ELYREON n8n Workflow Rescue

Registre officiel Autres

Static n8n workflow validator: workflow errors and repair suggestions. No execution.

Que pouvez-vous faire avec ELYREON n8n Workflow Rescue ?

  • Open Source & Local-First: Runs entirely on your local machine; zero external hosted infrastructure or cloud registry required.
  • Persistent Autonomous Workers: Dots maintain local SQLite state, surviving process restarts and machine reboots.
  • Event & Schedule Wakeups: Built-in cron/interval schedulers and idempotent event queues trigger autonomous workflows.
  • Bounded Drex Capabilities: Hard invariants prevent workspace escapes, credential reading, and arbitrary shell execution. Consequential mutations require task-scoped, time-limited (TTL) grants.
  • Tamper-Evident Receipts: Every evaluated action generates an immutable audit receipt with automated secret scrubbing.
  • Pluggable Model Routing: Supports cost-aware model tiers (cheap local vs. frontier escalation).
  • Optional Integrations: Seamless adapters for NousResearch Hermes Agent plugins and NOVA workflows.

Installation

Ajoutez ceci à la configuration MCP de votre client (Claude Desktop : claude_desktop_config.json ; Cursor : .cursor/mcp.json ; VS Code et les autres utilisent la même forme « mcpServers »).

Remote (streamable-http)
Claude Code
claude mcp add --transport http elyreon-n8n-workflow-rescue 'https://elyraon.com.co/agents/mcp' --header 'Authorization: <Free token: POST https://elyraon.com.co/agents/trial with JSON {}. Enter Bearer TOKEN.>'
Configuration JSON (Claude Desktop, Cursor, VS Code)
{
    "mcpServers": {
        "elyreon-n8n-workflow-rescue": {
            "type": "http",
            "url": "https://elyraon.com.co/agents/mcp",
            "headers": {
                "Authorization": "<Free token: POST https://elyraon.com.co/agents/trial with JSON {}. Enter Bearer TOKEN.>"
            }
        }
    }
}

Documentation

Tirée du README du projet sur GitHub, actualisée chaque semaine. Les liens ouvrent GitHub.

Sommaire
  1. Drex Dotz
  2. Highlights
  3. Architecture
  4. Installation
  5. 5-Minute Walkthrough
  6. Public Examples Included
  7. Known Limitations
  8. License

Drex Dotz

Persistent autonomous AI workers with bounded capabilities.
Give agents jobs, not root access.

Drex Dotz provides persistent, autonomous workers ("Dotz") that operate under a strict capability firewall. Instead of granting agents root access or unconstrained shell and filesystem permissions, every consequential action flows through the Drex Agent Firewall:

Dot
→ capability request
→ Drex
→ ALLOW / BLOCK / bounded grant
→ execution
→ receipt

Highlights

  • Open Source & Local-First: Runs entirely on your local machine; zero external hosted infrastructure or cloud registry required.
  • Persistent Autonomous Workers: Dots maintain local SQLite state, surviving process restarts and machine reboots.
  • Event & Schedule Wakeups: Built-in cron/interval schedulers and idempotent event queues trigger autonomous workflows.
  • Bounded Drex Capabilities: Hard invariants prevent workspace escapes, credential reading, and arbitrary shell execution. Consequential mutations require task-scoped, time-limited (TTL) grants.
  • Tamper-Evident Receipts: Every evaluated action generates an immutable audit receipt with automated secret scrubbing.
  • Pluggable Model Routing: Supports cost-aware model tiers (cheap local vs. frontier escalation).
  • Optional Integrations: Seamless adapters for NousResearch Hermes Agent plugins and NOVA workflows.

Architecture

                    USER / HERMES / API
                           │
                           ▼
                       DOT RUNTIME
                           │
              identity / lifecycle / events
                           │
          ┌────────────────┼────────────────┐
          ▼                ▼                ▼
      Researcher       Developer         Operator
          │                │                │
          └────────────────┼────────────────┘
                           ▼
                         DREX
                  capability firewall
                           │
             ┌─────────────┼─────────────┐
             ▼             ▼             ▼
           files         shell          web/API
        workspace       commands       connectors
                           │
                           ▼
                        RECEIPTS

Installation

Drex Dotz is distributed directly via GitHub (no PyPI required).

