This skill is a downstream specialization of the ctf-sandbox-orchestrator skill. It covers CTF sandbox challenges about provenance: dependency drift, registry pulls, build and release pipelines, artifact signing and the artifact a runtime consumer ends up using.
The agent walks the chain from source checkout to runtime consumer, compares declared and resolved versions, reconciles build-time and runtime state and reports the earliest point where the chain diverges. Replies are in Simplified Chinese unless English is requested.
What You Get
- A seven-step provenance chain from source checkout to runtime consumer
- A comparison approach for declared version, resolved version and shipped artifact
- A list of high-value evidence: lockfile drift, registry or mirror differences, install scripts and artifact hashes
- A report format that keeps the evidence as one short chain
Common Use Cases
- Finding where a dependency in a lab pipeline differs from the checked-in lockfile
- Comparing a signed artifact hash with the artifact a sandbox runtime loaded
- Tracing a build step or install script that changes a shipped image
What Makes It Unique
It treats checked-in manifests and live pipeline behavior as different things and asks for proof of runtime consumption, not only lockfile drift. It reports the earliest divergence point, 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. The agent reads manifests, logs and artifacts in the sandbox with the tools the user provides.