Marek Šuppa

Displaying account-specific Claude Code rate limits with ccstatusline

Mon, 13 Apr 2026 00:00:00 +0000

ccstatusline is a status bar for Claude Code – it makes it easy to see model info, git branch, token count, cost, context window, and all sorts of other things right from the Claude Code CLI. Since v2.1.0 (March 2026) it can also show rate limit utilization via the Anthropic OAuth API:

SessionUsage – how much of the current 5-hour block you’ve already burned through
WeeklyUsage – similar concept for the 7-day rolling window
BlockResetTimer / WeeklyResetTimer – countdown until the session / weekly reset

Cooldown periods for package updates

Tue, 31 Mar 2026 00:00:00 +0000

The last two weeks of March 2026 were brutal for software supply chains. A cascading campaign by a threat actor tracked as TeamPCP started with a compromised credential in Trivy (Aqua Security’s vulnerability scanner), which was then used to hijack Checkmarx KICS and then LiteLLM (an LLM proxy with ~97M monthly PyPI downloads). On top of that, axios – the most popular HTTP client on npm with ~100M weekly downloads – was compromised through a stolen maintainer token, with Google’s Threat Intelligence Group attributing the attack to North Korea’s UNC1069.

Storing API keys for Claude Code providers in KeePassXC

Wed, 25 Mar 2026 00:00:00 +0000

A growing number of providers expose Anthropic-compatible API endpoints for Claude Code:

They all work the same way: set ANTHROPIC_BASE_URL and ANTHROPIC_AUTH_TOKEN, and Claude Code talks to them instead of Anthropic. Their docs typically suggest putting the key straight into ~/.claude/settings.json:

json

{
 "env": {
 "ANTHROPIC_BASE_URL": "https://api.minimax.io/anthropic",
 "ANTHROPIC_AUTH_TOKEN": "<MINIMAX_API_KEY>"
 }
}

That, unfortunately, involves saving a plaintext secret on disk. Plaintext secrets anywhere near Claude Code are risky – it has been known to read .env files and leak their contents into session transcripts and tool-result logs under ~/.claude/, and settings.json itself is no safer. A popular workaround is a ~/.secrets file sourced from your shell profile, but that still means the key is stored in plaintext, just somewhere else.

Claude Code Cheat Sheet

Wed, 25 Mar 2026 00:00:00 +0000

A nicely organized cheat sheet of Claude Code commands, many of which are not widely known (e.g. /insights, /review). Automatically updated daily. See also the Hacker News discussion.

Using --wraps to get tab completions on fish shell aliases

Tue, 24 Mar 2026 00:00:00 +0000

When you write a wrapper function in fish, you can use the --wraps flag to inherit tab completions from the wrapped command:

fish

function claude-minimax \
 --wraps='claude' \
 --description 'Run Claude Code against MiniMax M2.7 API'
 ANTHROPIC_BASE_URL="https://api.minimax.io/anthropic" \
 ANTHROPIC_MODEL="MiniMax-M2.7" \
 claude $argv
end

The --wraps='claude' tells fish that claude-minimax is a wrapper around claude. This means that when you type claude-minimax -- and hit <Tab>, you get the same completion suggestions as claude --. Without it, fish has no way of knowing the relationship and you would get no completions at all.

Uploading images to GitHub PRs from the CLI

Sun, 22 Mar 2026 00:00:00 +0000

GitHub has no public API for uploading images to PR descriptions. The drag-and-drop in the web UI uses an undocumented endpoint (/upload/policies/assets) that requires browser session cookies – you simply can not call it with an API token or gh.

The GitHub CLI team explicitly closed this as “not planned”, with the explanation that the upload flow requires a browser to obtain temporary S3 credentials. Pretty understandable but still tough to live with, particularly in this agent era we live in.

How to get the deployed Git commit hash in Portainer CE

Thu, 19 Mar 2026 00:00:00 +0000

When Portainer Community Edition deploys a stack from a git repository, it clones the repo, runs docker compose up, and then deletes the .git directory. This means the usual git rev-parse HEAD trick doesn’t work inside the running container – there’s no git history left.

