What problem does it solve? Coding agents often start tasks in unfamiliar repositories without understanding the architecture, forcing them to rediscover system structure through slow, error-prone search. This Skill provides on-demand access to the System Context Compiler's verified machine model of the repository, so planning and edits are grounded in evidence-backed facts rather than assumptions. ## Core Features & Use Cases - Session Startup Context: Call system_context for a fused view of the system atlas, API surface, coverage, and omissions with an adaptive token budget. - Task-Specific Context Packs: Use task_context, component_context, flow_context, and impact_context to drill into components, runtime flows, and cross-layer change impact before editing. - Staleness Verification: Run verify_context to detect stale facts and re-index with scc index so outdated information never enters trusted context. - Use Case: Before modifying a shared API contract, call impact_context on the touched files to see downstream consumers, invariants, and failure behavior, then honor those constraints in your change. ## Quick Start Ask the agent to call system_context at the start of your session, then request task_context with your specific goal before planning any code changes.
