DEV Community: meddlesome

AI Agent vs Discord Webhook Message limitation

meddlesome — Tue, 01 Jul 2025 04:03:19 +0000

When you develop AI Agent, especially with n8n. You want to send message to Discord with Webhook. But it end up with error when LLM produced output more than 2000 characters as a limitation of Discord Webhook.

LLM Limitation

Solving by let LLM trying to split the message is not accurate, since 1 Token != 1 characters. then It will end-up with Multi-turn request which not efficient as possible with many attempt which wasted time and context windows(even with Context windows caching).

Tackle the problem with Tools calling

Since LLM is not accurate on counting, e.g. strawberry problem. Using Tools calling. It can help.

Simple Tools to count string length

Then I start with a simple tools as one liner to return string length to let Agent to split text and try again

return query.length

But this is not sufficient enough as

Still Multi-turn effort as time consume. Even LLM know the length, but LLM cannot split perfectly
Wasted Token (even with LLM context caching)

Efficient ways

Then I looking forward to create a single tools to let LLM calling tools split string chunks to array and let n8n workflow continue the rest.

I also create a reusable with Executed by another Workflow Trigger to perform sending Discord Webhook

Here is a JS Snipplet https://gist.github.com/meddlesome/b1e35ad133c726e0691b89c307245af5

// Input: Single string from previous node
const inputText = $input.first().json.text; // Adjust field name as needed

// Configuration
const MAX_LENGTH = 2000;
const SPLIT_PATTERNS = ['\n\n', '\n', '. ', ' ']; // Priority order for splitting

function splitText(text, maxLength) {
  if (text.length <= maxLength) {
    return [text];
  }

  const chunks = [];
  let remainingText = text;

  while (remainingText.length > 0) {
    if (remainingText.length <= maxLength) {
      chunks.push(remainingText);
      break;
    }

    let chunk = remainingText.substring(0, maxLength);
    let splitIndex = -1;

    // Find best split point
    for (const pattern of SPLIT_PATTERNS) {
      const lastIndex = chunk.lastIndexOf(pattern);
      if (lastIndex > maxLength * 0.5) { // Don't split too early
        splitIndex = lastIndex + pattern.length;
        break;
      }
    }

    // If no good split point found, force split at max length
    if (splitIndex === -1) {
      splitIndex = maxLength;
    }

    chunks.push(remainingText.substring(0, splitIndex).trim());
    remainingText = remainingText.substring(splitIndex).trim();
  }

  return chunks;
}

// Process the text
const textChunks = splitText(inputText, MAX_LENGTH);

// Return array of chunks
return textChunks.map((chunk, index) => ({
  json: {
    chunk: chunk,
    index: index,
    totalChunks: textChunks.length,
    length: chunk.length
  }
}));

It end-up with single turn and reducing time over 50% !

How to record screen with system sound (Including Microsoft Teams meeting) by QuickTime

meddlesome — Tue, 03 Dec 2024 14:35:22 +0000

Prerequisite

Installed Blackhole 2ch

How to Setup

Open "Audio MIDI Setup" app from Application/Utilities. If the Blackhole was installed. It should show BlackHole 2ch on the list
Create "Aggregate Device" > Select BlackHole 2ch > Rename the new profile to Input for QuickTime Record

Create "Multi-Output Device" > Select BlackHole 2ch AND your YOUR_CURRENT_SPEAKER_OR_HEADPHONE > Rename the new profile to Output for Screen Record

ℹ️ Below are also the 📝 checklist before recording, always check.

When to record

Open “Audio MIDI Setup” to check “Output for Screen Record” that both BlackHole 2ch AND YOUR_CURRENT_SPEAKER_OR_HEADPHONE were enabled
Open “Audio MIDI Setup” to check “Input for QuickTime Record” that BlackHole 2ch was enabled. If you want your 🎤 microphone sound recorded, you must also enable YOUR_MICROPHONE.

🚨 Warning !!! when to enable your 🎤 microphone on this profile. It will continue recording unless you have a hardware muted button.

Screen Record with QuickTime

File > New Screen Recording. Then ensure the audio input was correctly set with Option… > Microphone > Input for QuickTime Record

ℹ️ This also works with OBS or other streaming applications.

Record from Microsoft Teams

Choose 🔈 Speaker = Output for Screen Record(Aggregate)
Select 🎤 Microphone = YOUR_CURRENT_MICROPHONE

Record from other applications e.g. Chrome

Choose Audio Output to Output for Screen Record

🔊 While choosing Output for Screen Record, you cannot adjust the output volume for your speaker or headphones. To adjust the output volume, switch to your output to adjust the volume. Then switch back to Output for Screen Record again.

Some GKE pods are failing with strange error with Google Auth

meddlesome — Thu, 01 Jun 2023 12:40:26 +0000

Recently, I found a strange Pod behavior in Google Kubernetes Engine - GKE

When Rolling-out deployment. Most pods are healthy and up, but some pod error with CrashLoopBack.

