pakkinlau

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
outgoing katex syntax transformer(for diff or patch files).py — Make LLM-ish Markdown safe from ChatGPT to other place with other rendering format, such as Obsidian/Quarto (KaTeX/MathJax) etc.

PURPOSE
  Convert ChatGPT/Codex-style Markdown-with-math into a strict, portable
  Markdown+LaTeX subset that renders cleanly. The script normalizes delimiters, fixes common LaTeX
  pitfalls, and refuses to touch code/links/previous math.

pakkinlau

#!/usr/bin/env python3
"""
outgoing edits (.patch or .diff) validator.py — Validate that a unified-diff patch (or zip bundle of patches) is machine-applicable.

Goals:
- Catch syntactic diff corruption that breaks `git apply` / `patch`:
  * missing diff prefixes on blank hunk lines
  * hunks without proper @@ headers
  * missing ---/+++ headers for each diff --git section
  * malformed hunk count (observed line counts don't match @@ header)

pakkinlau

#!/usr/bin/env python3
"""
repo_snapshot_query.py

Utilities for querying a repo snapshot bundle produced by the
`agentic_repo_snapshot_curator.py` snapshot tool (or any tool that emits the same layout).

Snapshot layout (on disk)
-------------------------

pakkinlau

#!/usr/bin/env python
"""
Estimate token usage and cost from a ChatGPT web chat log (JSON export),
with:

- Configurable context window (to simulate truncation).
- Optional approximate reasoning token cost.

User-configurable parameters are at the top of this file.
"""

pakkinlau

#!/usr/bin/env python3
"""
codex_log_query.py

Small utility module for querying Codex-style JSONL work logs.

Assumed log shape
-----------------
Each line in the JSONL file is a JSON object of the form::

pakkinlau

#!/usr/bin/env python3
"""
repo_snapshot_query.py

Utilities for querying a repo snapshot bundle produced by the
`agentspine` snapshot tool (or any tool that emits the same layout).

Snapshot layout (on disk)
-------------------------

pakkinlau

#!/usr/bin/env python3
"""
gpt2mermaid.py — Deterministic Mermaid Normalizer (clipboard-in, clipboard-out)

Purpose
-------
Normalize raw Mermaid flowchart text copied to the clipboard into a
strict-parser-friendly form, then write the normalized result back to
the clipboard. The transformation is fully deterministic and avoids
intent inference; it focuses on syntax hygiene and common portability

pakkinlau

#!/usr/bin/env python3
"""
agentic_repo_snapshot_curator.py

Purpose: create per-root JSON snapshot bundles (index + sharded JSONL) with stats, while letting agents tune
which zones to include/skip without editing code.

Snapshot naming
- Snapshot id: repo=<slug>[__branch=<base> ]__ts=<UTC>__sig=<hash>[__tag=<tag>].
- Artifacts live in repo-snapshot-json/<snapshot_id>/; filenames are prefixed with the snapshot id.

pakkinlau

🎯 Big goal

():

• Context: <2–3 sentences describing why this goal matters, without solution hints.>
• Constraints & assumptions:
• Stakeholders / owners: <names/roles>

---

pakkinlau

<#
.SYNOPSIS
    One-shot installer & launcher for FaceFusion 3.3.0 (Windows 10/11).

.DESCRIPTION
    – Creates a Conda env “facefusion” with Python 3.12  
    – Installs CUDA 12.9 runtime + cuDNN 9.10 (NVIDIA GPUs)  
    – Installs FFmpeg 7.0.2 globally *and* inside the env  
    – Clones FaceFusion from GitHub and installs with ONNXRuntime-GPU  
    – Force-downloads all model weights (resumable)