Separate privileged acquisition from unprivileged clients.
RAPL access can require elevation; the Unix socket becomes a narrow, auditable boundary.Alternative not selected: Letting every Python or CLI process read protected counters directly.
A Rust daemon and CLI boundary for local energy and system-telemetry capture, designed around explicit evidence classes, deterministic validation and replayable traces.
Hardware energy counters can require elevated access, while developers and benchmark tools need a safe, reproducible, unprivileged workflow that does not overstate noisy measurements.
Chakib designed the daemon/client boundary, protocol and validation model, built the RAPL prototype and evolved it into a backend-independent trace-first foundation with deterministic test scenarios.
A privileged acquisition daemon reads raw counters and serves a strict local socket protocol. Unprivileged CLI and Python clients capture source descriptors and sample batches into replayable run artifacts; interpretation stays offline.
RAPL access can require elevation; the Unix socket becomes a narrow, auditable boundary.Alternative not selected: Letting every Python or CLI process read protected counters directly.
Raw evidence remains replayable while baselines, aggregation and benchmark claims can evolve offline.Alternative not selected: Embedding joules-per-token, scoring or report generation in the hot acquisition path.
The public v0 Rust prototype proves RAPL, protobuf framing and Unix-socket collection. A local protocol-first rewrite adds replayable trace artifacts and deterministic validation; that rewrite was not publicly reachable during the audit.
Rust, Linux, RAPL, Unix sockets
Protobuf, Binary frames, Typed descriptors
Deterministic daemon, Golden tests, Replayable traces