Skip to content
View JoshuaPinckard's full-sized avatar

Block or report JoshuaPinckard

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
JoshuaPinckard/README.md

Josh Pinckard

ToolsEnabled was founded and created by Joshua Pinckard. The original platform was developed by directing autonomous AI-agent fleets through the system's own evolving coordination architecture.

Research: multi-agent coordination, concurrent file mediation, and benchmarks for LLM code generation.

Selected work

  • ToolsEnabled (current). A workspace for Codex and Claude agents on Windows. Free software, MIT licensed.
  • LEAN-bench (2026). What language models do with trading-strategy specifications, graded by running every generated program in a pinned LEAN engine against a precomputed oracle bank.
  • byte-interval-mediation (2026). Read-receipt write admission for concurrent coding agents: a write is admitted only if the bytes it replaces are the bytes its author read.
  • agent-tree-fanout (2026). Span of control for hierarchical LLM agent trees: data from one deployed tree, a queueing model, and a reference bound.
  • Blast-Radius (2026). A scoring function that predicts which regions of a repository a change may semantically invalidate, in any language, with no parser. Design phase.
  • MyAssistant (2025). A Windows desktop agent that reads the screen with OCR, plans with OpenAI or local models, acts through the mouse and keyboard, and verifies each step.

Preprints

  • Byte-Interval Mediation: Read-Receipt Write Admission for Concurrent LLM Coding Agents. 2026. PDF
  • Fan-Out Is Not Fan-In: Span of Control for Hierarchical LLM Agent Trees. 2026. PDF

Contact

josh@toolsenabled.ai · https://toolsenabled.ai

Popular repositories Loading

  1. language-tutor-privacypolicy language-tutor-privacypolicy Public

    Privacy Policy for Language Tutor Extension

    HTML

  2. byte-interval-mediation byte-interval-mediation Public

    Read-receipt write admission for concurrent coding agents. The ToolsEnabled byte authority, extracted, with tests, design probes, data and a preprint.

    JavaScript

  3. agent-tree-fanout agent-tree-fanout Public

    Span of control for hierarchical LLM agent trees: text-free timing data, a queueing model, the fan-out bound and a preprint.

    TeX

  4. JoshuaPinckard JoshuaPinckard Public

    Profile

  5. MyAssistant MyAssistant Public

    Windows desktop agent (2025): reads the screen with OCR, plans with OpenAI or local models, acts through mouse and keyboard, verifies each step.

    Python

  6. co-read co-read Public

    Blast-Radius: predicting which regions of a repository a change may semantically invalidate, in any language, with no parser. Design phase.

    Python