Fix 3 IS connection tests (slack, outlook) in flow test suite

[bai-uipath]

The three connector-driven flow tasks were failing because the test connections live in Shared/uipath-maestro-flow, but the prompts and the Outlook post-check assumed Shared. Added a one-paragraph folder hint to the three task YAMLs and updated check_outlook_trigger_inbox.py to look in the right folder.

Test results

Task	Before	After
skill-flow-outlook-trigger-inbox	ERROR (timeout)	SUCCESS 1.000 (463s)
skill-flow-slack-channel-description	FAILURE 0.250	SUCCESS 1.000 (567s)
skill-flow-slack-weather-pipeline	FAILURE 0.000	SUCCESS 1.000 (1004s)

[github-actions]

Claude finished @bai-uipath's task in 1m 34s —— View job

---

PR Review: Fix 3 IS connection tests (slack, outlook) in flow test suite

• Read review criteria and project conventions
• Read full diff and existing PR conversation
• Read each changed file in full context
• Check for correctness, consistency, and conventions
• Post review findings

---

Summary

Fixes three connector-driven flow test tasks that were failing because test connections live in Shared/uipath-maestro-flow but prompts and the Outlook post-check assumed Shared. Adds a folder hint to the three task YAMLs and updates check_outlook_trigger_inbox.py to use folders get with the correct path.

Change-by-Change Review

1. tests/tasks/.../slack_channel_description/slack_channel_description.yaml

Severity: Low

4-line addition to initial_prompt directing the agent to the correct Orchestrator folder. Correct and well-placed (after the existing skill-loading instructions, before success_criteria).

Minor: line 29 uses -o json instead of --output json. Every other task YAML in this test suite uses the long form --output json (see smoke/registry_discovery.yaml:31, hitl/*.yaml, etc.). The project convention (CLAUDE.md) requires --output json when output is parsed programmatically — while this is a prompt hint rather than a harness command, using the long form keeps agent-facing instructions consistent with what the skill itself teaches.

2. tests/tasks/.../slack_weather_pipeline/slack_weather_pipeline.yaml

Severity: Low

Same 4-line addition as above. Same -o json vs --output json note applies (line 28).

3. tests/tasks/.../outlook_trigger_inbox/outlook_trigger_inbox.yaml

Severity: Low

Same 4-line addition. Same -o json note (line 45).

4. tests/tasks/.../outlook_trigger_inbox/check_outlook_trigger_inbox.py

Severity: OK

Clean, correct refactor:

• SHARED_FOLDER_PATH → TEST_FOLDER_PATH with updated path — naming matches the new semantics (line 27).
• _find_shared_folder_key() → _find_test_folder_key() — switches from folders list + filter loop to folders get with direct path lookup. This is a strict improvement: fewer API calls, simpler code, correct response shape handling (.get("Data", {}).get("Key") for single-object response vs the old list iteration).
• Error messages and docstrings updated consistently.
• All callers updated (line 69).
• No secrets or personal paths.

What's Missing

Nothing identified. This is a focused test-infrastructure fix with no skill-doc or CODEOWNERS implications.

Area Ratings

Area	Status	Notes
Frontmatter	N/A	No skill SKILL.md files changed
E2E Tests	OK	Existing tests fixed, no new tasks needed
Skill Body	N/A	No skill docs changed
References & Assets	N/A	No references changed
Repo Hygiene	OK	Scoped to the affected test tasks, no secrets, no cross-skill refs

Issues for Manual Review

• The test results in the PR description show all three tasks passing. Consider confirming CI passes on this branch before merge, since these are integration tests that depend on live tenant state.

Conclusion

Clean, well-scoped fix. The only nit is -o json → --output json in the three YAML prompt hints for consistency with the rest of the test suite. Approving with that optional polish.

Fix -o json → --output json →