Claude Skills by sanchitmonga22

Use a locally installed Claude Code CLI from Codex for a read-only second opinion, a resumable conversation, a summarized handoff, or an exact explicitly requested native Claude command. Trigger when the user asks Codex to ask, consult, invoke, or use Claude or Claude Code.

Delegate a bounded task from Codex to the local Claude Code CLI. Use when the user explicitly asks Claude to investigate, implement, rescue, or continue work. Defaults to read-only; write delegation is confined to a generated git worktree.

Continue work in Codex from a known Claude Code session UUID while preserving a local transcript snapshot and a concise handoff summary. Use when the user asks to import, transfer, or continue a Claude session in Codex.

Run an independent Claude Code review from Codex. Use for a normal defect review, an adversarial design challenge, a base-branch comparison, or an explicitly approved cloud ultrareview. Reviews are read-only and return structured findings.

Inspect and manage Claude Code background sessions from Codex. Use for repository-scoped status, recent logs, best-effort results, stopping, respawning, removing, or attaching to a known Claude background job.

Check whether the local Claude Code CLI is installed, authenticated, and compatible, or manage the optional per-repository Stop review gate. Use for setup, command-not-found, auth, version, missing capabilities, or review-gate status/enable/disable.

Verify that CC for Codex is loaded, installed, enabled, and connected to a ready local Claude Code CLI; optionally run one explicitly authorized read-only live smoke test. Use after installation or upgrade, or when diagnosing whether the bridge works from the Codex app or CLI.

Implement an approved Codex heavy-track plan, validate it with real commands, prove end-to-end behavior when applicable, obtain the second read-only Claude diff review, and report verified status. Use after heavy-plan approval or in explicitly autonomous mode.

Explore a substantial Codex task and produce an evidence-backed implementation plan with alternatives, risks, validation, and a read-only Claude challenge. Use as the planning phase of heavy-track or when asked for a full implementation plan.

Run the full Codex-first workflow for risky, multi-file, architectural, or multi-session work: resolve the base, explore, plan, challenge the plan, stop at one gate, implement, validate, review, fix, and report. Use for heavy track, full workflow, or when resuming substantial Codex work.