Skip to content
View jadmadi's full-sized avatar
🎯
Focusing
🎯
Focusing

Organizations

@WaqfTech

Block or report jadmadi

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
jadmadi/README.md

Jad Madi - Founder & toolsmith

jadmadi's GitHub stats

Salam there 👋 I'm Jad. I build tools for Arabic AI fluency benchmarking (Mahak مِحَكّ) and multi-agent governance & memory (Sila صِلَة, private preview).

I work in Go, TypeScript, and SQLite: local-first, fast, and read-only by default. Find me at jadmadi.net and x.com/jadmadi.

I look for thermal design partners, WaqfTech editor maintainers, and Arabic eval annotators for Mahak.

Featured AI devtools and MCP infrastructure

  • thermal (Go, SQLite, TUI, Web): GitHub-style contribution heatmap, multi-agent streak tracker, 9-box FinOps dashboard, workload replay simulator, and verifiable work receipts for 14+ AI coding agents (Devin, OpenCode, MiMoCode, Codex, Claude Code, Antigravity/Agy, command-code, codewhale, Droid, DeepSeek, Grok CLI, Muse, Nous Hermes, ZCode). Includes cross-agent disjoint token accounting, token yield, reasoning effort telemetry, and embedded local web dashboard. Install via curl -fsSL https://jadmadi.net/thermal/install.sh | sh. Essay: My MCP Stack: How I Run My Own Agent Tools.
  • sila (Go, SQLite, MCP, private preview): local-first continuity, alignment, and governance substrate for autonomous AI coding agents:
    • 58-Tool MCP Server: equips agents across Claude Code, Cursor, Windsurf, OpenCode, Codex, and Antigravity with unified cross-harness memory, goal leasing, and runtime orchestration.
    • Multi-Harness Ingestion: non-destructive, read-only loaders (mode=ro) that ingest session transcripts across 10+ AI CLI tools in milliseconds into pure-Go SQLite (modernc.org/sqlite, FTS5) with zero telemetry.
    • Project Skill Governor: defeats the silent 50-skill cutoff and prompt truncation defects, diagnoses LCS visual collisions, and dynamically serves skills on demand via MCP.
    • Universal Multi-Toolchain Clean (sila clean): safe disk hygiene and bloat reclamation across 9 developer ecosystems (Go, Rust/Cargo, pnpm, aube, npm, uv/pip, bun, zig, mise).
    • Roadmap Goals & Stage-Gates: coordinates epistemic goal leasing, focus binding, multi-agent proposal comparison, and automated preflight verification (sila goals, sila report).
    • Pre-Commit Discipline Gates (sila gate): local verification hooks enforcing test passes, formatting, changelog updates, and architectural scar-tissue scanning before commits.
    • Deterministic Rule Compiler: compiles single-source hub-and-spoke AGENTS.md rules into harness-native instructions (CLAUDE.md, .cursorrules, system prompts) and maps repository topologies.
    • Mission Control Cockpit: live 9-box interactive TUI cockpit (sila watch), telemetry snapshots (sila stats), morning standup briefings, and cross-session distillation.
    • Data Sovereignty Covenant (sila doctor --covenant): guaranteed 100% local custody, open relational FTS5 export, and pre-egress secret scrubbing.
    • Install via curl -fsSL https://sila.jadmadi.net/sila/install.sh | sh (Linux & macOS) or irm https://sila.jadmadi.net/sila/install.ps1 | iex (Windows).
  • surge (Go, CLI): lightning-fast, concurrent HTTP load testing tool in Go — tail latency percentiles (p50/p90/p95/p99), real-user traffic presets (baseline, realistic, capacity, spike), automated health verdicts, and JSON output for CI/CD performance regression gates.
  • skills (Agent skills, Markdown): curated agent skills for Claude Code, Cursor, Windsurf, Codex, Devin, and Antigravity — ADOO structured problem-solving, backend pre-flight check, codebase-artifact HTML docs, and Cloudflare Workers CPU/D1 audits. Available via npx skills add jadmadi/skills and on skills.sh.
  • skill-cabinet-go (Go, UPX): single-binary companion to subsy/skill-cabinet. Scans agent skill drawers (.agents, .claude, .codex, .cursor), serves a local web desk, and provides a scriptable CLI (ls, show, audit, delete, dedupe, snapshot/diff). Byte-identical parity with Node, 3.5 MB UPX-packed, self-updating.
  • open-streak (TypeScript, CLI): GitHub-style terminal activity heatmap and streak tracker for OpenCode.
  • mimo-streak (TypeScript, CLI): terminal activity heatmap, model breakdown, and cost tracking for MiMoCode.
  • agentmeter (Go, CLI, Web): local-first analytics and cross-tool comparison for Command Code, OpenCode, Muse, and DeepSeek Harness (dsh). Project page: jadmadi.net/project/agentmeter.

