All 23 tools
The complete 15-tool single-agent lifecycle plus 8 commercial Team tools for up to five registered installations. The full single-agent component retains Apache-2.0 rights; the Team component has its own commercial license.
Complete Local · $19 USD once · no subscription
The full proof runtime for a local agent team.
Run all 23 Living Stack tools: the complete single-agent proof lifecycle plus a private workspace for up to five installations, versioned policy history, metadata-only signed traces, and workspace audit export.
Stripe hosts payment entry. Living Stack never receives raw card credentials. The buyer's browser creates an Ed25519 installation key locally and saves a recovery kit before checkout; only the public key leaves the browser.
Exact purchase
What $19 buys.
Exact private package
Fulfillment is bound to a version, byte size, and SHA-256. Package bytes stay in a private vault and are released only after verified payment and entitlement checks.
Signed local activation
The checkout receipt, license grant, installation key, and package hash are bound together. Refunds and disputes revoke or suspend future access.
Human and agent paths
One offer, two explicit checkout modes.
Human checkout
Open the secure checkout, save the browser-generated local recovery kit, complete Stripe's required approval, then download the Stripe-webhook-verified fulfillment bundle. Living Stack never receives raw card credentials or the installation private key.
Agent checkout
An authorized MPP client can read the exact offer, obtain Stripe's payment challenge, and submit the identical bound request through the published agent endpoint. A successful MPP receipt commits the bound entitlement synchronously; discovery and preparation never charge.
Recovery
A paid order can be recovered without repeating payment. A stale, changed, replayed, refunded, or disputed binding fails closed.
Enforced mode (opt-in)
Hooks that can say no.
Off by default. Turn it on with living-stack-team-verify plan --enforce, then install. It ships in the Team component (0.2.0-beta.2) of the Complete Local 0.3.2-beta.1 bundle.
Completion claims
Claude Code cannot end a turn by claiming the work is done unless a test run the hook itself watched passed after the last edit.
Destructive actions
Destructive commands and writes outside the project folder are blocked unless you authorized that exact action.
Bench result
On 13 recorded hook inputs, it blocked 10 of 10 risky cases and 0 of 3 safe actions. On the same inputs, the best free hook tools we tested blocked at most 3 of the 10 each.
What it does not stop: writes made from inside interpreter one-liners or scripts (python -c, node -e, a script file); tests weakened until they pass trivially; a stop allowed after 3 blocked attempts in a row (recorded as unverified, with a warning); anyone who can use the local installation key re-signing the evidence log. Assertion counting covers node:assert only; other runners get pass/fail checks. Codex support is covered by unit tests only, not by the bench. Like the rest of Living Stack, Enforced mode does not guarantee agent behavior.
Single-agent proof
Scope, authorize, budget, verify, recover.
Complete Local includes the full 15-tool lifecycle for evidence-bound execution and signed traces.
Team proof
Coordinate up to five installations.
The same bundle adds shared policy, cross-installation trace verification, backups, drift checks, and workspace export.