Concurrency and Synchronization Audit

Use the Claude Code command as the canonical workflow instead of maintaining a duplicated snapshot.

Load the Canonical Command

1. Read .claude/commands/audit-concurrency/SKILL.md completely.
2. Read .claude/commands/_audit-common.md and .claude/commands/_audit-severity.md completely before auditing.
3. Follow the canonical command's dimension order by blast radius, delta-first skipping of unchanged dimensions, deduplication, and report format.

Adapt Claude Conventions to Codex

• Treat the user's text after the skill request as the command arguments described by the canonical skill's argument-hint.
• Resolve each /audit-<name> reference to that audit's SKILL.md under .claude/commands/, and hand off what the canonical delegates: lock-tracker internals and the access-model guard to /audit-ecs, ground-cover/sky pipeline semantics to /audit-exterior, sandbox trust-boundary questions to /audit-safety, and the debug-server command surface to /audit-tooling.
• Apply the speculative-fix guardrail to Vulkan-sync findings: barrier bugs are invisible to cargo test, so a finding stays a HYPOTHESIS until a BYRO_VALIDATION or RenderDoc capture confirms it.
• Prefer the repository's codebase-memory graph tools for lock acquisition sites, scheduler access declarations, and worker/data-flow discovery; use text search for barrier and stage-mask literals, Mutex/RwLock patterns, thread::spawn/rayon sites, shader and compute files, docs, and graph gaps.
• Interpret Claude Task-agent instructions as Codex sub-agent delegation only when delegation is available and allowed. Otherwise audit one dimension at a time.
• Preserve the canonical output path and do not create GitHub issues; publishing remains a separate audit-publish action.

Do not copy the canonical command body into this skill. The Claude command is the single source of truth.