OpenCode plugins and themes

The opencode-plugins fleet standardizes plugins on CalVer (v2026.9.0), builtUnder provenance tracking, and OpenCode 2.0.16 compatibility:

Open source contributions

  • subsy/skill-cabinet (PRs #4–#8, merged): core performance overhaul cutting initial scan time from ~7s to ~0.83s (8x speedup) with byte-identical output, storage reality link census, and disk redundancy reporting.
  • google-gemini/gemini-skills (PR #78, merged): fixed Codex plugin installation command syntax in official documentation.
  • zed-industries/extensions (PR #7409, merged): upstreamed mimo-code-theme to the official Zed extension registry.
  • muayyad-alsadi/fs-relay (PRs #2–#40, merged): advisory record locks across mounts, lock-race and write-behind fixes, CLI hardening, and a race-detector CI job for a networked filesystem.
  • danicat/kungfu (PR #2, merged): subcommand help, flag parsing, and exit-code fixes for a zero-dependency Agent Skills CLI gateway.

WaqfTech: Islamic tech (وقف تك)

  • mahak-bench (Astro, TypeScript, Go): open community benchmark that evaluates AI models on native Arabic fluency across everyday tasks (contracts, correspondence, creative writing, customer support, and instruction following). Includes web arena, Go CLI, and MCP server at mahak.waqf.dev.
  • waqftech-markdown-editor (TypeScript, React): lightweight Markdown editor with built-in auto-formatting for Islamic texts (Quranic brackets, Hadith quotation marks, classical ligatures, and guided Quranic verse search). Published as @waqftech/markdown-editor. Live demo.
  • waqf-license-draft (Legal, Spec): Waqf-DPL 1.0 (رُخصة وَقْف الرَّقْمِيَّة العامَّة), a digital public license framework governing software and content released as continuous endowment (وقف لله تعالى) across 7 tiered levels.
  • waqf-mcp (Astro, TypeScript, MCP): edge-native Islamic MCP federation gateway connecting Muslim developers and AI assistants to authentic Islamic sources. Live at mcp.waqf.dev.
  • Athan.waqf.app (TypeScript, Web): interactive 3D planetary observatory visualizing global Islamic prayer fronts and adhan continuity. Live at athan.waqf.dev.

Tooling and themes

  • zed-mimoCode-theme (Zed theme): warm dark and light theme pair for the Zed code editor with signature orange accents. Listed in the official Zed Extensions catalog.
  • Go-OPML (Go): high-performance concurrent OPML to JSON converter and RSS processor (GPL-3.0).
  • 91 Places to Launch (Web Tool): interactive distribution workbench ranking 91 high-authority startup directories and launch platforms by Domain Rating (DR 38–98) with real-time submission tracking.
  • SwapAudit (Go, CLI): swap and volatile memory auditor with surgical cleanup for Linux developer machines. Project page: jadmadi.net/project/swapaudit.

Live proof

Connect

Pinned Loading

  1. thermal thermal Public

    Multi-agent terminal heatmap, token FinOps, workload replay, and verifiable receipts for 15+ agents

    Go 2

  2. open-streak open-streak Public

    GitHub-style terminal activity heatmap for OpenCode usage data

    TypeScript 1

  3. WaqfTech/waqftech-markdown-editor WaqfTech/waqftech-markdown-editor Public

    TypeScript

  4. agentmeter agentmeter Public

    Local-first analytics and cross-tool comparison for Command Code, OpenCode, Muse, and DeepSeek Harness (dsh).

    Go 4