MAX_REQUIREMENTS = 8
 MAX_REQUIREMENT_CHARS = 180
 
+_IMPORT_OVERFLOW_CAPS = {
+    f"{prefix}-{kind}-cap-exceeded": cap
+    for prefix in ("planning-import", "external-dialect")
+    for kind, cap in (("item", MAX_REQUIREMENTS), ("text", MAX_REQUIREMENT_CHARS))
+}
+
 LOCAL_PATH_PATTERNS = (
     re.compile(r"/" + r"Users/[^\s`'\"<>]+"),
     re.compile(r"/" + r"Volumes/[^\s`'\"<>]+"),
@@ -122,6 +128,7 @@ def _build_import_result(
             source_label=source_label,
             native_dialect_profile_digest=native_dialect_profile_digest,
             dialect_profile=dialect_profile,
+            blockers=blockers,
         )
         if text and not blockers
         else None
@@ -192,22 +199,21 @@ def validate_import_result(result: dict[str, Any]) -> dict[str, Any]:
     """Verify that an import result cannot bypass draft, audit or freeze gates."""
 
     blockers: list[dict[str, Any]] = []
-    if result.get("schemaVersion") != PLANNING_IMPORT_RESULT_SCHEMA:
-        blockers.append({"code": "planning-import-schema-invalid"})
-    if result.get("enabledByDefault") is not False:
-        blockers.append({"code": "planning-import-default-enabled"})
-    if result.get("activationMode") != "explicit-command":
-        blockers.append({"code": "planning-import-activation-mode"})
-    if result.get("sourceTrusted") is not False:
-        blockers.append({"code": "planning-import-source-trust"})
-    if result.get("requiresReview") is not True:
-        blockers.append({"code": "planning-import-review-not-required"})
-    if result.get("auditRequired") is not True:
-        blockers.append({"code": "planning-import-audit-not-required"})
-    if result.get("freezeBlocked") is not True:
-        blockers.append({"code": "planning-import-freeze-not-blocked"})
+    gates = (
+        (result.get("schemaVersion") != PLANNING_IMPORT_RESULT_SCHEMA, "planning-import-schema-invalid"),
+        (result.get("enabledByDefault") is not False, "planning-import-default-enabled"),
+        (result.get("activationMode") != "explicit-command", "planning-import-activation-mode"),
+        (result.get("sourceTrusted") is not False, "planning-import-source-trust"),
+        (result.get("requiresReview") is not True, "planning-import-review-not-required"),
+        (result.get("auditRequired") is not True, "planning-import-audit-not-required"),
+        (result.get("freezeBlocked") is not True, "planning-import-freeze-not-blocked"),
+    )
+    for failed, code in gates:
+        if failed:
+            blockers.append({"code": code})
     if result.get("status") != "PASS":
         blockers.append({"code": "planning-import-result-not-pass"})
+        blockers.extend(_import_overflow_blockers(result.get("blockers")))
     candidate = result.get("candidatePlan")
     candidate_digest = None
     _validate_markdown_collection(result.get("markdownCollection"), blockers)
@@ -313,12 +319,17 @@ def _candidate_plan(
     source_label: str,
     native_dialect_profile_digest: str | None,
     dialect_profile: dict[str, Any] | None,
+    blockers: list[dict[str, Any]],
 ) -> dict[str, Any] | None:
     parsed = _parse_structured_input(data)
     title = _extract_title(text, parsed)
-    requirements = _extract_requirements(text, parsed)
-    if not requirements:
+    descriptions = _extract_requirements(text, parsed)
+    blockers.extend(import_content_blockers(descriptions, prefix="planning-import", field="requirements"))
+    if not descriptions or blockers:
         return None
+    requirements = [
+        {"id": f"R-IMPORT-{index + 1}", "description": description} for index, description in enumerate(descriptions)
+    ]
     source_digest = sha256_hex(data)
     criteria = [
         {
@@ -413,7 +424,49 @@ def _extract_title(text: str, parsed: dict[str, Any] | None) -> str:
     return "Imported draft plan"
 
