docs: tighten global agent guidance

fix(skills): clarify codex-host review coverage

docs: clarify plan-loop closeout flow

refactor(skills): deduplicate loop skills and strengthen execute-continuously

fix(skills): autonomous closing chain — self-test, simplify, review-loop run without user gates

fix(skills): strengthen verification language — no excuse to skip automated e2e

fix(skills): strengthen e2e verification and simplification in plan-loop and self-test

fix: lower new-task confidence threshold from 95% to 90%

fix: harden cross-skill invocation and add gotchas to workflow skills

fix(self-test): add surface identification and user-action gotchas

fix: explicitly say 'invoke the self-test skill' to ensure agents load it

fix: require self-test after behavioral fixes in review-loop

fix: adjust self-test emphasis per feedback

docs: align agent-philosophy.md self-test language with AGENTS.md

docs: strengthen self-test emphasis across skills and AGENTS.md