Flux

Publish events and query shared world state via Flux state engine. Use when agents need to share observations, coordinate on shared data, or track entity sta...

Security Scan

The skill is internally coherent: it implements a CLI wrapper around a Flux HTTP API, requests a single service token (FLUX_TOKEN), and its instructions and code align with the described purpose.

Purpose & Capability

Name/description (publish/query Flux world state) align with the included files and script: the bash CLI calls the Flux HTTP endpoints. Required credential FLUX_TOKEN and optional FLUX_ADMIN_TOKEN match documented API usage. Minor mismatch: SKILL.md / README say curl (and jq/python) are required/used but the registry metadata does not list required binaries.

Instruction Scope

SKILL.md instructs the agent to call the provided flux.sh script for publish/get/list/delete/admin actions and to contact the documented Flux API endpoints. The script only uses FLUX_* env vars and standard utilities (curl, jq, python3). It also exposes a directory endpoint that lists namespaces/entities (which may reveal global state across tenants on the public instance) — this is expected for a discovery feature but worth awareness.

Install Mechanism

No install spec; skill is instruction/script-only. No downloads or archive extraction. Script is included in the bundle so nothing external is fetched by the installer.

Credentials

Only FLUX_TOKEN is required (primaryEnv). FLUX_ADMIN_TOKEN and FLUX_URL are optional and justified by admin actions and local override. This is proportionate. Note: the skill will send requests to the default external host https://api.flux-universe.com — ensure you trust that service before providing credentials.

Persistence & Privilege

always is false and the skill does not request persistent platform-level privileges. It does not modify other skills or system-wide config. The agent may invoke the skill autonomously (normal default).

Assessment

This skill is a straightforward CLI wrapper for a Flux HTTP API and appears to do what it says. Before installing: (1) Confirm you trust the external endpoint (https://api.flux-universe.com) before setting FLUX_TOKEN there; the token grants API access. (2) Only provide FLUX_ADMIN_TOKEN if you need to run admin updates — it's a higher-privilege credential. (3) Ensure the runtime has curl and either jq or python3 (the script uses them but the package metadata didn't declare required binaries). (4) Be aware the skill can list namespaces/entities (discovery); avoid sharing tokens that span tenants if you need strict isolation.

Flux Skill

Flux is a persistent, shared, event-sourced world state engine. Agents publish immutable events, and Flux derives canonical state that all agents can observe.

Key Concepts

• Events: Immutable observations (temperature readings, status changes, etc.)
• Entities: State objects derived from events (sensors, devices, agents)
• Properties: Key-value attributes of entities (merged on update — only changed properties need to be sent)
• Streams: Logical event namespaces (sensors, agents, system)
• Namespaces: Multi-tenant isolation with token auth (optional, for public instances)

Prerequisites

Public instance: https://api.flux-universe.com (namespace purchased at flux-universe.com — name auto-assigned at purchase, e.g. dawn-coral) Local instance: http://localhost:3000 (default, override with FLUX_URL env var)

Authentication: Set FLUX_TOKEN to your bearer token. Required for the public instance. Optional for local instances with auth disabled.

Namespace Prefix

All entity IDs must be prefixed with your namespace: yournamespace/entity-name

Example with namespace dawn-coral:

./scripts/flux.sh publish sensors agent-01 dawn-coral/sensor-01 \
  '{"temperature":22.5}'
./scripts/flux.sh get dawn-coral/sensor-01

Entity IDs without a namespace prefix will be rejected on auth-enabled instances.

Getting Started

First, verify your connection:

./scripts/flux.sh health

Then check the directory to see what's available on the Flux Universe:

./scripts/flux.sh get flux-core/directory

The directory lists all active namespaces, entity counts, and total entities — a good way to discover what data is flowing through the system.

Scripts

Use the provided bash script in the scripts/ directory:

• flux.sh - Main CLI tool

Common Operations

Publish Event

./scripts/flux.sh publish <stream> <source> <entity_id> <properties_json>

# Replace dawn-coral with your namespace
# Example: Publish sensor reading
./scripts/flux.sh publish sensors agent-01 dawn-coral/temp-sensor-01 '{"temperature":22.5,"unit":"celsius"}'

Query Entity State

./scripts/flux.sh get <entity_id>

# Replace dawn-coral with your namespace
# Example: Get current sensor state
./scripts/flux.sh get dawn-coral/temp-sensor-01

List All Entities

./scripts/flux.sh list

# Filter by prefix
./scripts/flux.sh list --prefix scada/

Delete Entity

./scripts/flux.sh delete <entity_id>

# Example: Remove old test entity
./scripts/flux.sh delete test/old-entity

Batch Publish Events

# Replace dawn-coral with your namespace
./scripts/flux.sh batch '[
  {"stream":"sensors","source":"agent-01","payload":{"entity_id":"dawn-coral/sensor-01","properties":{"temp":22}}},
  {"stream":"sensors","source":"agent-01","payload":{"entity_id":"dawn-coral/sensor-02","properties":{"temp":23}}}
]'

Check Connector Status

./scripts/flux.sh connectors

Admin Config

# Read runtime config
./scripts/flux.sh admin-config

# Update (requires FLUX_ADMIN_TOKEN)
./scripts/flux.sh admin-config '{"rate_limit_per_namespace_per_minute": 5000}'

Use Cases

Multi-Agent Coordination

Agents publish observations to shared entities:

# Replace dawn-coral with your namespace
# Agent A observes temperature
flux.sh publish sensors agent-a dawn-coral/room-101 '{"temperature":22.5}'

# Agent B queries current state
flux.sh get dawn-coral/room-101
# Returns: {"temperature":22.5,...}

Status Tracking

Track service/system state:

# Replace dawn-coral with your namespace
# Publish status change
flux.sh publish system monitor dawn-coral/api-gateway '{"status":"healthy","uptime":3600}'

# Query current status
flux.sh get dawn-coral/api-gateway

API Endpoints

Event Ingestion:

• POST /api/events — Publish single event (1 MB limit)
• POST /api/events/batch — Publish multiple events (10 MB limit)

State Query:

• GET /api/state/entities — List all entities (supports ?prefix= and ?namespace= filters)
• GET /api/state/entities/:id — Get specific entity

Entity Management:

• DELETE /api/state/entities/:id — Delete single entity
• POST /api/state/entities/delete — Batch delete (by namespace/prefix/IDs)

Real-time Updates:

• GET /api/ws — WebSocket subscription

Connectors:

• GET /api/connectors — List connectors and status
• POST /api/connectors/:name/token — Store PAT credential
• DELETE /api/connectors/:name/token — Remove credential

Admin:

• GET /api/admin/config — Read runtime config
• PUT /api/admin/config — Update runtime config (requires FLUX_ADMIN_TOKEN)

Namespaces (auth mode only):

• POST /api/namespaces — Register namespace (returns auth token)

Notes

• Events auto-generate UUIDs (no need to provide eventId)
• Properties merge on updates — only send changed properties, existing ones are preserved
• Timestamp field must be epoch milliseconds (i64) — required by the API, auto-generated by flux.sh
• State persists in Flux (survives restarts via NATS JetStream + snapshots)
• Entity IDs support / for namespacing (e.g., scada/pump-01)