← ClaudeAtlas

remote-session-reviewlisted

Use when an agent needs to review remote session task status, output, audit records, artifacts, diffs, tests, runtime-memory updates, interrupts, revocation state, or residual risk before declaring Remote Dev Skillkit work complete.
EitanWong/remote-dev-skillkit · ★ 0 · Code & Development · score 62
Install: claude install-skill EitanWong/remote-dev-skillkit
# Remote Session Review Use this skill before telling the user a remote session task is complete. ## Review Checklist - Evidence source is known: session id, task id, gateway, bundle path, artifact ids, and audit source are explicit or discovered from the current context. - Runtime memory used by the task is identified, and any newly discovered safe facts are written back only after evidence review. - Before declaring completion, probe read-only context such as `rdev doctor`, `rdev mcp tools`, session status, task status, artifact listings, audit exports, workspace path, framework install path, network reachability, tunnel/mesh selection, and verifier availability when those details are not already explicit. - If session id, task id, gateway, workspace, evidence bundle path, artifact location, adapter choice, tunnel/mesh authorization, interrupt policy, framework install path, or expected verifier is unclear, ask before declaring completion. - Do not invent gateway URLs, paths, ticket codes, root keys, release URLs, checksums, workspace roots, adapter choices, or interrupt policies from examples or placeholders. - Keep path and configuration neutral. Evidence paths, workspace roots, artifact locations, verifier paths, framework directories, and gateway URLs must come from task metadata, current context, MCP/CLI output, manifest metadata, or explicit human/operator confirmation. - Treat relay, tunnel, mesh, VPN, SSH, proxy, and DNS facts as evide