DEV Community: Suneet Chugh

Building XENON: Bringing an Autonomous Voice & Chat SRE Teammate to Discord with Switch

Suneet Chugh — Sat, 07 Nov 2026 13:55:15 +0000

Building XENON: Bringing an Autonomous Voice & Chat SRE Teammate to Discord with Switch

Hackathon Submission: Built for the Flip the Switch Hackathon hosted by WeMakeDevs, powered by Switch AI.

1. The Bottleneck: Why AI Agents Are Trapped in Developer Terminals

Over the past year, autonomous AI coding agents (Claude Code, Cursor, OpenCode, Codex) have fundamentally upgraded our personal velocity. But in production engineering, software engineering is not a solo sport.

When a critical incident strikes, a deployment fails, or a PR is waiting on tests, your team doesn’t live in your isolated terminal window. Your team gathers in Discord, Slack, or Microsoft Teams.

Today, human engineers are forced to act as fragile, manual human bridges:

Copy-pasting error traces from Discord into an AI terminal.
Waiting for the agent to recommend a fix.
Copy-pasting the markdown or patch back into Discord.
Running the deployment manually.

We asked: What if the AI agent sat directly inside our Discord server as a full engineering peer—listening in voice channels, triaging incidents, and running privileged rollbacks upon voice authorization?

Meet XENON.

2. Introducing XENON

XENON is an open-source autonomous developer teammate powered by Switch.

Instead of living trapped in one engineer's local shell, XENON:

Collaborates in Discord channels and voice calls: Tag @XENON in chat or summon him into a Discord voice channel.
Processes Real-Time Voice: Ingests voice instructions using Whisper Speech-to-Text and replies aloud via low-latency neural TTS.
Orchestrates Real Developer Work: Connects directly to GitHub REST/GraphQL APIs and CI/CD webhooks via structured tool execution.
Enforces Strict RBAC: Ensures only authorized repository maintainers and guild administrators can approve critical deployments or rollbacks.

3. High-Level Architecture

Here is how the end-to-end data flow operates:

                  ┌────────────────────────────────────────┐
                  │          Team Discord Server           │
                  │   [ #war-room Chat ]   [ Voice Call ]  │
                  └───────────┬────────────────────┬───────┘
                              │                    │
                              ▼                    ▼
                    [ Discord Gateway ]    [ Audio Stream ]
                              │                    │
                              ▼                    ▼
                     [ Switch Framework ]   [ Whisper STT ]
                    (Session Persistence)          │
                              │                    │
                              └─────────┬──────────┘
                                        │
                                        ▼
                            [ RBAC Security Guard ]
                            (Owner / Role Verification)
                                        │
                                        ▼
                            [ Autonomous Agent Loop ]
                             (MCP Developer Tools)
                                        │
                      ┌─────────────────┴─────────────────┐
                      ▼                                   ▼
              [ GitHub & CI/CD ]                 [ Neural Edge-TTS ]
             (Direct API Actions)               (Discord Voice Out)

4. Key Engineering Challenges & Solutions

A. The Bot-to-Bot Communication Trap

A common architectural antipattern is attempting to have an AI Discord bot trigger other developer bots (like deployment shippers) by typing commands into text channels.

The reality: The official Discord API standard explicitly requires bots to ignore other bots (if (message.author.bot) return;) to prevent infinite cascading loops.

The Fix: XENON bypasses bot-chat intermediaries entirely. Armed with Model Context Protocol (MCP) tool schemas, XENON interfaces directly with GitHub APIs, Docker daemons, and Vercel/CI webhooks under cryptographically verified credentials.

B. High-Fidelity Audio Without Crashing Your Budget

Relying on expensive, metered third-party TTS APIs during long hackathon debugging sessions quickly exhausts quotas. Furthermore, Discord’s built-in /tts command only reads text aloud on the local user’s operating system—it does not broadcast into voice channels.

The Fix: We built a dedicated voice pipeline combining:

Whisper STT for low-latency voice command parsing from Discord voice streams.
Microsoft edge-tts for zero-cost, ultra-natural neural audio generation that streams directly into the active @discordjs/voice connection.

C. The Switch Backbone

Switch (sandbox-quantum/switch) handles the heavy lifting of:

Decoupling Discord's messy gateway event structures from the underlying agent loop.
Providing seamless state and session persistence across multiple channel threads.
Allowing our entire team to inspect XENON's reasoning steps in real-time.