But Portainer does remember the commit it deployed. The stack API endpoint exposes it in GitConfig.ConfigHash:

bash

curl -s -H "X-API-Key: $PORTAINER_API_KEY" \
 "$PORTAINER_URL/api/stacks/$PORTAINER_STACK_ID" \
 | python -m json.tool | grep ConfigHash

json

"ConfigHash": "5d346dc4e8b34235c11950e123358ebd85cfa3e8"

This is the exact commit SHA that Portainer cloned when it last (re)deployed the stack (here’s how the update mechanism works). You can query it at container startup to bake the version into your app. Here’s a minimal snippet that writes it to a file (using only Python stdlib, no extra dependencies needed in a slim image):

Google Docs can be exported to Markdown via URL change

Mon, 16 Feb 2026 00:00:00 +0000

Google Docs now supports exporting directly to Markdown by tweaking the URL. Take any Google Doc URL like:

https://docs.google.com/document/d/1mt8aYM88Jj5qkep1xYC5vj0wBlbX2u6gdxhf_puaiQI/edit?tab=t.0

Replace everything after the document ID with export?format=md:

https://docs.google.com/document/d/1mt8aYM88Jj5qkep1xYC5vj0wBlbX2u6gdxhf_puaiQI/export?format=md

If you use curl to test it out, you might get just a 307 redirect:

bash

curl -I "https://docs.google.com/document/d/1mt8aYM88Jj5qkep1xYC5vj0wBlbX2u6gdxhf_puaiQI/export?format=md"
# HTTP/2 307
# location: https://doc-0s-...googleusercontent.com/...

Pass -L to follow it and get the actual Markdown:

Hugo can merge multiple sections into one stream

Wed, 11 Feb 2026 00:00:00 +0000

When building a Hugo site with multiple content types (blog, TIL, links), you can merge them into a single stream on the homepage using union:

go-html-template

{{ $blog := where site.RegularPages "Section" "blog" }}
{{ $links := where site.RegularPages "Section" "links" }}
{{ $all := union $blog $links }}
{{ range sort $all "Date" "desc" }}
 ...
{{ end }}

This is how sites like simonwillison.net create a unified content stream from diverse content types.

Write the Paper First

Tue, 10 Feb 2026 00:00:00 +0000

Jason Eisner’s timeless advice on academic writing. The core idea is simple but very impactful: start writing the paper before you have all the results. Writing forces you to clarify your thinking, reveals gaps in your argument, and often redirects the research itself (e.g. what other experiments to do in order to arrive at a coherent story).

skLEP: A comprehensive Slovak NLU benchmark

Tue, 01 Jul 2025 00:00:00 +0000

First comprehensive benchmark for evaluating Slovak natural language understanding. Presented at ACL 2025.

Five things I wish I knew the day I started University

Wed, 10 Oct 2018 00:00:00 +0000

Things I wish someone told me the day I become a university student (for the first time).

How to get the most out of RoboCup(Junior)

Thu, 20 Jul 2017 00:00:00 +0000

The international RoboCup(Junior) is a unique event. This post summarizes what I would want to know before I participated for the first time.

How I built an app for FirefoxOS

Mon, 23 Sep 2013 00:00:00 +0000

… and nearly funded a company.

Lavabit

Fri, 09 Aug 2013 00:00:00 +0000

A few notes on a privacy focused email service and its shutdown.

Notes from Nederlands: take two

Wed, 10 Jul 2013 00:00:00 +0000

Been there, seen that. Nederlands for the first time.

Notes from Belgium

Sat, 06 Jul 2013 00:00:00 +0000

Not being somewhere for the first time also has its bright sides. Especially for Belgium.

Notes from Nederlands

Sat, 22 Jun 2013 00:00:00 +0000

Been there, seen that. Nederlands for the first time.

A tale of a physicist and a craftsman

Thu, 30 May 2013 00:00:00 +0000

Time is the only limit when a physicist meets a craftsman.

Writing a blog

Mon, 01 Apr 2013 00:00:00 +0000

Those who blog often say that others should start too. Unfortunately, they don’t say when.