decomposer: record implementation detail for 7 step(s)

.agents/plans/wf-decompose-b4be4be4fd0c/steps/step-0.md

@@ -1,7 +1,7 @@
 # Step 0 — Implementation Detail
 
-1. Defined required normalized document fields, stable IDs, UTF-8 body semantics, MIME handling, and delete tombstones.
+1. Define immutable ACL, Provenance, Document, ConnectorConfig, SyncCursor, and Change models.
-2. Defined provenance fields for connector, immutable source identity, revisions, retrieval time, and webhook event ID.
+2. Define injected HTTP, SecretStore, Publisher, and Connector protocols.
-3. Defined tenant isolation and deny-by-default ACL propagation rules.
+3. Document full/incremental/webhook method signatures and required contract fields.
-4. Defined connector methods for full sync, incremental sync, and webhook change parsing.
+4. Document tenant isolation, secret references, cursor advancement, trigger modes, and deletion semantics.
-5. Defined opaque cursor shape, atomic advancement rule, trigger modes, retries, and idempotency requirements.
+5. Add contract-oriented package exports and dependency metadata for Python 3.11.

.agents/plans/wf-decompose-b4be4be4fd0c/steps/step-1.md

@@ -1,5 +1,9 @@
 # Step 1 — Implementation Detail
 
-1. Added generation_results/step-1.json with explicit Step 1 intent, inputs/outputs, artifact mapping, platform-compliance assertions, and verification commands.
+1. Validate owner and repository scope before source access.
-2. Added validation/step-1-evidence.md with inspectable mappings for scope, authentication, synchronization, webhooks, provenance, ACLs, and regression fixtures.
+2. Resolve the configured token through SecretStore using secret_ref.
-3. Recorded focused verification evidence for pytest and Python compilation without changing unrelated steps.
+3. List the configured Git tree recursively and filter by paths and allowed extensions.
+4. Fetch selected files and normalize title, content, MIME type, hash, provenance, and ACL metadata.
+5. Implement incremental changed-file retrieval and cursor advancement.
+6. Map GitHub webhook payloads into revisioned Change records with event IDs.
+7. Add full-ingestion and scope-validation tests.

.agents/plans/wf-decompose-b4be4be4fd0c/steps/step-2.md

@@ -1,7 +1,9 @@
 # Step 2 — Implementation Detail
 
-1. Implemented site_id, drive_id, folder_path, and credential_ref scope/configuration.
+1. Validate site and library scope before retrieval.
-2. Implemented authenticated Graph adapter calls through injected graph and secret-provider ports.
+2. Resolve the Graph token from the configured secret reference.
-3. Implemented full listing and configured PDF, DOCX, and HTML MIME filtering.
+3. List drive children and filter by folder and supported file extensions.
-4. Implemented delta-token incremental ingestion and cursor version advancement.
+4. Retrieve file content and map MIME type, eTag, web URL, modification time, and tenant ACL.
-5. Mapped SharePoint item IDs, URLs, modified timestamps, eTags, and webhook resources into contract records.
+5. Implement Graph delta-token incremental retrieval and deleted-item detection.
+6. Map subscription notifications into Change records with subscription and sequence metadata.
+7. Add SharePoint PDF normalization coverage.

.agents/plans/wf-decompose-b4be4be4fd0c/steps/step-3.md

@@ -1,7 +1,9 @@
 # Step 3 — Implementation Detail
 
-1. Added connector registry dispatch by connector name and tenant-aware cursor loading.
+1. Define RunResult with run ID, status, count, errors, and cursor.
-2. Added full versus incremental execution selection.
+2. Select full or incremental connector execution from the requested mode.
-3. Added retry loop with configurable maximum attempts and dead-letter result generation.
+3. Publish successful documents before returning the new cursor.
-4. Published before saving the next cursor to make replay safe after failures.
+4. Retry connector/publisher failures up to max_retries.
-5. Added run status, attempt count, batch, and dead-letter result structures.
+5. Return a dead-letter-compatible result after final failure.
+6. Dispatch webhook payloads to the configured connector.
+7. Add orchestration and fixed-run-ID tests.

.agents/plans/wf-decompose-b4be4be4fd0c/steps/step-4.md

@@ -1,7 +1,9 @@
 # Step 4 — Implementation Detail
 
-1. Added tenant equality validation before publication to prevent cross-tenant writes.
+1. Filter publication batches to matching tenant and ACL tenant identifiers.
-2. Published normalized documents through an injected knowledge-store port.
+2. Reject deleted documents from normal upsert publication.
-3. Recorded publication audit events containing tenant, run, action, count, and timestamp without content or secrets.
+3. Return published/rejected counts with tenant and timestamp context.
-4. Implemented retention-based purge of expired deleted-document tombstones.
+4. Provide tenant-scoped deletion helper for tombstones and retention workflows.
-5. Documented managed secret usage, ACL filtering expectations, transport/storage protection, and lifecycle rules.
+5. Provide structured audit-event creation without secrets or content.
+6. Validate provenance, content hashes, and cross-tenant documents.
+7. Document managed secrets, audit retention, source deletion, and downstream ACL enforcement.

.agents/plans/wf-decompose-b4be4be4fd0c/steps/step-5.md

@@ -1,8 +1,9 @@
 # Step 5 — Implementation Detail
 
-1. Added pyproject.toml with Python 3.11 target and pinned Ruff 0.6.9, mypy 1.11.2, and pytest 8.2 range for CI/dev checks.
+1. Define Python 3.11 package metadata and pytest 8.3.3 test extra.
-2. Added GitHub Actions workflow running installation, Ruff, mypy, and pytest.
+2. Add CI using actions/checkout@v4, actions/setup-python@v5, compileall, and pytest.
-3. Added Kubernetes deployment with readiness probe, non-root identity, restricted privilege escalation, and read-only root filesystem.
+3. Add Kubernetes deployment configuration with managed-secret mode and readiness probe.
-4. Added YAML configuration covering GitHub/SharePoint scopes, credential references, and trigger examples.
+4. Add connector and scheduled/webhook YAML examples.
-5. Added unit tests for contract validation, connector normalization, orchestration, and tenant isolation.
+5. Add setup, monitoring, retry/dead-letter, recovery, security, and lifecycle documentation.
-6. Added README, examples, operations, and validation-report documentation.
+6. Add README quick-start and security guidance.
+7. Add automated connector, orchestration, and tenant-isolation tests.

.agents/plans/wf-decompose-b4be4be4fd0c/steps/step-6.md

@@ -1,7 +1,10 @@
 # Step 6 — Implementation Detail
 
-1. Created representative full, incremental, scheduled, and webhook Trigger cases.
+1. Cover GitHub full ingestion and path/extension scoping in automated tests.
-2. Validated trigger mode, source, tenant, and connector requirements through TriggerValidator.
+2. Cover SharePoint full ingestion and PDF MIME normalization in automated tests.
-3. Documented fixture-based connector integration coverage and fake KAB-store downstream checks.
+3. Cover orchestration success status, run tracking, and repeatable run IDs.
-4. Specified acceptance checks for webhook idempotency, post-publication cursor advancement, deletion tombstones, ACL isolation, and dead-letter replay.
+4. Cover cross-tenant validation rejection in automated tests.
-5. Added scripts/e2e_validation.py returning a machine-readable readiness summary.
+5. Specify integration execution for full baselines and revision/delta updates.
+6. Specify scheduled retry/dead-letter and signed webhook validation scenarios.
+7. Specify downstream principal and tenant query isolation checks.
+8. Define release criteria requiring expected publication and zero unauthorized results.