Home

Agent Harness Bootstrap

Two Claude Code skills and a viewer. One writes specs both you and the AI understand. One fits an agent team to your repo. The third lets you see the whole thing and switch any part of it off.

This wiki is the reference. The README is the argument for why any of it exists, and the landing page is the short version of that. Start there if you have not read either.

Start here

If you want to	Read
Install it and run it once	Getting Started
Understand why the build is small on purpose	Tailored Build
Look up a seat, a rule, a hook, a command, a flag	the reference pages below
See what your harness actually looks like	Harness View
Work out why something is not firing	Troubleshooting
Ask the obvious question	FAQ

Reference

These five pages are generated from the repository by scripts/build_wiki.py. They are not maintained by hand, which is the only reason a wiki's counts can be trusted:

• Agent Reference - the seats, their model and effort, and when each is installed
• Rule Reference - which rules load in every session and which are path-scoped
• Hook Reference - what blocks what, and the exit code that does it
• Command Reference - every slash command the harness installs
• Flag Reference - what each answer in the questionnaire turns on

If you edit one of those pages in the wiki UI, the next build overwrites it. Change the asset instead, and the page follows.

The one-paragraph version

spec-builder turns whatever you brought - an idea, a meeting transcript, legacy documents, a bare repo - into one contract written to international standards, with stable requirement IDs. harness-bootstrap reads your code, then builds the .claude/ harness that fits it: a roster derived from the contract and the modules that exist, rules scoped to paths that exist, and hooks that block rather than advise. The guardrails are shell scripts and exit codes, so swapping every agent from Opus to Haiku leaves the safety floor byte-identical. harness-view reads the result and scores it, with no model in the loop.

Proof you can run yourself

python eval/guardrail_eval.py                        # the safety floor, per hook flavour
python benchmark/benchmark.py                        # harnessed vs a bare repo
python harness-bootstrap/scripts/port.py --self-test # the Cursor and Codex adapter
python scripts/check_numbers.py                      # every published figure vs the scripts

Nothing on this wiki is a number somebody remembered. If a page and a script disagree, the script is right and the page is a bug.