Skip to content
View loafecho's full-sized avatar

Block or report loafecho

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
loafecho/README.md

Tony Garcia

Customer ops leader who builds AI tools that people can trust. Founder of loafecho.

Selected work

  • print-compliance: screens STL and 3MF files for geometry that resembles regulated weapon parts and returns PASS, REVIEW or BLOCK with a reason a person can check.
  • ygo-duel-agent: trains a small language model to play Yu-Gi-Oh on one consumer GPU, using a Rust duel runner on the open-source EDOPro engine and evaluations fixed before each training round.
  • surface-pro7-linux-camera: kernel and userspace patches for the Surface Pro 7 cameras and touchscreen driver on Linux, with the fault-injection scripts and offline checkers used to test them.
  • glia-agent: a maintenance agent for a Markdown notes vault that can only propose changes; a person approves each one and a separate executor applies it as one revertible commit.
  • agent-sandbox: least-privilege microVM sandboxes for AI agents, plus a Rust tool server that holds risky actions until a person approves them.
  • homelab-as-code: infrastructure as code for a one-node Proxmox homelab (OpenTofu, Ansible, Argo CD), with firewall rules tested against the ports that are actually listening and a drift gate that answers PASSED, DRIFT or ERROR.

How I build

  • Checks that must be able to fail. A test only counts if removing the thing it protects turns it red. glia-agent switches off each of its guards in turn and requires the suite to fail; the camera repo's evidence checker recomputes each verdict from the raw test lines instead of trusting the summary.
  • Approval before action. In glia-agent and agent-sandbox the agent proposes or requests, a person approves, and only then does anything change. The rules for what an agent may never touch live in code, not in a prompt.
  • Honest limits. Every README has a Limits section that says what was not tested and where the tool falls short, such as geometry that can send a file to review but cannot reject it, or patches tested on one device.

Pinned Loading

  1. glia-agent glia-agent Public

    Propose-only maintenance agent for a Markdown knowledge vault, with an approval ladder enforced in code

    Python

  2. agent-sandbox agent-sandbox Public

    Sandboxed boxes and an approval switchboard for running AI agents with least privilege

    Rust

  3. ygo-duel-agent ygo-duel-agent Public

    Self-play training and evaluation harness for a Yu-Gi-Oh! duel agent

    Rust

  4. print-compliance print-compliance Public

    File-level screening engine for STL/3MF prints: slice-based bore detection, transparent rule scoring, measured false-positive floor

    Python

  5. surface-pro7-linux-camera surface-pro7-linux-camera Public

    Linux kernel patches and test tooling for the Surface Pro 7 cameras and touch

    Python

  6. homelab-as-code homelab-as-code Public

    Homelab infrastructure as code: OpenTofu, Ansible and Kubernetes manifests

    Shell