ruv
d46eb8a4b2
harness/ruview (@ruvnet/ruview): - guardrails: digit gate now sees numbers inside code spans; F1-style metric tokens followed by ':' or a nearby number are no longer scrubbed (fail-open regressions in the honesty gate) - mcp-server: tools/call requests serialize through a FIFO promise chain (hardware/mutating tools never overlap) while ping/tools/list stay immediate; stdin close drains in-flight responses before exit - tools: which() no longer memoizes negative lookups tools/ruview-mcp (@ruvnet/rvagent): - index: realpath invoked-directly guard — library import no longer connects a stdio transport to the consumer's process - http-transport: explicit allowedOrigins is exact-match only (localhost any-port convenience applies only with no configured allowlist); session map gains maxSessions=64 + 5min idle TTL sweep - train-count: job records persist the child pid and reconcile stale 'running' status after a server restart (exit-code marker or dead pid) - config: cog binary candidates ordered by process.arch .github/workflows/ruview-npm-release.yml: port the full ADR-265 D1 gate (version-literal check, unpacked-size budget, tarball-install smoke test) from npm-packages.yml so the publish path enforces what the header claims. Tests: harness 30→36, rvagent 99→112, all passing. Co-Authored-By: claude-flow <ruv@ruv.net>
npx @ruvnet/ruview — RuView WiFi-sensing operator harness
An AI agent harness that knows how to operate RuView (WiFi-DensePose): onboard a
newcomer, provision an ESP32 CSI node, calibrate a room, train pose models, and —
crucially — refuse to overstate accuracy. Minted from the RuView monorepo via
metaharness and hardened per ADR-182.
WiFi sensing infers coarse pose/presence/breathing from Channel State Information.
It is not a camera. Every accuracy number this harness emits must be MEASURED
against a baseline — that rule is enforced in code (ruview_claim_check).
Quick start
npx @ruvnet/ruview # onboard — pick a setup path
npx @ruvnet/ruview claim-check --file REPORT.md # the honesty guardrail (non-zero exit on untagged claims)
npx @ruvnet/ruview verify # run the deterministic proof (VERDICT: PASS)
npx @ruvnet/ruview doctor # self-check (tools + optional kernel/host)
npx @ruvnet/ruview --help
The operator tools are pure Node and run with zero install weight — the
package has no dependencies at all (ADR-263 O3). doctor / install can
additionally use @metaharness/kernel + a host adapter if you install them
(npm i @metaharness/kernel @metaharness/host-claude-code); everything else
runs without them.
Tools (ruview_*)
Exposed both as CLI verbs and as an MCP server (npx @ruvnet/ruview mcp start):
| Tool | What it does |
|---|---|
ruview_onboard |
Pick docker-demo / repo-build / live-esp32; print the next command |
ruview_claim_check |
Lint text for untagged / overstated accuracy claims (guardrail) |
ruview_verify |
Run verify.py deterministic proof → VERDICT |
ruview_node_monitor |
Assert CSI is flowing on an ESP32 (read-only) |
ruview_calibrate |
ADR-151 room pipeline (baseline→enroll→train-room→room-watch) |
ruview_node_flash |
Build+flash firmware (Windows/ESP-IDF; mutating, guarded) |
Every tool is fail-closed: missing repo / python / binary / port → an honest negative, never a fabricated success.
Skills
Host-neutral playbooks in skills/ (onboard, provision-node, calibrate-room,
train-pose, verify). npx @ruvnet/ruview skill <name> prints one.
Use as a Claude Code MCP server
The bundled .claude/settings.json registers the ruview MCP server
(npx -y @ruvnet/ruview mcp start). Drop this package's .claude/ into a repo, or run
npx @ruvnet/ruview install --host claude-code.
Hosts
claude-code (bundled), and via metaharness host adapters: codex, opencode, copilot, pi-dev, hermes, rvm, github-actions.
License
MIT © ruvnet