 
-def _extract_requirements(text: str, parsed: dict[str, Any] | None) -> list[dict[str, str]]:
+def _import_overflow_blockers(value: Any) -> list[dict[str, Any]]:
+    if not isinstance(value, list):
+        return []
+    return [safe for item in value[:2] if (safe := _safe_import_overflow(item)) is not None]
+
+
+def _safe_import_overflow(value: Any) -> dict[str, Any] | None:
+    if not isinstance(value, dict):
+        return None
+    code, field, cap = value.get("code"), value.get("field"), value.get("cap")
+    if not isinstance(code, str) or not isinstance(field, str) or type(cap) is not int:
+        return None
+    if code not in _IMPORT_OVERFLOW_CAPS or field not in {"requirements", "validationHints"}:
+        return None
+    if cap != _IMPORT_OVERFLOW_CAPS[code]:
+        return None
+    safe: dict[str, Any] = {"code": code, "field": field}
+    for key in ("count", "cap", "chars", "index"):
+        number = value.get(key)
+        if type(number) is int and 0 <= number <= 2**63 - 1:
+            safe[key] = number
+    return safe
+
+
+def import_content_blockers(values: list[str], *, prefix: str, field: str) -> list[dict[str, Any]]:
+    """Refuse discarded content with bounded metadata instead of a partial draft."""
+    if len(values) > MAX_REQUIREMENTS:
+        return [{"code": f"{prefix}-item-cap-exceeded", "field": field, "count": len(values), "cap": MAX_REQUIREMENTS}]
+    for index, value in enumerate(values):
+        if len(value) > MAX_REQUIREMENT_CHARS:
+            return [
+                {
+                    "code": f"{prefix}-text-cap-exceeded",
+                    "field": field,
+                    "index": index + 1,
+                    "chars": len(value),
+                    "cap": MAX_REQUIREMENT_CHARS,
+                }
+            ]
+    return []
+
+
+def _extract_requirements(text: str, parsed: dict[str, Any] | None) -> list[str]:
     extracted: list[str] = []
     if parsed is not None:
         raw_requirements = parsed.get("requirements")
@@ -440,10 +493,7 @@ def _extract_requirements(text: str, parsed: dict[str, Any] | None) -> list[dict
         )
         if fallback:
             extracted.append(fallback)
-    cleaned = [_clean_line(item, limit=MAX_REQUIREMENT_CHARS) for item in extracted if item.strip()]
-    return [
-        {"id": f"R-IMPORT-{index + 1}", "description": item} for index, item in enumerate(cleaned[:MAX_REQUIREMENTS])
-    ]
+    return [_clean_line(item, limit=None) for item in extracted if item.strip()]
 
 
 def _decode_text(data: bytes, blockers: list[dict[str, Any]]) -> str:
@@ -536,11 +586,11 @@ def _safe_identifier(value: str) -> str:
     return re.sub(r"[^a-z0-9_.-]+", "-", lowered).strip("-")[:80]
 
 
-def _clean_line(value: str, *, limit: int) -> str:
+def _clean_line(value: str, *, limit: int | None) -> str:
     cleaned = re.sub(r"\s+", " ", value).strip()
     for pattern in LOCAL_PATH_PATTERNS:
         cleaned = pattern.sub("[redacted-path]", cleaned)
-    return cleaned[:limit]
+    return cleaned if limit is None else cleaned[:limit]
 
 
 def _check_positive_cap(value: int, field: str) -> None:

{agent_lifecycle_kit-2.36.0 → agent_lifecycle_kit-2.37.0/src/agent_lifecycle_kit.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: agent-lifecycle-kit
-Version: 2.36.0
+Version: 2.37.0
 Summary: Provider-neutral lifecycle control layer that helps coding agents plan, execute, validate, and prove software tasks to completion.
 Author: Agent Lifecycle Kit contributors
 License: Apache-2.0
@@ -33,7 +33,7 @@ Dynamic: license-file
 
 **Agent Lifecycle Kit (ALK)** coordinates coding-agent work through a verifiable finish. It keeps the requested outcome, reviewed plan, execution boundaries, evidence and acceptance decisions connected while the external agent changes the project.
 