Digging down to Pod Event, everything good. Except unhealthy. Going more to The log of unhealthy pod. Some hint is showing

Error: Could not load the default credentials. Browse to https://cloud.google.com/docs/authentication/getting-started for more information.
    at GoogleAuth.getApplicationDefaultAsync (/usr/src/app/node_modules/google-auth-library/build/src/auth/googleauth.js:183:19)

We found some interesting fix by adding Environment variable to pod.

DETECT_GCP_RETRIES=3

🎉🎉🎉 Now pods are fully up without error.

Cause of this issue is when rolling deployments, multiple pods made request to GCP for authentication, sometime it timeout without retry then fail.

Solving kubectl authentication error on GKE

meddlesome — Fri, 27 Jan 2023 09:11:32 +0000

When you executing kubectl command to your cluster on GKE. It will thrown some error like

$ kubectl get namespaces
error: The gcp auth plugin has been removed.
Please use the "gke-gcloud-auth-plugin" kubectl/client-go credential plugin instead.
See https://cloud.google.com/blog/products/containers-kubernetes/kubectl-auth-changes-in-gke for further details

As GKE change the way to authentication more safely. then we need to upgrade our GKE authentication

But Even installing gcloud components install gke-gcloud-auth-plugin and set export USE_GKE_GCLOUD_AUTH_PLUGIN=True might be still not working. Since an existing "Kube context" (credentials) still exist. You need to remove and fetch kube context again.

Here is overall trick to solve the problem

$ gcloud components install gke-gcloud-auth-plugin
$ echo 'export USE_GKE_GCLOUD_AUTH_PLUGIN=True' >> ~/.bashrc
$ source ~/.bashrc
$ gcloud components update
$ cp ~/.kube/config ~/.kube/config.backup && rm ~/.kube/config
$ gcloud container clusters get-credentials {{GKE_CLUSTER_NAME}} --region {{GKE-REGION}} --project {{GCP_PROJECT_NAME}}

How to access Google Cloud Storage Bucket with Service account JSON

meddlesome — Tue, 20 Dec 2022 16:27:44 +0000

This method for gsutil to access Google Cloud Storage bucket with shared Service account with private key as JSON

Sample JSON Structure

{
  "type": "service_account",
  "project_id": "xxx",
  "private_key_id": "xxx",
  "private_key": "-----BEGIN PRIVATE KEY-----\name-----END PRIVATE KEY-----\n",
  "client_email": "xxx@xxx.iam.gserviceaccount.com",
  "client_id": "xxx",
  "auth_uri": "https://accounts.google.com/o/oauth2/auth",
  "token_uri": "https://oauth2.googleapis.com/token",
  "auth_provider_x509_cert_url": "https://www.googleapis.com/oauth2/v1/certs",
  "client_x509_cert_url": "https://www.googleapis.com/robot/v1/metadata/x509/xxx.iam.gserviceaccount.com"
}

Using gsutil config -e to set method to authentication
Choose oAuth2 and provided full path of service account JSON file
Then you can access with gsutil ls gs://{bucket-name}

Safe to interact with Kubernetes CLI

meddlesome — Wed, 20 Apr 2022 16:28:10 +0000

When you interacting with Kubernetes CLI like kubectl or any related commands. It normally interact with current context of kubeconfig file with Project, Cluster, Namespace.

It will mess you up when interacting with multiple projects, clusters, namespaces. Since you cannot sure what is context that you will pull a trigger. And quite error-prone when interact with that.

Current Context Prompt

Ensure on your current context of kubectl by showing current context prompt on your screen when typing related Kubernetes commands.

Prerequisites

When you using zsh with romkatv/powerlevel10k theme. It feature "Show in Command" which support kubecontext which can show current context at right hand side of each prompt and always show when you run related Kubernetes commands.

Here is a sample guide to customize on romkatv/powerlevel10k configuration file ~/.p10k.zsh which you can open on any text editor, then edit and reload your shell session.

Show on Command

This will specific a list of command that will trigger to show current context while you typing-in, feel free to adding more if you have more related Kubernetes tools.

typeset -g POWERLEVEL9K_KUBECONTEXT_SHOW_ON_COMMAND='kubectl|helm|kubens|kubectx|oc|istioctl|kogito|k9s|helmfile|fluxctl|stern|ytt'

Customize context detail

You can specific on what Kubernetes context information to show on the prompt when you typing-in any related Kubernetes commands

Many options are available and already commented on ~./p10k.zsh file, but here is my recommended pattern that quite clear when you working with multiple contexts.

