fix(fleet): handle non-object JSON command logging

[Bre77]

Intent

Fix production regression: Commands._log_request_result (tesla_fleet_api/tesla/fleet.py, added in PR #64 / v1.6.4 as DEBUG-only troubleshooting logging with explicit no-behavior-change intent) unconditionally called data.get('response') assuming a dict body. Any JSON-legal non-dict response (real case: Teslemetry list_authorized_clients returning null) crashed AttributeError after the HTTP request already succeeded, surfacing as HTTP 500 in a captain's production Home Assistant Powerwall reconfigure flow (HACS integration energy reconfigure broke; router-beta avoided it only by pinning 1.6.3). Fix: guard _log_request_result with isinstance(data, dict) before calling .get(); for any non-dict body (null, list, scalar) log result=success and return early, preserving the function's 'must never break a successful request' intent from PR #64. Added regression tests in tests/test_command_logging.py for null/list/scalar JSON bodies, and fixed a latent None-vs-unset ambiguity in the test file's _fake_response helper (it previously treated an explicit json_body=None the same as 'not passed', silently defaulting to {}) so the null-body test can actually exercise a null body. Also added one line to AGENTS.md documenting the invariant. Deliberately scoped narrowly to this one bug per the task brief - did not touch the separate key-permissions hardening issue, which a different crew is handling in its own PR.

What Changed

• Guarded _log_request_result so successful requests with JSON null, lists, or scalar bodies no longer crash command-result logging.
• Added regression coverage for non-dict JSON response bodies and fixed the test response helper so explicit None bodies are exercised correctly.
• Documented REST endpoints that may return non-object JSON bodies across the public docs.

Risk Assessment

✅ Low: The change is narrowly scoped to preventing debug logging from raising on JSON-legal non-dict response bodies, with focused regression coverage and no observed behavioral risk beyond preserving the successful response value.

Testing

Inspected the targeted diff, ran the command-logging regression suite successfully, captured an API-level transcript for null/list/scalar JSON responses, restored uv-generated egg-info changes, removed local test caches, and confirmed the worktree was clean afterward.

Evidence: Non-dict JSON request transcript

CASE null body from list_authorized_clients returned: None log: DEBUG:command=list_authorized_clients transport=fleet result=success CASE list body returned: [{'id': 1}] log: DEBUG:command=clients transport=fleet result=success CASE scalar body returned: True log: DEBUG:command=flag transport=fleet result=success

CASE null body from list_authorized_clients
returned: None
log: DEBUG:command=list_authorized_clients transport=fleet result=success

CASE list body
returned: [{'id': 1}]
log: DEBUG:command=clients transport=fleet result=success

CASE scalar body
returned: True
log: DEBUG:command=flag transport=fleet result=success

Pipeline

Updates from git push no-mistakes

✅ **intent** - passed

✅ No issues found.

✅ **Rebase** - passed

✅ No issues found.

✅ **Review** - passed

✅ No issues found.

✅ **Test** - passed

✅ No issues found.

• git diff 68f79c5b95d1ad5394fae1ec4247bca99d341af3..70d2479c2e3db3241fc85fe214e2c0dc1cc43bd4 -- tesla_fleet_api/tesla/fleet.py tests/test_command_logging.py AGENTS.md
• uv run pytest tests/test_command_logging.py
• mkdir -p /tmp/no-mistakes-evidence/01KXAXMHRPTR061M4VTS6C4NRH && uv run python - <<'PY' > /tmp/no-mistakes-evidence/01KXAXMHRPTR061M4VTS6C4NRH/non_dict_json_request_transcript.txt ... PY
• git status --short after cleanup

✅ **Document** - passed

✅ No issues found.

✅ **Lint** - passed

✅ No issues found.

✅ **Push** - passed

✅ No issues found.