5. Code Spotlight: The RBAC Security Guard

When an AI agent has the authority to run tests, inspect code, and trigger rollbacks, security cannot be an afterthought. Here is how XENON guards privileged execution paths:

typescript

// src/security/rbac.ts
import { GuildMember, PermissionsBitField } from "discord.js";

export interface SecurityContext {
  userId: string;
  isOwner: boolean;
  hasDeployPrivilege: boolean;
}

export function verifyExecutionPrivilege(
  member: GuildMember,
  allowedRoleIds: string[]
): SecurityContext {
  const isOwner = member.id === member.guild.ownerId;
  const isAdmin = member.permissions.has(PermissionsBitField.Flags.Administrator);
  const hasRole = member.roles.cache.some((role) => allowedRoleIds.includes(role.id));

  const hasDeployPrivilege = isOwner || isAdmin || hasRole;

  if (!hasDeployPrivilege) {
    console.warn(`[SECURITY] Blocked unauthorized command attempt by ${member.user.tag} (${member.id})`);
  }

  return {
    userId: member.id,
    isOwner,
    hasDeployPrivilege,
  };
}

6. What’s Next for XENON

Building with Switch during this hackathon proved that AI agents are ready to move out of solitary confinement and into team chat.

Our roadmap includes:

Multi-Platform Support: Expanding Switch routing to mirror incidents simultaneously across Discord and Slack.
Proactive Log Streaming: Having XENON detect Sentry anomalies and automatically open a Discord war-room thread with reproduction steps before an engineer is paged.

Try it out & Contribute!

GitHub Repo: (https://github.com/AWDAX/XENON_AI-Autonomous-Agent)
Powered by Switch: github.com/sandbox-quantum/switch
Hackathon: WeMakeDevs Flip the Switch

If you found this interesting, drop a reaction or let us know in the comments how your team is connecting AI agents to chat platforms!

Why Vector RAG Fails Coding Agents: How We Built CAVDAX with FalkorDB & Model Context Protocol

Suneet Chugh — Sun, 18 Oct 2026 18:25:10 +0000

Why Vector RAG Fails Coding Agents: How We Built CAVDAX with FalkorDB & Model Context Protocol

Author: Suneet Chugh

Track: Track 01: Agents That Act on Connected Data | Track 02: Knowledge Graphs & Long-Term Memory

Project: CAVDAX (Contextual Action & Verified Decision Autonomous eXecution)

Hackathon: WeMakeDevs × FalkorDB Graph Hacks 2026

Repository: github.com/suneetchugh-dev/CAVDAX

Live Demo: cavdax.sahnirmaan.live

💥 The Incident: When Vector RAG Blew Up Our Checkout

Last week, an autonomous coding agent was tasked with updating an e-commerce checkout service. It queried the repository’s vector knowledge base, retrieved an Architecture Decision Record titled ADR-001: Adopt TokenAuth v1, and generated 40 lines of code wiring up a deprecated token authentication protocol.

The code was syntactically flawless. All unit tests passed.

There was just one catastrophic problem: ADR-001 had been superseded two weeks earlier by ADR-002: Enforce OAuth2 Security.

Why did the agent retrieve the dead architecture decision? Because to a vector database, ADR-001 and ADR-002 share virtually identical embeddings: "checkout", "authentication", "token", "payment", "security". Cosine similarity is completely blind to time, directionality, and causality.

This is the dirty secret of modern agentic workflows: Vector RAG suffers from fatal "Silent Cache Rot."

In this article, I want to break down why vector search alone cannot govern autonomous agents, and how my sister Geetika and I built CAVDAX—a deterministic, graph-grounded execution gate powered by FalkorDB and the Model Context Protocol (MCP) that validates code changes against live causal graphs in under 2 milliseconds.

⚡ The Fundamental Flaw: Similarity vs. Causality

When an engineer works in a codebase, their mental model is not an unstructured cloud of floating semantic vectors. It is a Causal Directed Acyclic Graph (DAG):

When an engineer supersedes an ADR or commits a breaking change:

They know all historical claims extracted from the old ADR are instantly invalidated.
Any architectural decision depending on those claims must be reviewed before code is written.

Vector embeddings cannot represent this:

No Temporal Invalidation: Cosine similarity has no concept of SUPERSEDES.
No Directional Transitivity: If A → B and B → C, vector similarity cannot compute multi-hop ancestry.
No Pre-Execution Safety: File-system tools run unrestricted without verifying provenance.

🛡️ Enter CAVDAX: Enclosed Action via FalkorDB

We designed CAVDAX around a radical premise:

An autonomous coding agent should never have direct, unrestricted access to the file system. Every file modification tool must be enclosed inside a pre-execution Cypher gate that traverses live causal provenance before writing a single byte.

🔬 Why FalkorDB Was Irreplaceable

When designing this gate, we evaluated three database paradigms:

Relational Databases (PostgreSQL): Running recursive multi-hop traversals (WITH RECURSIVE) across arbitrary DAG depths caused recursive JOIN latency spikes (> 45ms), completely destroying agent responsiveness.
Traditional Graph Databases (Neo4j / Amazon Neptune): Round-trip HTTP/Bolt query latencies hovered between 25ms and 80ms. For a synchronous tool call that an agent invokes repeatedly, this overhead was unacceptable.
FalkorDB (GraphBLAS Engine): FalkorDB stores graphs as sparse adjacency matrices and computes multi-hop path traversals using linear algebraic matrix multiplication over the Redis binary protocol.

