[codex] Polish watchlist skill guardrails

[dd3ok]

Summary

• Narrow the watchlist skill trigger wording for pending results and scope pre-authorized recording to the current repo/workflow.
• Clarify safe source links, due_at wording, and timezone precedence in runtime and lifecycle docs.
• Compress the SKILL.md Add section while preserving the guardrails; Add section is now 249 words and the body is 745 words.
• Document strict-safety conservatism and open-item field expectations in English and Korean READMEs.
• Add regression checks for the new guardrail, concise Add section, and documentation markers.

Validation

• python3 -m unittest discover -s evals -p 'test_*.py' — 72 tests OK
• python3 evals/check_release_metadata.py — passed
• python3 evals/check_policy_markers.py — passed
• python3 evals/check_semantic_cases.py — 23 cases passed
• python3 evals/check_watchlist.py examples/WATCHLIST.example.md --strict-format --strict-safety --require-archive-section — passed
• python3 evals/check_watchlist.py .agents/skills/watchlist-md/assets/WATCHLIST.template.md --strict-format --strict-safety --require-archive-section — passed
• python3 .agents/skills/watchlist-md/scripts/validate_watchlist.py .agents/skills/watchlist-md/assets/WATCHLIST.template.md --strict-format --strict-safety --require-archive-section — passed
• git diff --check origin/main...HEAD — passed

[gemini-code-assist]

Code Review

This pull request updates the watchlist-md skill and its documentation to clarify field requirements, timezone precedence for ID generation, and safety policies regarding sensitive links. Key changes include the addition of the --strict-safety flag documentation, refined validator rules for open items, and updated test cases to verify these guidelines. Feedback suggests improving consistency between the skill definition and the READMEs by explicitly listing all required fields and using precise field names like done_when instead of descriptive labels.