Traction scoreGitHub stars can be faked, so popularity alone can be misleading. Traction Score looks for broader signs of recent attention, adoption, and active maintenance.
Design-harness provides a structured environment for technical leads and architects to conduct literature reviews, vendor selection, and competitive analysis. It organizes raw sources into ideas and final outputs using a three-layer markdown system, allowing an agent to handle the organization while the human makes the final adjudication.
The tool is built for contested design decisions that require a clear audit trail. By linking disparate papers and half-formed ideas into a cohesive system, it transforms fragmented research into a defendable technical design.
WHO IT'S FOR
Technical leads / architects
documenting evidence for system design decisions
Research engineers
conducting literature reviews and due diligence
Product managers
performing competitive analysis and vendor selection
Compatible AgentsThe repository documents support for these agents. The skills may also work with other agents that can load SKILL.md files, but they may need some setup or small changes.
Installs: The single design-harness skill folder · Other SKILL.md-compatible agents
Before you start
A SKILL.md-compatible agent
Git available in the terminal
Python 3 already installed for the skill's documented runtime commands
Run from: A directory where the repository can be cloned as ./design-harness
Run this documented command block in a terminal to clone the repository and copy the design-harness folder. The shown destination, `~/.claude/skills/`, is Claude-specific; when installing for another SKILL.md-compatible agent, replace that destination as you enter the copy command with the selected agent's own skill directory. The command text below preserves the author's documented form.
Use the design-harness skill to file these papers onto the board in [workspace path]: [paper file paths or URLs]. For this first pass, create one source card per paper, grade it, and anchor its claims to the source. Do not assemble the final output yet.
Use the design-harness skill in [workspace path] to organize these judgments as idea cards: [your judgments]. Link each judgment to relevant existing evidence, cluster related ideas, and surface conflicts without deciding between them for me.
Use the design-harness skill to build the visual canvas for the existing workspace at [workspace path] and give me the local clickable result. Do not assemble or change the final output.