Skip to content

Instantly share code, notes, and snippets.

View sshh12's full-sized avatar
πŸ‘¨β€πŸ’»
Where there is code, there is code.

Shrivu Shankar sshh12

πŸ‘¨β€πŸ’»
Where there is code, there is code.
View GitHub Profile
@sshh12
sshh12 / README.md
Created October 2, 2026 21:38
asdfmovieish: the prompt, workflow architecture (~370 Claude subagents), stats and repo layout for making a full asdfmovie-style episode in code + ElevenLabs

asdfmovieish: a full asdfmovie-style episode, drawn in code and voiced by ElevenLabs, built by ~370 Claude subagents

One prompt, one Claude Code session (about 19 hours of wall-clock, including two usage-limit stalls), and a pile of Workflow scripts produced a 2:29, 1080p, 25 fps original episode in the style of TomSka's asdfmovie: 25 scripted skits, a title card, credits and an end card; every frame drawn with Python + Skia, every voice, sound effect and music cue generated with ElevenLabs. This gist has the prompt, how the workflows were used, the stats, the repo layout and how to reproduce it. The six workflow scripts that actually ran are attached as wf*.js.

It is a fan tribute and is not affiliated with TomSka. No official footage or audio is in the output; the research corpus (downloaded official episodes) is for analysis only and is not part of this gist.

@sshh12
sshh12 / observatory-models.html
Last active September 7, 2026 21:41
Astra demo: rental research to a photo-aligned 3D rooftop observatory concept
<!doctype html>
<html lang="en"><head><meta charset="utf-8"><meta name="viewport" content="width=device-width,initial-scale=1"><title>Astra β€” From Rental Research to a Rooftop Observatory</title>
<style>
:root{color-scheme:dark;--bg:#0a111b;--panel:#101e2c;--line:#294051;--muted:#91a6b9;--text:#e5eef5;--cyan:#8cd8df;--gold:#efbe6c;--pink:#df8ba7}*{box-sizing:border-box}body{margin:0;background:var(--bg);color:var(--text);font:14px/1.5 system-ui,-apple-system,sans-serif}button,input,select{font:inherit}button,select{border:1px solid var(--line);border-radius:8px;padding:9px 12px;background:#152839;color:var(--text)}button{cursor:pointer}button:hover{border-color:var(--cyan)}button.active{background:#214653;border-color:var(--cyan)}a{color:var(--cyan)}input[type=range]{accent-color:var(--cyan);width:100%}input[type=number],input[type=date]{width:100%;background:#0c1824;color:var(--text);border:1px solid var(--line);border-radius:6px;padding:7px}header{max-width:1600px;margin:auto;padding:25px 30px 20px}.eyebrow
@sshh12
sshh12 / README.md
Last active August 21, 2026 21:15
Claude Code 2.1.238 remote/internal-model flow β€” reconstructed pseudocode and diagram

Claude Code remote/internal-model flow

This Gist documents a reconstructed control flow observed in the public Claude Code native package beginning with version 2.1.236 and still present in 2.1.238.

It is explanatory pseudocode, not Anthropic source code. Function names are expanded, unrelated application logic is omitted, and unknown private behavior remains explicitly marked as unknown.

