Return a compact architecture summary suitable for injecting into an AI context window. Use this at the start of a session to orient the agent on the repo's major modules, their sizes, and the strongest coupling seams. Pass a path_scope to restrict the summary to a subdirectory. Avoid calling this repeatedly in a single session — cache the result and use targeted tools (cluster_of_file, impact_neighbors) for follow-up queries.

Return cross-cluster dependency edges that touch a file or cluster. Use this to identify architectural seams — places where two modules are tightly coupled across a boundary. Call this before refactoring or extracting a subsystem to understand what depends on it and what it depends on. Prefer high_coupling_seams when you want the strongest seams repo-wide.

Classify a user prompt into architectural scope and retrieval strategy. Use this as the first step when you receive a free-text task and are unsure which Big Indexer tools to call next. It maps the prompt to a scope type (symbol / file / cluster / repo) and a suggested retrieval sequence. Do NOT use this if you already know the target file or symbol — call the targeted tools directly.

Return the architectural cluster that contains a given source file. Use this when you need to know which module boundary or subsystem a file belongs to before making changes, so you avoid cross-cluster side effects. Do NOT use this for symbol lookup — use search_symbols instead.

Return staged architecture context for a task prompt using scope-first escalation. Use this as a one-shot alternative to chaining classify_prompt → cluster_of_file → impact_neighbors manually. It internally escalates from symbol → file → cluster → repo scope and stops as soon as enough context is gathered. Prefer this for general task prompts. Use individual tools when you need fine-grained control.

Return the strongest cross-cluster coupling seams in the repo or around a focal point. Use this to find the highest-risk architectural boundaries — pairs of clusters with the most inter-cluster dependencies. Call with no arguments for a repo-wide hotspot list. Pass a file or cluster to scope the result to its neighbourhood. Use boundary_edges when you already know the focal file/cluster.

Return the architectural blast radius of a symbol or file change. Use this before editing a function, class, or module to understand which other files and clusters are likely to be affected. Depth controls how many hops away from the focal point to traverse. Prefer depth=1 for quick checks and depth=2–3 for larger refactors.

Search for symbols (functions, classes, methods) by name or description. Use this to locate a specific function or class before navigating to it or understanding its context. Requires an index DB for ranked semantic results; falls back to graph name-matching when no DB is present. Prefer this over grep when you want architecture-aware ranking. Do NOT use for cluster lookup — use cluster_of_file instead.

Convert a natural-language task description into COV (code-operation vocabulary) tokens. Use this to extract the structured intent of a task before searching for behavioral twins. COV tokens represent the canonical operations implied by the task (e.g. "validate", "persist", "emit-event"). Do NOT use this for symbol search — use search_symbols instead.