Single-Command Install
pip install "git+https://github.com/colt2822/Drex-Dotz.git@v0.1.0"
Two-Command Install (Explicit Firewall Dependency)

If your environment prefers explicit dependency pre-installation:

pip install "git+https://github.com/colt2822/Drex-Agent-Firewall.git@v0.1.3"
pip install "git+https://github.com/colt2822/Drex-Dotz.git@v0.1.0"

5-Minute Walkthrough

1. Initialize Runtime Environment

Seed the runtime with the three standard example workers:

dotz init

Output:

Initialized Drex Dotz runtime at ./dotz_runtime with 3 example Dotz.
2. List Registered Workers
dotz list

Output:

DOT                  STATE        UPDATED             
-------------------------------------------------------
developer            IDLE         1760000000.0        
operator             IDLE         1760000000.0        
researcher           IDLE         1760000000.0        
3. Inspect the Researcher Dot
dotz show researcher

Output:

Researcher status: IDLE

Workspace:
  ./dotz_runtime/dots/researcher/workspace

Model policy:
  default: cheap
  escalate: frontier

Capabilities:
  ✓ web.read
  ✓ files.read
  ✓ notes.write
  ✗ shell.execute
  ✗ secrets.read
  ✗ filesystem.modify
  ✗ money.spend
  ✗ filesystem outside workspace

Drex:
  enforcement active (pack: read-only-research)
4. Assign a Safe Research Task
dotz assign researcher "research latest security advisories and record notes" --exec

Output:

Assigned task 'research latest security advisories and record notes' to Dot 'researcher' -> Task ID: task_c8f2b1
Executing task...
Result: Completed task
5. Inspect Receipts

Inspect the immutable audit trail generated by the capability firewall:

dotz history researcher

Output:

ACTION          DECISION   CAPABILITY      TARGET                    RECEIPT_ID
--------------------------------------------------------------------------------
task_execute    ALLOW      research        notes                     rcpt_9f123a
6. Observe a Safe Drex BLOCK

When a Dot attempts an action outside its authorized capabilities—such as attempting shell execution or escaping its workspace—Drex immediately halts execution with zero upstream side effects:

from dotz.runtime import Runtime

rt = Runtime()
task_id = rt.assign_task("researcher", "Attempt shell execution")

# Researcher attempts shell command without 'shell.execute' capability
result = rt.execute_dot_action(
    dot_name="researcher",
    task_id=task_id,
    capability="shell.execute",
    action="shell_execute",
    args={"command": "cat /etc/passwd"},
)

print(result)
# {'allowed': False, 'error': 'Capability shell.execute is denied by Dot policy', 'receipt_id': 'rcpt_...'}

The forbidden command is blocked before invocation, and a BLOCK receipt is recorded in the audit store.


Public Examples Included

Drex Dotz ships with three bounded worker archetypes:

  1. Researcher (dotz/examples/researcher/): Web research, reading local documentation, and compiling notes. Denied shell execution and destructive filesystem modification.
  2. Developer (dotz/examples/developer/): Workspace code navigation and test execution. Consequential file edits require explicit, time-bounded mutation grants.
  3. Operator (dotz/examples/operator/): Read-only log inspection and status checks without generic root access.

Known Limitations

  • Process-Level Mediation: Enforcement operates at the Python runtime and guarded adapter layer; it does not replace kernel cgroups or virtual machine sandboxing.
  • Local Persistence: State is stored in local SQLite databases; cross-machine synchronization and cloud multi-tenancy are deferred to future milestones.

License

Apache-2.0. See LICENSE, NOTICE, and THIRD_PARTY_NOTICES.md.

Questions fréquentes

Qu'est-ce que ELYREON n8n Workflow Rescue ?

Static n8n workflow validator: workflow errors and repair suggestions. No execution.

Comment installer ELYREON n8n Workflow Rescue ?

Il fonctionne comme serveur distant. Dans Claude Code : claude mcp add --transport http elyreon-n8n-workflow-rescue 'https://elyraon.com.co/agents/mcp' --header 'Authorization: <Free token: POST https://elyraon.com.co/agents/trial with JSON {}. Enter Bearer TOKEN.>'. Pour les autres clients, collez la configuration JSON de cette page dans leurs réglages MCP.

Faut-il une clé API pour ELYREON n8n Workflow Rescue ?

Oui. Il attend Authorization, à renseigner dans sa configuration.

ELYREON n8n Workflow Rescue est-il open source ?

Oui. Son code est sur GitHub sous licence Apache-2.0.

Qui maintient ELYREON n8n Workflow Rescue ?

Il est publié par colt2822. AI.tn le référence sans le maintenir ni le recommander.

Issu du registre MCP officiel. Être listé n'est pas une recommandation : vérifiez le code et les permissions d'un serveur avant de lui donner accès à vos données.