This skill is a downstream specialization of the ctf-sandbox-orchestrator skill. It covers challenges where the decisive behavior depends on request timing, async ordering, lock gaps or stale state, such as duplicate actions, stale reads or balance and privilege drift.
The agent identifies mutable state, builds a small deterministic concurrent sequence, compares a baseline run with a racing run and proves the final state drift from a clean reset. Replies are in Simplified Chinese unless English is requested.
What You Get
- A list of mutable boundaries to record: transaction scope, locks, retries, idempotency keys and cache invalidation
- A baseline-versus-race comparison method that changes one variable at a time
- A short chain template from request ordering to resulting drift
- Guidance to state the root cause: missing lock, weak idempotency, stale cache, delayed commit or retry side effect
Common Use Cases
- Reproducing a duplicate-action or stale-read bug in a CTF challenge
- Showing how request ordering changes a balance or privilege value in a lab
- Documenting a lock gap between a check and a write in a sandbox service
What Makes It Unique
It demands a minimal deterministic sequence and final-state proof from a clean reset, which discourages noisy stress runs. It hands off to the queue worker drift skill when the path becomes queue-dominated, and it runs only after the orchestrator skill has set sandbox assumptions.
Under The Hood
The skill is instruction text only: a SKILL.md, one reference checklist and an agent config file that disables implicit invocation. It ships no scripts and does not itself fetch pages, call APIs, run commands or write files. Concurrent test inputs would come from the agent’s own tools in the sandbox.