This skill is a downstream specialization of the ctf-sandbox-orchestrator skill. It covers challenges where request interpretation changes between a proxy, middleware and a backend parser: path normalization, header canonicalization, Host derivation and transfer framing.
The agent maps every parsing hop, builds paired requests that differ in one dimension, captures proxy and backend evidence and reduces the result to one minimal differential path. Replies are in Simplified Chinese unless English is requested.
What You Get
- A hop-by-hop map of proxy, gateway, app server and downstream parsing
- A paired-request method that changes one canonicalization dimension at a time
- Guidance to compare proxy logs, backend logs and route matches
- A root-cause statement format: path drift, header ambiguity, framing differential or host derivation
Common Use Cases
- Explaining why a proxy and a backend route the same path differently in a lab
- Documenting Host or forwarded-header ambiguity in a sandbox CTF
- Recording a length or chunk framing mismatch between two hops
What Makes It Unique
It insists on one changed field per request pair so the cause stays attributable, and it asks for backend route evidence as well as front proxy logs. It hands off to the runtime routing skill when no framing issue is involved, 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 or payloads and does not itself fetch pages, call APIs, run commands or write files. Any requests are sent with tools the user provides, against sandbox targets they are authorized to test.