This skill is a specialization of the ctf-sandbox-orchestrator skill for challenges where the decisive branch lives in browser state. The agent identifies the active persistence surface, records origin, scope and expiry, and ties each stored item to one concrete effect such as a request header, a rendered branch or a cached response.
It compares clean-state and changed-state runs with one variable altered at a time. The skill states that it runs only after the orchestrator has set sandbox assumptions.
What You Get
- A three-step workflow: map browser state surfaces, tie state to runtime behavior, reduce to the decisive persistence chain
- A reference checklist covering cookie attributes, storage buckets, service worker scopes, cached responses and boot-time globals
- A comparison method that changes one variable between clean-state and changed-state runs
- A hand-off rule to the broader web-runtime skill when the problem leaves browser persistence
Common Use Cases
- Explaining why a page in a CTF behaves differently after load because of stored client state
- Documenting which cookie or storage key sets a role, route or feature flag
- Checking whether a service worker or offline cache changes the response a page receives
What Makes It Unique
It keeps each stored item attached to one origin and one observed effect, and it separates UI-only state from state the backend accepts. Its listed pitfalls include listing storage contents without showing which item matters, and treating cached UI data as server-side authorization.
Under The Hood
The skill is instruction text only: a SKILL.md, one reference checklist and an agent configuration file. It ships no scripts. The method involves loading pages in a browser, reading and changing cookies and storage, and replaying requests, so the agent’s own browser and shell tools do that work in the sandbox the orchestrator set up. The skill does not fetch pages or call APIs on its own.