ronaldpetty

OSI layer	User space	Kernel space
Layer 7 — Application	OIDC, OAuth2, HTTP, HTTPS, HTTP/2, HTTP/3, gRPC, TELNET, CAC-based login flows	eBPF sometimes, for L7 observability when paired with parsers/agents
Layer 6 — Presentation	TLS libraries, HTTPS, certificate handling, CAC certificate use	Usually not kernel; TLS is mostly user-space, though kernel TLS exists in some systems
Layer 5 — Session	OAuth/OIDC login sessions, TLS se

ronaldpetty

%%manim -qm -v WARNING OIDCSimpleIDTokenFlow

from manim import *
import numpy as np

class OIDCSimpleIDTokenFlow(MovingCameraScene):
    def make_lane(self, title, x, width=3.0, height=6.4):
        body = RoundedRectangle(
            corner_radius=0.12,
            width=width,

ronaldpetty

%%manim -qm -v WARNING OIDCAppsK8sAmbientFlow

from manim import *
import numpy as np


class OIDCAppsK8sAmbientFlow(MovingCameraScene):
    # ---------- helpers ----------
    def make_lane(self, title, x, width=3.0, height=6.4):
        body = RoundedRectangle(

ronaldpetty

# This example session in on Mac
# 
# % docker compose up -d ssh-server                                                                                                                                                             
# [+] up 2/2
#  ✔ Network qday-test_sshnet Created
#  ✔ Container ssh-pqc-server Started
# 
# % ssh -vvv -p 2222 -o StrictHostKeyChecking=no -o UserKnownHostsFile=/dev/null -o Ciphers=aes128-ctr \                              
#   testuser@localhost 2>&1 | grep -E "kex: "
# debug1: kex: algorithm: ecdh-sha2-nistp256

ronaldpetty

Codex provides a slash command to check status called /status. This command cannot be ran outside of Codex.

To get similar results (for now just the timing/amount), this script reviews local storage to report similar infomration via CLI tool

# create script in local `$HOME/bin` directory
mkdir ~/bin

ronaldpetty

Date	Model	Lab	Purpose	Size	Interesting fact
2019-02-14	GPT-2	OpenAI	General language generation	Up to 1.5B	Staged release due misuse concerns; full 1.5B release came later in 2019. ([OpenAI][1])
2020-05-29	GPT-3	OpenAI	General LLM / few-shot learning	175B	Made “prompting” and few-shot learning a mainstream concept. ([arXiv][2])

ronaldpetty

Multi-Agent Git Branch + Worktree Workflow

Assumptions:

main branch has normal project code only.
agent/codex branch has main + AGENTS.md.
agent/claude branch has main + CLAUDE.md.
main should not keep AGENTS.md or CLAUDE.md.
task details are passed as URLs in the agent prompt.

ronaldpetty

class MyMeta(type):
    def __new__(cls, name, bases, attrs):
        mod_attrs = {}
        for attr_name, attr_value in attrs.items():
            if attr_name.startswith("__"):
                mod_attrs[attr_name] = attr_value
                continue

            cap_name = attr_name.capitalize()

ronaldpetty

import subprocess
import os
import time

# Start the Python REPL in interactive mode, with a pipe for stdin and stdout
proc = subprocess.Popen(
    ["python3", "-i"],
    stdin=subprocess.PIPE,
    stdout=subprocess.PIPE,
    stderr=subprocess.STDOUT  # Merge stderr into stdout for simplicity

ronaldpetty

...
- initcontainers
  - name: blocker
  - command:
  - sh
  - -x
  - -c
  - |
  broken = true
  till_not_broken broken: