Telemetry on corpus usage

There is no telemetry on the consumption of the corpus, which makes it's maintenance left to heuristics ( cfo needs to review xyz when stale )

Human effort allocation is currently blind or left to hardcoded decisions. Human review time is the scarcest resource in an agent-written corpus. Spending it without a quantitative approach is spending most of it badly.

Heavily consumed concepts carry more risk. An error in a concept loaded into a thousand contexts a day propagates further than one nobody opens. This is blast radius, and it is currently unobservable and the concept might not even have a stale field.

Cold concepts are undiagnosable. A concept that is never read might be redundant, might be badly titled, might be described in terms no consumer queries for, or might be the one thing pulled once a year for an audit.

Usage data does not resolve which, but without it you cannot even ask.

At runtime of the corpus querying some form of unique-id logging is updated < human_bod_unique_id_usage.md or agent_x_unique_id_usage.md > could be generated and potentially harvested through git for analysis

Start by locating the runtime corpus-querying entry point and reviewing the proposed human_bod_unique_id_usage.md and agent_x_unique_id_usage.md files. Determine how unique-id usage records could be generated and harvested through git; done means the issue's telemetry data is available for usage analysis.

Written by the indexing model from the issue text.