What problem does it solve? Reading an entire document corpus or repository by hand is slow and error-prone. This Skill performs read-only, source-grounded summarization, extraction, comparison, and synthesis over bounded document sets or sealed repository roots, returning grounded content without mutating files or claiming governance authority. ## Core Features & Use Cases - Document mode: Submit bounded UTF-8 documents with label and content fields for mechanical extraction (context.documents.extract) or multi-source reasoning (context.documents.reason). - Repository mode: Provide a canonical repo_root and exact expected_repo_head for inventory, dependency tracing, or architectural synthesis (context.repository.extract / context.repository.reason). - Bounded routing contract: Send one JSON routing request per work unit to the co-packaged coordinator.py runtime, with quality_profile and effort_class tuned to the task depth. - Use Case: Ask the agent to trace control flow across a codebase or audit a set of log transcripts, and receive a grounded synthesis that you can spot-check against the original sources. ## Quick Start Ask the agent to summarize these documents or trace this repository using the context capability, providing either labeled document contents or the repository root with its expected head commit.
