Update to Skillquiver

{
  "description": "Dispatches a bounded reviewer and preserves verified findings. Use when completed work needs read-only review before merge.",
  "included_files": [
    {
      "relative_path": "agents/openai.yaml",
      "size_in_bytes": 230
    },
    {
      "relative_path": "code-reviewer.md",
      "size_in_bytes": 5271
    }
  ],
  "name": "requesting-code-review",
  "skill_md_contents": "---\nname: requesting-code-review\ndescription: Dispatches a bounded reviewer and preserves verified findings. Use when completed work needs read-only review before merge.\n---\n\n# Requesting Code Review\n\nReview completed work against its requirements before it cascades. Give the\nreviewer the work product and exact review range, never the whole session\nhistory.\n\n## Choose the path\n\nHandle a standalone bounded read-only code review directly.\n\n- For a user's standalone bounded read-only review, inspect and report it\n  directly. Read only the named file and directly required context. Do not edit\n  the code. Do not enumerate the workspace.\n- For completed implementation work, dispatch a reviewer after a meaningful\n  task, major feature, complex fix, or before merge.\n- For one small file, use at most one reviewer unless its result lacks named\n  evidence; explain the missing evidence to any follow-up reviewer.\n\nFor the direct path, format each finding as\n`- <Severity>: <path>:<line> - <defect>. <impact and reasoning>.` Use a plain\n`path:line` when a valid clickable absolute path is unavailable.\nEvery finding must name the defect, impact, and reasoning.\nNever output a placeholder, empty link, or unfinished finding.\n\n## Define the review range\n\nPrefer the base commit recorded before implementation. Otherwise derive it from\nthe confirmed base branch:\n\n```bash\nBASE_SHA=$(git merge-base <base-branch> HEAD)\nHEAD_SHA=$(git rev-parse HEAD)\n```\n\nNever default to `HEAD~1`; it silently omits earlier commits in a multi-commit\ntask. On Windows, run these Bash commands in a Bash-capable shell or use their\nPowerShell equivalents.\n\n## Dispatch\n\nFill [code-reviewer.md](code-reviewer.md) with:\n\n- `[DESCRIPTION]`: concise summary of the completed work;\n- `[PLAN_OR_REQUIREMENTS]`: authoritative behavior and constraints;\n- `[BASE_SHA]` and `[HEAD_SHA]`: complete range to inspect.\n\nUse the host's general-purpose worker or closest equivalent. Keep the review\nread-only and require exact file/line evidence, calibrated severity, reasoning,\nand a merge verdict.\n\n## Preserve and resolve findings\n\nMaintain one accumulator of verified findings across every reviewer response.\nA later \"no additional findings\" must never erase an earlier verified issue.\n\nVerify each finding against the code and requirements before acting. Fix\nCritical and Important defects before continuing. When a finding is wrong,\nrespond with technical evidence from code or tests rather than deference. After\na material fix, request a focused re-review of the changed evidence, not an\nunbounded restart.\n\nThe final user-facing review is the severity-ordered synthesis of the\naccumulator with exact file and line references. Do not forward an intermediate\nreviewer message as the final verdict. State reviewed scope and any unverified\nsurface explicitly.\n"
}