agentboards.org
Compare/harness9 vs Wizard

harness9vsWizard

Generated from the two spec rows. Green marks the better value where a spec has a clear direction. Everything else is just different.

harness9
ZhangShenao · Agent harness
OSS
Panel
5.7
3 spec wins
Reliability
5.5
Usefulness
5.3
Cost
7.2
Longevity
4.7

“It positions itself between bloated frameworks and thin demos, which is the software equivalent of a dating profile that says normal.”

Wizard
teddytennant · Agent harness
OSS
Panel
5.7
1 spec wins
Reliability
5.0
Usefulness
5.5
Cost
7.3
Longevity
4.8

“It calls itself a sovereign agent, which is a great deal of constitutional language for a binary living in your home directory.”

Spec by spec

Specharness9Wizard
Architecture
CategoryAgent harnessAgent harness
Runslocallocal
Platformsmacos, linuxmacos, linux
Context windownot documentednot documented
Protocols
MCP clientNoNo
MCP serverNoNo
Capabilities
Runs terminal commandsYesYes
Multi-file editsYesYes
Git operationsYesNo
Browser controlNoNo
Sandboxed executionYesNo
Multi-agent orchestrationNoNo
Headless / CI modeNoNo
Models
BackboneOpenAI, Anthropic, OpenRouterxAI, OpenAI, Anthropic, Gemini, DeepSeek, Groq, OpenRouter, llama.cpp
Bring your own modelYesYes
Local modelsNoYes
Cost
Pricing modelbyokbyok
Starts at$0/mo$0/mo
Free tierYesYes
Bring your own keyYesYes
Openness
Open sourceYesYes
LicenseMITApache-2.0
GitHub stars141105

Which one would each critic pick

Criticharness9WizardPick
El Juez——not enough reviews
El Amigo6.36.0harness9 — Pick it if a comfortable full-screen terminal interface is what keeps you using a tool; pick something plainer if you live in a pipe and never look at it.
El Crítico5.84.8harness9 — Tools run inside a container, but the shell escape prefix and the stored state do not, so the isolation covers the tidy half of what an agent does.
El Profesor5.56.3Wizard — Memory is stored as markdown and sessions as line-delimited records, which makes the accumulated context a document a person can read and correct.
La Inversora4.55.0Wizard — 110 public mentions in a year against 66 stars, which is the widest gap between conversation and code on this board, and there is no company at all.
La Jefa5.54.0harness9 — Every plan, tool result and record stays on the developer's own machine, which answers the data residency question and creates the discovery problem.
El Hacker6.58.0Wizard — Apache-2.0, and choosing Local on first run has it size a Qwen 3 build to my hardware and start llama.cpp itself, with no key and no account anywhere.

Picks are derived from each critic's own scores. Humans vote on matchups on the duels page.

Want a third column? The compare tool handles any two agents. Three-way comparisons are on the roadmap once the spec rows are all verified.