Skip to content
View Vick606's full-sized avatar
6 followers · 11 following

Achievements

Achievement: Pull SharkAchievement: YOLO

Achievements

Achievement: Pull SharkAchievement: YOLO

Block or report Vick606

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. Markdown is supported. 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
Vick606/README.md

Software engineer focused on Python, AI security, and developer tooling.

I build tools that make AI systems measurable and harder to break: detecting prompt injection buried in agent tool outputs, running repositories in isolated environments, and producing structured reports that downstream systems can trust.


Recent work

Subcanopy Guard — context-aware indirect prompt injection scanner for AI agent tool outputs. Sliding-window density plus stylometric discontinuity.

86.5% caught on AgentDojo v1 · 0.85 ms per scan · zero runtime dependencies · pip install subcanopy-guard

RepoProbe — clone, containerize, and test any Python repo, then emit a structured JSON evaluation report.

v0.1.0 shipped · 22 unit tests passing · verified on pallets/click


Stack

Tech stack

Focus for 2026

Shipping open-source tooling for AI security and evaluation. Subcanopy Guard and RepoProbe are live. Next: deeper coverage of the OWASP LLM Top 10, starting with the obfuscation and decoding layer for prompt injection detection.


Reach me

Pinned Loading

  1. subcanopy-guard subcanopy-guard Public

    Context-aware indirect prompt injection scanner for AI agent tool outputs. 86.5% caught on AgentDojo v1, 0.7ms scans, zero runtime dependencies.

    Python 1

  2. verirust verirust Public

    Rust CLI that compiles and verifies code submissions, emitting a deterministic ACCEPTED or REJECTED verdict. Built for RLVR training pipelines.

    Rust 1

  3. subcanopy-gateway subcanopy-gateway Public

    HTTP gateway for Subcanopy Guard. Scan prompts and tool outputs for injection attempts, store scan history, search known attack patterns.

    Python 1

  4. issue-scout issue-scout Public

    Curate GitHub issues into a versioned dataset for LLM bug-fixing evaluation.

    Python

  5. sid-ea-findings-lab sid-ea-findings-lab Public

    Structured findings for LLM agent and RAG vulnerabilities. Each scenario ships vulnerable, fixed, and hard-negative variants with traces, deterministic tests, and Semgrep rules.

    Python 1

  6. repo-probe repo-probe Public

    Lightweight CLI that clones a Python repo, detects its test setup, runs tests in Docker, and reports structured results.

    Python