Getting startedDraft
Everything you need to start is free.
The framework is open source and runs from your repo with no account. A free humanmaxxing.ai account is the second step, not a paywall: it adds the reports and the tracking you would otherwise have to build yourself.
Free · open source · no account
1Install the framework
Open your coding agent in your repo and paste this. It clones the framework, interviews you, builds a team of specialist agents from your own code, and installs the workflows they run in. It stops for your approval before touching anything.
Clone https://github.com/pgartlan/humanmaxxing into _humanmaxxing/ in this repo, then read _humanmaxxing/getting-started.claude.md and follow it.
github.com/pgartlan/humanmaxxing · Read it before you run it.
It builds your agent team
This is the part that makes the rest work. The framework does not ship a generic roster. It reads your repo and proposes the specialists your code actually needs, and you approve the roster before a single agent file is written.
- SurveyReads your repo the way a new senior hire would: frameworks, test runners, auth boundaries, migrations, and the business domains with enough code to deserve their own expert. Read-only.
- ProposeA roster of four to eight agents, each naming the finding in your repo that earned its place. You approve, trim or rename.
- GenerateAgent files written with your real commands and paths. Reviewers are read-only; builders get exactly the tools they need.
- WireReviewers are mapped to your paths, so plan review and code review call the right specialist when its files change.
A typical rostercode-reviewertest-engineersecurityarchitectureperformanceone expert per business domainHow the team is built →
What happens after you paste
- InterviewOne message, nine questions, answered once. The one it will not guess: the rule in your domain that can never silently drift.
- SurveyIt checks what you said against the repo: manifests, CI, tests, branch protection. Read-only.
- PlanAn install order built for your repo. It stops here. Nothing is written until you approve.
- InstallOne piece at a time: your project contract first, then your agent team, then the workflows. A progress ledger lets you stop on Tuesday and resume on Thursday.
What you get
starta ticket becomes an investigated, reviewed planreview-plana team of reviewers reads the plan before any code existsreview-codeindependent code review with a second-opinion modelcheckpointdocs, quality checks and the commitshipcommit to PR to merge, behind blocking gatesbootstrap-agentsbuilds your team of specialist agents from your repobootstrap-claude-mdsfolder-level conventions where they are earnedtemplatesthe behavioral contract and the procedures behind it
Also free · one organization
2Connect a free account
Sign up, add the reporting hook to your repo, and open a pull request. From then on the reports write themselves and the numbers keep themselves.
How it connects
- Create the accountThe dashboard is hosted by OpsVis, so sign-up and sign-in happen on its site.
- Create a project keyOn the dashboard’s Admin tab. Save it in your repo’s untracked .env.local as OPSVIS_CLI_API_KEY.
- Paste the install promptFrom the same tab, into Claude Code. It adds the OpsVis telemetry plugin and stops so you can read the diff before you commit.
Reports that arrive on their own
- A retro of the session after every PR: where the time went, what stalled, which reviewers found real issues
- An email when spend crosses a threshold you set
Tracking you do not have to build
- Human time against machine time, for every change
- Cost per PR, and what each agent costs you
- Reviewer performance: issues found, and cost per issue found
- Cycle time from ticket pickup to merge, stage by stage
- 30 days of history
Free up to 50,000 events a month. An event is one agent turn.
What shows up Sample data
Cost per PR
Model spend per merged pull request
$8.70-41% vs 12 wk ago
$8.70 week of Nov 2
Table
| Week | USD |
|---|---|
| Aug 17 | $14.80 |
| Aug 24 | $13.90 |
| Aug 31 | $15.20 |
| Sep 7 | $12.60 |
| Sep 14 | $11.90 |
| Sep 21 | $12.40 |
| Sep 28 | $10.80 |
| Oct 5 | $10.10 |
| Oct 12 | $10.60 |
| Oct 19 | $9.40 |
| Oct 26 | $9.10 |
| Nov 2 | $8.70 |
Reviewer performance
Confirmed issues found in the last 12 weeks, by reviewer agent
148code-reviewer · highest of 8
What is free, exactly
| Feature | Framework only | With a free account |
|---|---|---|
| A team of specialist agents built from your repo | ✓ Included | ✓ Included |
| All seven skills and both templates | ✓ Included | ✓ Included |
| Plan review and code review teams | ✓ Included | ✓ Included |
| Blocking gates before a PR can open | ✓ Included | ✓ Included |
| Runs from your repo, yours to adapt | ✓ Included | ✓ Included |
| Real-time dashboard: the two clocks, cost, cycle time | — Not included | ✓ Included |
| Cost per agent and reviewer performance | — Not included | ✓ Included |
| Session retro after every PR | — Not included | ✓ Included |
| Spend thresholds with email alerts | — Not included | ✓ Included |
| 30-day history | — Not included | ✓ Included |
| Price | $0 | $0 |
Nothing in either column is a trial. The framework never needs the account, and the account never locks the framework.
What the account changes in your repo
One reporting hook. It sends one event per agent turn. Nothing is reported until you add it, and removing it stops the reporting. The skills, the reviews and the gates keep working either way.
When you would pay
When you want the weekly retro run for you, teams and roles, alerts in Slack, or history past 30 days. Until then there is nothing to buy. Pricing →
Questions
Do I need an account to use the framework?
No. The skills, templates and workflows run from your repo with no account and no network call to us.
Is the free account a trial?
No. It does not expire. It covers one organization up to the monthly event allowance.
What happens if I pass the free allowance?
Dashboards keep working and we tell you. Nothing is cut off mid-month; you choose whether to move to Growth.
Which agents does it work with?
Built and run daily on Claude Code. The skills are plain Markdown and shell, so other agents can follow them; hook enforcement is Claude Code-specific today.
Can I stop?
Yes. Remove the hook and reporting stops. Everything the framework installed in your repo is yours and keeps working.