Skip to content

Hyperia observability wiring — pulse hello/backoff, agent_died, feed consumption map #86

Description

@kordless

Plan: docs/plans/hyperia-observability-wiring.md — the reconciliation doc for Hyperia's Maximus epic (#102 theirs). Most of the pipe EXISTS: n8 pulse.rs already POSTs busy/idle to :9800/api/pulse/liveness with the pane token (poker should treat it as authoritative); Hyperia pulls fleet/events via :9801/mcp and tool-runs via :9802. n8-side gaps: (a) initial hello/idle pulse at monitor start so fresh containers register before first busy, (b) pulse error backoff, (c) agent_died events (overlaps podman plan P5). SSE feed exists at /fleet/events/stream; MCP polling suffices otherwise.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions