gurdiga

claude-tool-profiling

A Claude Agent Skill that measures how long individual Claude Code tool calls take — Bash, Edit, MCP tools, everything.

It uses Claude Code's built-in OpenTelemetry instrumentation with the console exporter, so no Docker collector or observability backend is needed. Spans land in stdout, a small awk pipeline extracts tool.execution durations.

Why

When you're scripting lots of tiny tool calls, it helps to know the per-call floor. For example, the benchmark below shows that Bash "echo hello" has an overhead ~20× higher than an MCP stdio tool — which changes how you'd design a multi-step workflow.

gurdiga

Health Data

Apple Health exports and analysis scripts.

Contents

• apple_health_export/ — Apple Health export (XML + clinical records)
• sleep_cycles.py — Parses sleep data and prints nightly sleep cycles
• sleep_cycles_2.py / sleep_cycles_2.js — Same report, fed by an iOS Shortcut instead of the full export (see sleep_cycles_2.md)
• Makefile — Convenience targets

gurdiga

Pull Request Guidelines

When creating PRs:

1. Create as draft PR (gh pr create --draft)
2. Assign to the user (--assignee @me)
3. Add the "🌟 NEW FEATURE" label (--label "🌟 NEW FEATURE")
4. Do not hard-wrap long Markdown text lines in PR descriptions
5. Do not include Co-Authored-By: Claude line in PR descriptions (co-authorship is already mentioned in commits)

gurdiga

About this

This is a little hack to fix the issue reported by a lot of Blogspot bloggers here:

https://support.google.com/blogger/thread/94150705/blogger-quick-edit-pencil-icon-has-disappeared https://support.google.com/blogger/thread/94163591/edit-tool-on-blog

NOTE: A little downside is that the “Pencil” word under the posts is visible to everyone (not only the blog authors).

How to use

gurdiga

.ONESHELL:
MAKEFLAGS=

export CURL_PARAMS=--silent --connect-timeout 5 --retry-delay 0 --retry 5

start: list-albums # download-photos

list-albums:
	@function main() {
		list_albums "/albums/c111/SandraDodd"

gurdiga

tell application "Notes"
	repeat with eachNote in notes in folder "Log"
		tell eachNote
			set content to body as text
			set {year:y, month:m, day:d, hours:h, minutes:n} to creation date
			
			set m to m as number
			set d to d as number
			set n to n as number
			set h to h as number

gurdiga

function getCookieByName(name) {
  return unescape(
    (document.cookie.split(/; /g)||[])
      .filter(function(pair) {
        return pair.split('=')[0] === name;
      })[0]
      .split('=')[1]
  );
}

gurdiga

Per-directory Bash history (w/o aliasing cd)

I use Bash’s PROMPT_COMMAND variable:

The value of the variable PROMPT_COMMAND is examined just before Bash prints each primary prompt. If PROMPT_COMMAND is set and has a non-null value, then the value is executed just as if it had been typed on the command line.

The source code should be pretty straight forward, but if not, please ask in the comments. Put this in your .bashrc or similar:

# per-directory Bash history

gurdiga

JS interface composition with promises

Inspired by Mr. Robert C. Martin’s episodes on SOLID principles that I’ve watched lately, and by the idea of “programming to interfaces” I’ve tried to come up with a schema that would allow me to have the concerns separated, but still composable.

These are a few modules from an Angular project.

So, I have an AuthenticationService module that does user account house-keeping:

// app/authentication-service/authentication-service.js
(function() {

gurdiga

function randomPassword(lenght) {
  function random() {
    return parseFloat('.' + crypto.getRandomValues(new Uint32Array(1))[0]);
  }

  var characters = '0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ',
    password = '',
    i;

  for (i = 0; i < 10; i++)