cc-peer only ever sees local Claude Code sessions — ~/.claude/sessions, same-uid Unix socket or named pipe. There's no cross-machine leg, and there deliberately shouldn't be one built into cc-peer's own protocol handling: Claude's own cloud bridge is device-attestation-signed with a non-extractable key, which is a real boundary to respect, not a gap to work around.
The clean path is a cc-peer bridge for agent-comms (ExaDev/agent-comms), the same shape as its existing pi/claude-code/codex/mcp/opencode bridges. cc-peer stays exactly what it is today — a local peer speaking Claude Code's own protocol — and the bridge process is the thing that also joins the local agent-comms mesh and forwards traffic both ways.
Cross-machine reach then comes from agent-comms' own transport, not anything cc-peer has to build: its federation layer today, moving onto wire-mesh's relay/capability-token model per ExaDev/agent-comms#51. cc-peer doesn't need to depend on wire-mesh, or on agent-comms' transport internals, at all — only on the bridge existing.
Blocked by ExaDev/agent-comms#87 (the bridge itself).
cc-peer only ever sees local Claude Code sessions —
~/.claude/sessions, same-uid Unix socket or named pipe. There's no cross-machine leg, and there deliberately shouldn't be one built into cc-peer's own protocol handling: Claude's own cloud bridge is device-attestation-signed with a non-extractable key, which is a real boundary to respect, not a gap to work around.The clean path is a cc-peer bridge for agent-comms (ExaDev/agent-comms), the same shape as its existing pi/claude-code/codex/mcp/opencode bridges. cc-peer stays exactly what it is today — a local peer speaking Claude Code's own protocol — and the bridge process is the thing that also joins the local agent-comms mesh and forwards traffic both ways.
Cross-machine reach then comes from agent-comms' own transport, not anything cc-peer has to build: its federation layer today, moving onto wire-mesh's relay/capability-token model per ExaDev/agent-comms#51. cc-peer doesn't need to depend on wire-mesh, or on agent-comms' transport internals, at all — only on the bridge existing.
Blocked by ExaDev/agent-comms#87 (the bridge itself).