Skip to content

Releases: EricFinland/proof

Release list

Proof v2.0.0

Proof v2.0.0 Latest
Latest

Choose a tag to compare

View all tags
@EricFinland EricFinland released this 12 Jun 12:42
v2.0.0
8998ac4

Major upgrade. Proof now closes the loop: it catches the lie, makes the agent fix it, and keeps score.

New

  • Honesty ledger + proof stats: every verification is recorded; see your agent's honesty rate, lies caught, clean streak, and worst-offending check.
  • proof check: verify any claim text without a transcript. Works from any coding agent, shell, or CI step. proof check "tests pass" --root . --json
  • .proof.toml: pin exact test/build/typecheck/lint commands per repo, set the serve command and base URL for http checks, tune max fix cycles. Config beats auto-detection; the claim's own command beats both.
  • HTTP v2: response body assertions (returns 200 with "ok"), automatic local server boot (.proof.toml serve, package.json dev/start, or Procfile) with clean process-tree teardown, and live verification of deploy claims.
  • Fix loop: after a FAIL, the agent is re-blocked with the previous receipts and "attempt N of 3" until it actually fixes the problem; Proof gives up gracefully after max cycles instead of nagging forever.
  • Runner detection for Maven, Gradle (wrapper aware), .NET, Make, Mix, Composer, and Bun/pnpm/yarn/npm by lockfile.
  • --json on verify/check/stats for CI pipelines; --version; Claude Code plugin manifest.

Numbers

  • 199 tests (was 71), CI green on Linux and Windows across Python 3.11-3.13
  • Still pure Python standard library, zero dependencies
  • Still strict: any failed claim fails the turn; missing tooling is INCONCLUSIVE, never a false PASS

Install

npx skills add EricFinland/proof
Assets 2
Loading

Proof v1.0.0

Proof v1.0.0

Choose a tag to compare

View all tags
@EricFinland EricFinland released this 12 Jun 11:07
v1.0.0
225ae98

First stable release.

Proof arms a Claude Code Stop hook that fires whenever the agent claims work is complete ("tests pass", "it works", "deployed"). An independent verifier then runs the real checks and returns a strict PASS/FAIL/INCONCLUSIVE verdict with command-output receipts. The agent can no longer self-certify.

Highlights

  • Automatic: arm once per project, zero configuration after that
  • 8 verifier strategies: tests, build, typecheck, lint, command, http, repro, filecheck
  • Strict verdict: any single failed claim fails the whole turn
  • Independent by construction: the verifier runs in a fresh context and trusts only execution output, never prose
  • Crash-safe hook with a per-claim recursion guard, verified once per session
  • Pure Python standard library, no dependencies
  • Cross-platform: CI green on Linux and Windows across Python 3.11-3.13
  • 71 tests, including an end-to-end acceptance test that reproduces the demo

Install

cd <your project>
python /path/to/proof/scripts/proof.py arm

See the README for the full demo, including a real-world catch where a project's "all tests pass" claim was false because the suite did not even import.

Loading