Brief diagram

 CLI / host override       ANTHROPIC_MODEL       account/org default
#!/usr/bin/env python3
"""CLI for xAI Grok web search and X search.
================================================================================
SERVER-SIDE TOOL REFERENCE (xai_sdk 1.6.1, grok-4.5)
================================================================================
Regenerate with: scripts/inspect_grok_tools.py {local|trace|selfreport|all}
There are TWO layers, and conflating them is the main source of bugs:
@sshh12
sshh12 / identity.html
Last active August 8, 2026 20:46
Exploring Claude/GPT Knowledge Cutoffs
<!doctype html>
<meta charset="utf-8">
<title>Identity vintage - which era does each model think it is from?</title>
<style>
body { background:#f9f9f7; color:#0b0b0b; margin:0;
font:13px/1.45 system-ui, "Segoe UI", sans-serif; }
.wrap { max-width:1500px; margin:24px auto 60px; padding:0 18px; }
h1 { font-size:20px; margin:0 0 6px; }
.sub { color:#52514e; max-width:960px; margin:0 0 14px; }
@sshh12
sshh12 / pr-artifact.sh
Created July 20, 2026 22:09
pr-artifact β€” attach images & videos to a pull request without committing binaries
#!/usr/bin/env bash
#/ pr-artifact β€” attach images & videos to a pull request without committing binaries.
#/
#/ HOW IT WORKS
#/ Screenshots, screen recordings and diagrams don't belong in git history β€” they
#/ bloat every clone, forever. But GitHub *release assets* live outside the git tree:
#/ they're blob storage hung off a tag, not commits. This script uploads your media
#/ as release assets and hands back a stable
#/ https://github.com/<owner>/<repo>/releases/download/<tag>/<file>
#/ URL you can paste straight into a PR description or comment.

Goal: Build a local browser app (WebGPU). Left panel shows reference.png (a Mars surface photo taken ~100m above ground). Right panel is a free-fly 3D render of a 100mΓ—100m tile that is geologically consistent with the reference β€” a newly synthesized tile in the same morphology and style, not a pixel-for-pixel reconstruction of the same patch. It should be nearly photorealistic, comparable in realism to the reference; matching it pixel-by-pixel is not a goal. Generality is the core constraint. The pipeline must infer what it needs from the reference and synthesize from that. Pointing it at a different reference image must change the output appropriately β€” geometry, materials, and scatter. No Mars-specific or photo-specific hardcoded constants, no magic numbers that only work for this image. If something is baked to this photo, it's wrong. Geology engine: From the reference, infer the dominant process(es) β€” e.g. aeolian dunes/ripples, impact/ejecta fields, layered/sedimentary, fluvial, talus β€” and the statisti

export const meta = {
name: 'remaining-roadmap',
description: 'Plan + adversarially critique + dependency-sequence ALL remaining roadmap sprints (5 surface-scale LOD, 6 aerial/orbital, 7 Venus, 8 Earth, 9 transient effects); generate the imagegen voxel validation anchors for the new bodies/regimes; and BUILD the one LOD-independent tractable body (Venus) end-to-end with a [V] critic gate. Produces a vetted execution blueprint + real validation targets + an attempted Venus, validating/critiquing as it goes.',
phases: [
{ title: 'Engine facts' },
{ title: 'Per-sprint plans' },
{ title: 'Adversarial critique' },
{ title: 'Sequencing blueprint' },
{ title: 'Validation anchors' },
{ title: 'Venus build' },
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>PlanetLab β€” Procedural Planet, Moon & Small-Body Builder</title>
<style>
:root{
--bg:#0a0c10; --panel:#12161d; --panel2:#1a1f29; --line:#283040;
--txt:#cfd8e6; --dim:#8a97ad; --accent:#ff8a4c; --accent2:#5cc8ff;

Magnifica Humanitas

Encyclical Letter of Pope Leo XIV on safeguarding the human person in the age of artificial intelligence (15 May 2026)

*A plain-language translation of Pope Leo XIV's encyclical, rendered for a reader who is secular and technical rather than Catholic, and who may have little or no familiarity with the Bible or Christian thought. I have tried to preserve the meaning sentence by sentence, and to keep every claim the original makes β€” including its theological ones. Where the original reaches for the language of the Church, I have kept the claim and put it in plainer words, leaving its convictions standing as the author's stated convictions, so that you can follow the argument whether or not you share the belief. Biblical stories, figures, and images are briefly explained the first time they appear. The original's scholarly citation footnotes (references to earlier Church documents) have been dropped; named authors and works are folded into the text. The paragraph numbers are preserved