The workshop floor: a multi-agent SDD harness, CLIs that do one thing well, and tooling kept intentionally small. Everything it needs, nothing it doesn’t.
A standardized multi-agent harness for Claude Code and OpenCode: Spec-Driven Development roles (Leader, Spec Author, Implementer, Reviewer) installed into any project with a single command. Its process also ships as six standalone skills (npx skills add jordimarsal/harness).
The problem. Coding agents work well in the small and drift in the large. Teams need one repeatable process — specs, roles, gates — independent of stack.
Highlights
7 stacks covered (Java/Spring, Python, Node/TS, Rust, and more) with per-stack conventions.
Single-command install; conventions template includes SonarQube/code-quality rules.
Born from daily use across personal projects — this portfolio is built with it.
Its process ships as six standalone skills, installable with npx skills add jordimarsal/harness.
An agentic benchmark for local GGUF models: prompt speed, tool-calling, code generation and multi-step agent work graded under one identical harness, with the raw receipts of every call — 100% local, no cloud, no APIs.
The problem. Public benchmarks are trained on, and “how smart is it” is not the question a local rig answers. ggufbench asks which GGUF to run on your own machine: every model runs the same harness, every scoreboard row states the task set it really ran, and exclusions are recorded with a reason — never silent.
Highlights
Four dimensions under one identical harness: prompt speed, tool-calling, code generation and multi-step agent work — three runs per task (seeds 42/43/44) on the same 2× RTX 5080 rig, reported as mean ± stdev.
The current task set: 18 tool, 72 code and 72 agent tasks — versioned (v1 → v5) so every scoreboard row states the set it really ran; a single 32768-token reasoning budget and thinking modes measured separately.
100% local — no cloud, no APIs, no telemetry — and receipts included: every call leaves timings, tool calls and per-task traces for a live scoreboard, and 13 retired models stay as record only.
0.85best global score — minicpm-v-4.7-35b-a3b q4_k_m, thinking off
CLI that renders Markdown with Mermaid diagrams straight to PDF, engineered with TDD, mypy strict and CI.
The problem. Docs with diagrams rot when the toolchain is manual. This CLI makes the Markdown→PDF step a one-command, reproducible build — diagrams included.
Highlights
Quality as a feature: full TDD suite, mypy strict typing and CI on every push.
