basebuild-execution-advisorlisted

# Basebuild Execution Advisor Use `get_execution_advice` only after an idea or plan has a persisted bounded assessment. Pass exactly one `ideaId` or `planId`. ## Interpretation - Treat the recommendation as advice, never as permission to launch. - Prefer the role that matches the next action: `planner` for design/spec work; `coder` for implementation. - Read `excluded` before proposing a route the user cannot execute. - Read every factor and `sourceFreshness`; a high aggregate score does not override a hard compatibility exclusion. - `high` confidence means the route has fresh public profile and local availability evidence. `medium` or `low` means material evidence is missing or stale. - A user override is intentional and wins while compatible. Explain any hard gate that prevents it. - Missing capacity means unknown—not unlimited and not exhausted. Ask the user when the decision depends on uncertain quota or timing. - Stale capacity can inform a comparison but cannot justify claiming that a provider has enough quota. ## Privacy boundary The ranking runs locally. Tool output is allowlisted to persisted assessment metadata, provider/model identifiers, coarse capacity evidence, public profile references, factor explanations, and freshness/confidence. It excludes credentials, account identifiers, project text, source, messages, questionnaire answers, raw usage, diffs, logs, and absolute paths. When the active chat uses an external provider, returning tool output crosses tha