CLOUD_PROVIDER/PROJECT_NAME/CLUSTER_NAME/NAMESPACE
Same as above sample images

typeset -g POWERLEVEL9K_KUBECONTEXT_DEFAULT_CONTENT_EXPANSION='${${P9K_KUBECONTEXT_CLOUD_NAME:+/$P9K_KUBECONTEXT_CLOUD_NAME/$P9K_KUBECONTEXT_CLOUD_ACCOUNT/$P9K_KUBECONTEXT_CLOUD_CLUSTER}:-$P9K_KUBECONTEXT_NAME}${${:-/$P9K_KUBECONTEXT_NAMESPACE}:#/default'

Parsing JSON on Jenkins , a problem with sh

meddlesome — Mon, 07 Feb 2022 07:07:10 +0000

When you want to process a JSON file in Jenkins pipeline, you can do in various ways

Pipeline Utility Steps readJSON
jq
JSON Parser in Python, NodeJS. etc

This article will focus on the problem with jq vs Jenkins sh since it introducing some interesting problem.

Sample Data

file.json

{
    "lorem": {
        "key": "Lorem Ipsum",
        "dash-key": "Lorem Ipsum",
        "underscore_key": "Lorem Ipsum",
    }
}

In the jq, a syntax to get the JSON value from a key is

$ cat file.json | jq -r .lorem.key
Lorem Ipsum
$ cat file.json | jq -r .lorem.underscore_key
Lorem Ipsum

But, it will have a problem when your JSON key contains "-" dash.

$ cat file.json | jq -r .lorem.dash-key
jq: error: key/0 is not defined at <top-level>, line 1:
.lorem.dash-key
jq: 1 compile error

With some limitation with jq, a key string contains dash - have to combine pattern with single-quote and double-quote when query. Possible in many pattern to get a value.

Dash Issue with jq Github Issue, Stack Overflow

$cat file.json | jq -r '.lorem."dash-key"'
Lorem Ipsum

$cat file.json | jq -r .lorem.'"dash-key"'
Lorem Ipsum

$cat file.json | jq -r '.lorem["dash-key"]'
Lorem Ipsum

$cat file.json | jq -r .lorem.\"dash-key\"
Lorem Ipsum

When we bring the jq query into Jenkins with a key name as a variable. It will lead to next problem, as Jenkins have some weird way to manipulating string with single-quote and double-quote in sh function.

Fun read Adventures with escaping quotes in Jenkins and Gist of weird escape on Jenkins

Then, we need to escape double-quote with Jenkins sh and jq with this pattern \\\"

SOME_KEY = "dash-key"
value = sh(returnStdout: true, script: "cat file.json | jq -r .lorem.\\\"${SOME_KEY}\\\"").trim()

NextJS APIs validator with Middleware

meddlesome — Tue, 26 May 2020 11:46:05 +0000

Validate the request.body is a must for every APIs development. NextJS is able to use Connect compatible Middlewares for an extendable feature on top of each request/response like ExpressJS.

Here is a guide to integrating the express-validator, a wrapper of validator.js inside your NextJS application.

First, Install express-validator in NextJS project

yarn add express-validator

Next, Create NextJS Middleware Helper Method as /lib/init-middleware.js

export default function initMiddleware(middleware) {
    return (req, res) =>
      new Promise((resolve, reject) => {
        middleware(req, res, (result) => {
          if (result instanceof Error) {
            return reject(result)
          }
          return resolve(result)
        })
      })
  }

Next, Create NextJS Validator Middleware to handle error and response out as /lib/validate-middleware.js

export default function validateMiddleware(validations, validationResult) {
  return async (req, res, next) => {
    await Promise.all(validations.map((validation) => validation.run(req)))

    const errors = validationResult(req)
    if (errors.isEmpty()) {
      return next()
    }

    res.status(422).json({ errors: errors.array() })
  }
}

Now it's time to integrate validate rules and middleware into your NextJS API Routes. You can use any validator.js's validator functions as sample below.

import initMiddleware from '../../../lib/init-middleware'
import validateMiddleware from '../../../lib/validate-middleware'
import { check, validationResult } from 'express-validator'

const validateBody = initMiddleware(
    validateMiddleware([
        check('first_name').isLength({min:1, max: 40}),
        check('day').isInt({ min: 1, max: 31}),
        check('gender').isIn(['male','female']),
        check('mobile_phone').isMobilePhone(['th-TH']),
        check('boolean').isBoolean(),
    ], validationResult)
)

export default async (req, res) => {
  switch (req.method) {
    case "POST":
      await validateBody(req, res)

      const errors = validationResult(req)
      if (!errors.isEmpty()) {
        return res.status(422).json({ errors: errors.array() })
      }

      nextFunction(req, res)

      break;
    default:
      res.status(404).json({ message: "Request HTTP Method Incorrect." })
      break;
  }
}

Now the express-validator will be working on top of each HTTP Request to validate your request.body in NextJS API Route :)