self-review-connector

name	self-review-connector
description	Mechanical audit of the ADME OSDU connector — implementation correctness, test coverage, artifacts completeness, and security. Produces a scored report.
disable-model-invocation	true

Self-Review: ADME OSDU Connector

Scoring

Band	Score	Verdict
90+	—	READY
75–89	—	ALMOST
50–74	—	NEEDS WORK
< 50 or BLOCKER	—	NOT READY

Severity weights: BLOCKER (−10), MAJOR (−3), MINOR (−1).

---

Section A — Implementation (max 30)

Check each item. Award points for PASS, deduct for failures.

#	Check	How	Points
A1	AdmeOsduLakeflowConnect extends LakeflowConnect	grep "class AdmeOsduLakeflowConnect" connector/lakeflow/adme_osdu.py	3
A2	All 4 abstract methods implemented	grep for def list_tables, def get_table_schema, def read_table_metadata, def read_table	4
A3	read_table returns (Iterator, offset|None)	Read _read_domain, _read_legal_tags, _read_entitlements return types	3
A4	Pagination terminates — empty results return start_offset unchanged	Read _read_domain: check if not records: return iter([]), start_offset	4
A5	Governance tables return None offset	Check _read_legal_tags and _read_entitlements return (iter(rows), None)	3
A6	incremental_filter_template set on domain configs	grep incremental_filter_template connector/lakeflow/adme_osdu.py	3
A7	_flatten_record extracts all _DOMAIN_SCHEMA fields	Compare _flatten_record output keys against _DOMAIN_SCHEMA names	3
A8	No hardcoded real tokens or secrets	grep -r "eyJ|Bearer " connector/lakeflow/ — must find none	BLOCKER
A9	HTTP client uses with self._new_client() as client: pattern (closes cleanly)	Read _read_domain, _read_legal_tags, _read_entitlements	2
A10	Unknown table raises ValueError	Read read_table, get_table_schema, read_table_metadata	1

---

Section B — Testing (max 30)

#	Check	How	Points
B1	All 10 base AdmeConnectorTests tests present	grep "def test_" tests/unit/test_suite.py | wc -l ≥ 10	3
B2	All ADME-specific tests present	grep "def test_" tests/unit/adme_osdu/test_adme_osdu_lakeflow_connect.py | wc -l ≥ 8	3
B3	Simulate tests all pass	PYTHONPATH=. pytest tests/unit/adme_osdu/ -v — 0 failures	BLOCKER
B4	conftest.py autouse mock fixture present	grep "autouse=True" tests/unit/adme_osdu/conftest.py	2
B5	Mock handles cursor pagination (returns empty on cursor)	grep "body.get.*cursor" tests/unit/adme_osdu/conftest.py	2
B6	Mock handles incremental filter (returns empty on query)	grep "body.get.*query" tests/unit/adme_osdu/conftest.py	2
B7	Corpus has ≥ 3 records per domain table	python3 -c "import json; d=json.load(open('connector/simulator/corpus/wellbore.json')); print(len(d))" ≥ 3	2
B8	Legal tags corpus has correct wrapper format	python3 -c "import json; d=json.load(open('connector/simulator/corpus/legal_tags.json')); assert 'legalTags' in d"	2
B9	Entitlements corpus has correct wrapper format	assert 'groups' in json.load(open('connector/simulator/corpus/entitlements.json'))	2
B10	dev_config.json.template present (documents live config format)	ls tests/unit/adme_osdu/dev_config.json.template	2
B11	dev_config.json absent (not committed)	ls tests/unit/adme_osdu/dev_config.json must NOT exist	MAJOR
B12	Stubs directory present and functional	python3 -c "import sys; sys.path.insert(0,'tests/stubs'); import respx, tenacity"	3
B13	Root conftest.py injects stubs	grep "tests/stubs" conftest.py	1
B14	Test mode env var documented	grep CONNECTOR_TEST_MODE tests/unit/adme_osdu/conftest.py	1

---

Section C — Artifacts (max 20)

#	Check	How	Points
C1	CLAUDE.md present and has ADME-specific content	grep admesbx CLAUDE.md	3
C2	connector_spec.yaml present and valid YAML	python3 -c "import yaml; yaml.safe_load(open('connector_spec.yaml'))"	3
C3	connector_spec.yaml documents all required connection params	Check for base_url, data_partition_id, adme_api_client_id, tenant_id	3
C4	README.md present	ls README.md	2
C5	requirements.txt and requirements-dev.txt present	ls requirements*.txt	2
C6	.claude/agents/ has connector-dev and connector-tester	ls .claude/agents/	2
C7	.claude/skills/ has test-and-fix-connector and validate-connector-auth	ls .claude/skills/	2
C8	.claude/commands/ has develop-connector and validate-connector	ls .claude/commands/	1
C9	pyproject.toml has testpaths and pythonpath configured	grep testpaths pyproject.toml	2

---

Section D — Security (max 20)

#	Check	How	Points
D1	No hardcoded JWT tokens (eyJ...) in source	grep -r "eyJ" connector/ tests/ --include="*.py" | grep -v stub must be empty	BLOCKER
D2	No hardcoded passwords or secrets	grep -ri "password.*=" connector/ --include="*.py" — only test stubs	MAJOR
D3	dev_config.json in .gitignore	grep dev_config.json .gitignore	2
D4	static_access_token is Optional with no default	grep static_access_token connector/models/config.py shows Optional[str] = None	3
D5	No ssl_verify=False or cert bypass	grep -r "verify=False|ssl_verify" connector/ --include="*.py" must be empty	MAJOR
D6	No subprocess or eval in connector code	grep -r "subprocess|eval(" connector/lakeflow/ --include="*.py" must be empty	MAJOR
D7	Stub azure.identity raises on instantiation (not silently succeeds)	grep "raise RuntimeError" tests/stubs/azure/identity/__init__.py	2
D8	Corpus JSON has no real PII (real person names, real SSNs)	Review connector/simulator/corpus/*.json — synthetic data only	3

---

Section E — Consistency (max 10)

#	Check	How	Points
E1	Tables in TABLES constant match those in test replay_config and corpus directory	Compare AdmeOsduLakeflowConnect.TABLES with connector/simulator/corpus/ filenames	3
E2	Schema field names in _DOMAIN_SCHEMA match _flatten_record output keys	Manual comparison	3
E3	connector_spec.yaml params match _build_runtime option keys	Compare spec parameters[].name with options[...] calls in adme_osdu.py	2
E4	incremental_filter_template uses {watermark} placeholder (not {cursor} etc.)	grep incremental_filter_template connector/lakeflow/adme_osdu.py	2

---

Output Format

# ADME OSDU Connector Self-Review

## Results

| Section | Score | Max |
|---------|-------|-----|
| A Implementation | X | 30 |
| B Testing | X | 30 |
| C Artifacts | X | 20 |
| D Security | X | 20 |
| E Consistency | X | 10 |
| **Total** | **X** | **110** |

**Verdict: READY / ALMOST / NEEDS WORK / NOT READY**

## Findings

### BLOCKERS
- (none) / list

### MAJOR
- (none) / list

### MINOR
- (none) / list