Skip to content
View ryanportfolio's full-sized avatar
💭
extracting wisdom @ corewise.video
💭
extracting wisdom @ corewise.video
3 followers · 1 following

Achievements

Achievement: Galaxy BrainAchievement: Pair Extraordinairex4Achievement: Pull Sharkx4Achievement: YOLOAchievement: Quickdraw

Achievements

Achievement: Galaxy BrainAchievement: Pair Extraordinairex4Achievement: Pull Sharkx4Achievement: YOLOAchievement: Quickdraw

Highlights

  • Pro

Block or report ryanportfolio

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.

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
ryanportfolio/README.md

A plotter drafts a technical elevation on the left; the same sheet is marked up and checked on the right.

Ryan Allen. I build AI systems and developer tools. All of it ships through a review pipeline I built.

Products

Truenote screenshot Truenote. Citation-first RAG for support teams. Hybrid vector, full-text and trigram retrieval. CoreWise screenshot CoreWise. Extract wisdom from any link. Chat with any model. Deep research.
willaicite screenshot willaicite. Deterministic AEO and GEO audit, no model in the scoring path. Kine Fractal screenshot Kine Fractal. A rules engine for buying and selling index funds, with backtesting.
CoreWise Academy screenshot CoreWise Academy. Free course catalog on working with AI, each with objectives and exercises. MAIMCOIL screenshot MAIMCOIL. Pixel-art bullet heaven with achievements and leaderboard.

Developer tools

Harness Firmware. Self-syncing project template for AI coding agents, 20 bundled skills. STK. Claude Code hook that clamps oversized Read results into line-numbered outlines.
savetokens.tips. Usage optimization guide, every figure measured from logged commands. SDLC Audit. Scores AI-agent development discipline from GitHub metadata, deterministically.
tracebench. Replayable evals harvested from failed agent sessions.

Engineering

dealstream. Go services ingesting eight retailer feeds into Postgres, ClickHouse and Redis. strictcall. LangGraph agent with Pydantic-validated tool contracts over a SQL warehouse.

Desktop utilities

Local PTT. Offline push to talk for Windows. Global hotkey, int8 Whisper on CPU in real time. SecureWall. Default-deny Windows firewall with scoped outbound blocking.
ZipFlow. Double click a ZIP, get the folder, the ZIP deletes itself.

Prototypes

fullbuild.ai prototypes: immersive web, digital storytelling, WebGL, motion, systems.

How the work gets reviewed

The model that wrote the code never grades it.

Two differently built inspection arms check the same drawing from opposite sides, each leaving its own correction marks, with a stamp press waiting at the end.

The review stage, one diff, two independent passes. Pass one is a handoff audit: a fresh session with no memory of the build, told to break the work. Pass two is a cross-vendor review: the same diff through a different vendor’s model. Each marks the lines it flagged, on its own side of the diff, and the two disagree.

Both passes are skills I wrote: handoff audit and /codex-review. How I score all of it.

Those skills ship with Harness Firmware, the self-syncing template I start every repo from. The part that does the work is a reference library the agent writes back into: every gotcha that bites once gets recorded as a pitfall, and the next session reads it before touching that area. Knowledge accumulates in the repo instead of evaporating when a session ends.

Where to go next

fullbuild.ai carries this work with case notes and live prototypes. Background is on my about page.

Pinned Loading

  1. Harness-Firmware Harness-Firmware Public template

    Self-syncing Claude Code project template: rule kernel, helpful skills, committed memory library, session hooks. Improvements flow both ways between the template and the projects it spawns.

    JavaScript 1

  2. savetokens savetokens Public

    Token-saving techniques for Claude Code and Codex, published as a component datasheet. Measured before-and-after figures, refreshed daily. Source for savetokens.tips

    HTML 1

  3. Truenote Truenote Public

    RAG for business | Citations or nothing | Support teams need an answer they can verify

    TypeScript 1

  4. willaicite willaicite Public

    Free, deterministic GEO audit that scores whether AI answer engines (ChatGPT, Perplexity, Claude, Google AI Overviews) can retrieve, read, and quote your website

    TypeScript 1 1

  5. STK STK Public

    STK (Session Token Killer) Claude Code hook that clamps oversized Read tool results into line-numbered outlines. RTK kills tokens per command, STK per session. 85% of my oversized agent context was…

    JavaScript 1

  6. Corewise.Academy Corewise.Academy Public

    Free library of original guides on working with AI. Source and scripts for corewise.academy.

    MDX 1