What problem does it solve?
Cross-origin pages often leak one bit of state (match vs no-match, logged-in vs anonymous) through side channels that scanners never chain together. This Skill provides a systematic methodology for combining inclusion methods, leak techniques, and detectable differences into working binary oracles that exfiltrate secrets character by character during authorized assessments.
Core Features & Use Cases
- Full XS-Leak matrix coverage: Maps inclusion methods (iframes, pop-ups, HTML elements, fetch) to leak techniques (event handlers, timing, connection-pool and event-loop limits, CORB, postMessage, Performance API) so no variant is missed.
- Oracle construction and confirmation: Provides ready-to-adapt JavaScript skeletons for timing oracles, hash-navigation onload leaks, and scriptless nested-object status leaks, plus statistical guidance to avoid false positives.
- Defense-resistant techniques: Covers connection-pool saturation and event-loop timing that work even under Site Isolation, framing protection, and reduced timer precision.
- Use Case: During an authorized engagement, a victim's private search endpoint returns different responses when a query matches. You frame the endpoint, use the hash-navigation onload oracle to test candidate prefixes, and recover a private token one character at a time, then document the finding with reproduction steps and remediation advice.
Quick Start
Ask the agent to assess a state-dependent cross-origin endpoint for XS-Search leaks and build a char-by-char exfiltration oracle using the appropriate inclusion method and leak technique.