Traversing 4 hops across thousands of nodes in FalkorDB took 1.84ms.

FalkorDB gave us the exact speed profile required to place a formal graph database directly inside the synchronous execution loop of an AI tool call.

💻 The 3 Core Cypher Queries Behind the Gate

1. Bounded Multi-Hop Verification (`verifyActionGate`)

When the agent invokes apply_code_change(decision_id: "DEC-102"), the gate intercepts the call and runs this Cypher query:


cypher
MATCH (d:Decision {id: $decisionId})
WITH d, d.status AS decision_status
OPTIONAL MATCH path = (d)-[:DEPENDS_ON*1..4]->(f:Fact)-[:EXTRACTED_FROM]->(s:Source)
WITH d, decision_status,
     collect({
       path_nodes: [n IN nodes(path) | labels(n)[0] + ':' + coalesce(n.id, n.ref)],
       fact_stale: coalesce(f.is_stale, false),
       source_status: coalesce(s.status, 'ACTIVE')
     }) AS dependencies
RETURN d.id AS decision_id,
       decision_status,
       CASE 
         WHEN decision_status = 'INVALIDATED_NEEDS_REVALIDATION' THEN 'BLOCKED_DECISION_INVALIDATED'
         WHEN size(dependencies) = 0 THEN 'BLOCKED_NO_GROUNDING'
         WHEN ANY(dep IN dependencies WHERE dep.fact_stale = true) THEN 'BLOCKED_STALE_FACT'
         WHEN ANY(dep IN dependencies WHERE dep.source_status = 'SUPERSEDED') THEN 'BLOCKED_SUPERSEDED_SOURCE'
         ELSE 'APPROVED'
       END AS gate_verdict,
       [dep IN dependencies WHERE dep.fact_stale = true OR dep.source_status = 'SUPERSEDED' | dep.path_nodes] AS violation_paths


**2. The Cascading Invalidation Wave (cascadeInvalidation)**
MATCH (newSource:Source {id: $newSourceId})
MATCH (oldSource:Source {id: $oldSourceId})
MERGE (newSource)-[:SUPERSEDES]->(oldSource)
SET oldSource.status = 'SUPERSEDED',
    oldSource.superseded_at = timestamp()
WITH oldSource
MATCH (oldSource)<-[:EXTRACTED_FROM]-(staleFact:Fact)
SET staleFact.is_stale = true, 
    staleFact.invalidated_at = timestamp()
WITH staleFact
MATCH (d:Decision)-[:DEPENDS_ON*1..4]->(staleFact)
SET d.status = 'INVALIDATED_NEEDS_REVALIDATION',
    d.invalidated_at = timestamp()
RETURN count(DISTINCT d) AS affected_decisions, 
       collect(DISTINCT d.id) AS invalidated_decision_ids



**3. Act IV: The Autonomous Revalidation Loop (revalidateDecision)**


MATCH (newSource:Source {id: $newSourceId})<-[:EXTRACTED_FROM]-(newFact:Fact {id: $newFactId})
MATCH (oldDecision:Decision {id: $oldDecisionId})
CREATE (newDecision:Decision {
  id: $newDecisionId,
  title: $title,
  status: 'VALID',
  created_at: timestamp()
})
CREATE (newDecision)-[:DEPENDS_ON]->(newFact)
CREATE (newDecision)-[:REPLACES]->(oldDecision)
SET oldDecision.status = 'REPLACED_ARCHIVED'
RETURN newDecision.id AS active_decision_id

DEV Community: Suneet Chugh

Building XENON: Bringing an Autonomous Voice & Chat SRE Teammate to Discord with Switch

Building XENON: Bringing an Autonomous Voice & Chat SRE Teammate to Discord with Switch

1. The Bottleneck: Why AI Agents Are Trapped in Developer Terminals

2. Introducing XENON

3. High-Level Architecture

4. Key Engineering Challenges & Solutions

A. The Bot-to-Bot Communication Trap

B. High-Fidelity Audio Without Crashing Your Budget

C. The Switch Backbone

5. Code Spotlight: The RBAC Security Guard

6. What’s Next for XENON

Try it out & Contribute!

Why Vector RAG Fails Coding Agents: How We Built CAVDAX with FalkorDB & Model Context Protocol

Why Vector RAG Fails Coding Agents: How We Built CAVDAX with FalkorDB & Model Context Protocol

💥 The Incident: When Vector RAG Blew Up Our Checkout

⚡ The Fundamental Flaw: Similarity vs. Causality

🛡️ Enter CAVDAX: Enclosed Action via FalkorDB

🔬 Why FalkorDB Was Irreplaceable

💻 The 3 Core Cypher Queries Behind the Gate

1. Bounded Multi-Hop Verification (verifyActionGate)

1. Bounded Multi-Hop Verification (`verifyActionGate`)