kenji mark: a pixel kensaurus head on limekenji skills
162 agent skills, 63 slash commands, 6 subagents

You say the job.
The playbook runs.

Installed workflows for Claude Code, Cursor, Codex CLI, and Gemini CLI. They ask first, test first, and stop only when the work is proven.

npx @kensaurus/skills --auto

Then remember one command: /kenji and the job in your own words. It picks the playbook and runs it.

Animated capture of real output. The command /kenji with no job prints: What do you want to do? Say the job in your own words and I will pick the playbook and run it. Then a table of jobs and the skill for each, such as fix a bug and ship the fix: workflow-fix-and-ship. You say: fix this bug and ship it. workflow-fix-and-ship triages with evidence, fixes with a test, verifies in the browser. Tests green, PR open.

What happens after install. You say the job in chat. The matching playbook runs, proves the result, and opens the PR.

THE COMPLETION GATE

Diagram of the completion gate. 1, the state file holds the checklist on disk. 2, the Stop hook runs when the agent tries to stop and sends it back while items are open, giving up after three tries with no progress. 3, when every item is checked, a judge in a fresh context compares the request, the diff, and test evidence, and returns PASS, CONTINUE, or BLOCKED. Animated capture of real output. The state file has two open items and the agent says it is done. The Stop hook returns decision: block and lists the two items. Once both are checked, the hook returns an empty object and the agent may stop.

OTHER WAYS TO INSTALL

MORE FROM KENSAURUS