-Release 2.36 secures imports. Release 2.35 documents the [external AI SDLC CLI contract](docs/guides/ai-sdlc-integration.md) and eight bounded consumer cases. Wheel/sdist include both guides and case JSON; the plugin ships the paired CLI contract. Native authority and release gates remain required.
+ALK secures imports and documents the [external AI SDLC CLI contract](docs/guides/ai-sdlc-integration.md) and eight bounded consumer cases. Wheel/sdist include both guides and case JSON; the plugin ships the paired CLI contract. Native authority and release gates remain required.
 
 The provider-neutral workflow works with Codex, Claude Code, Qwen Code, Goose, OpenInterpreter, Pi, Grok Build, or another CLI. Provider commands, model choice and secret handling stay in adapters or host-local profiles.
 
@@ -42,7 +42,7 @@ Release 2.28 adds [bounded lifecycle and accounting baselines](docs/reference/pe
 ## Quick start
 Start with [install ALK and make the first run](docs/guides/install-and-first-run.md). Release 2.23 adds [compact S0/S1 authoring](docs/reference/tier-proportional-plan-packages.md) with explicit checklist completion and unchanged approval gates. Release 2.22 adds [conservative plan-review reuse](docs/reference/conservative-delta-review.md) with independent current applicability and no automatic full reruns. Release 2.21 adds [command accounting, bounded preparation and response recovery](docs/reference/source-native-workflow-accounting.md). Release 2.20 adds [target-bound initialization](docs/reference/cli.md#release-220-composed-initialization-and-digest-display), [bounded local continuation](docs/reference/workflow-continuation.md) and batched Git snapshots. Release 2.19 adds [safe artifact authoring and planning preflight](docs/reference/safe-artifact-authoring.md), [bounded review handoff](docs/reference/review-mesh.md#bounded-review-handoff) and thin skills with complete references. Release 2.18 bounds process/Git diagnostics, hardens signing and distinguishes OpenCode 1.x/2.x compatibility ([adapter notes](docs/adapters/opencode.md)); 2.17 strengthens [workflow validation and recovery](docs/reference/workflow-completeness.md). Release 2.16 introduced cumulative accounting within verified host capabilities; see [retention maintenance](docs/operations/retention-maintenance.md).
 
-It includes macOS, Linux and Windows instructions, Python 3.11-3.15, [PyPI](https://pypi.org/project/agent-lifecycle-kit/) installation with `python -m pip install agent-lifecycle-kit==2.36.0`, and common `agent-lifecycle version` errors. See [2.15 migration notes](docs/reference/cli.md#migrating-authority-artifacts-to-215) and [authority boundaries](docs/reference/authority-artifact-boundaries.md).
+It includes macOS, Linux and Windows instructions, Python 3.11-3.15, [PyPI](https://pypi.org/project/agent-lifecycle-kit/) installation with `python -m pip install agent-lifecycle-kit==2.37.0`, and common `agent-lifecycle version` errors. See [2.15 migration notes](docs/reference/cli.md#migrating-authority-artifacts-to-215) and [authority boundaries](docs/reference/authority-artifact-boundaries.md).
 
 For a quick start, run:
 

{agent_lifecycle_kit-2.36.0 → agent_lifecycle_kit-2.37.0}/tests/cli/fixtures/external-consumer-cases.json

@@ -4,7 +4,8 @@
   "testedVersions": [
     "2.34.0",
     "2.35.0",
-    "2.36.0"
+    "2.36.0",
+    "2.37.0"
   ],
   "versionSchema": "agent-lifecycle-version.v1",
   "continuationSchema": "agent-workflow-continuation-receipt.v1",