Document manual-to-scheduled ladder policy

[Fearvox]

Summary

• Add docs/ops/manual-to-scheduled-ladder-v0.md.
• Document the 0-2 / 3-4 / 5+ clean-cycle ladder.
• Require public_safety on future gate outputs used for batch or scheduled-automation proposals.
• Explicitly state this change creates no schedules, trigger changes, autopilot reactivation, skill mutation, or runtime mutation.

Verification

• git diff --check origin/main...HEAD
• rg -n "0-2|3-4|5\\+|No schedule creation|no trigger change|autopilot reactivation|public_safety|This document is policy only" docs/ops/manual-to-scheduled-ladder-v0.md
• public-surface added-lines scan: hits were limited to policy terms such as secret/token/OAuth/SSH, with no actual credentials or private paths.

[Copilot]

Pull request overview

Adds an ops policy document that defines when manual Workbench gate workflows may progress to operator-approved batching and then to scheduled-automation proposals, with a required public_safety field for future gate outputs used as evidence.

Changes:

• Introduces a 0–2 / 3–4 / 5+ clean-cycle ladder for manual → batch → scheduled-proposal eligibility.
• Defines what qualifies as a “clean manual cycle” and what evidence must be preserved in closeout.
• Requires a top-level public_safety block in future gate outputs used to justify batching or scheduled-automation proposals.

[Copilot]

Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.

[Fearvox]

Capy /capy: Copilot second-pass review is currently erroring. Please run a bounded review on the current HEAD f56bc78e5b4c009852c8f2c52c477e5bec746186 only. Findings first, focus on ladder-rule drift, wording inconsistencies, YAML copy-paste safety, and any reason this should stay FLAG instead of moving forward.

[Copilot]

Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.

[Fearvox]

Reviewer surface status update:

• All substantive Copilot findings on this PR have been addressed on-branch (f56bc78) and the corresponding inline threads were replied to.
• The repo's last public Capy review evidence says the Git Dialogue responder remains disabled, so Capy silence should not be read as an unresolved code defect here.
• The remaining blocker is Copilot review availability only; recent re-requests have produced transient unable to review responses and no new substantive findings.

Verdict: FLAG until Copilot produces either a fresh finding or an explicit clean pass.