Skip to content

Repository files navigation

SK — Verifiable Execution CLI

Tamper-proof execution proofs for your scripts.

Run anything. Prove everything. Detect tampering instantly.

Install

pip install -e .

Or run directly:

PYTHONPATH=. python3 -m sk.cli run <script>

Usage

Generate a proof

sk run build.sh
sk run deploy.py
sk run test_suite.py

Output:

SK: executing build.sh
SK: command = bash build.sh
---
[your script output here]
---
SK: exit code = 0
SK: duration  = 0.014s
SK: script    = 9cf8eb6adaab4377...
SK: stdout    = da5cc1cd147a099d...
SK: proof     = f10f3aa670c6dcfc...

Proof generated: proof_1714430000.json

Verify a proof

sk verify proof_1714430000.json

Output:

SK Verify: proof_1714430000.json
  proof_id:   abb9c0fa-fd4b-4468-985d-1813f68d3e5b
  script:     /path/to/build.sh
  exit code:  0
  duration:   0.014112s
  checks run: 5

  All integrity checks passed.

VERIFICATION PASSED

Detect tampering

If anyone modifies the proof file, stdout, stderr, or the script itself:

  FAIL: proof_hash mismatch — proof file has been tampered with
  FAIL: stdout_hash mismatch — stdout has been modified

VERIFICATION FAILED

What's in a proof?

Each proof file contains:

Field Description
proof_id Unique UUID for this execution
script Absolute path to the executed script
script_hash SHA-256 of the script at execution time
stdout / stderr Captured output
stdout_hash / stderr_hash SHA-256 of each output stream
returncode Exit code
timestamp_start / timestamp_end Execution window
duration_seconds Wall-clock duration
environment Python version, OS, architecture, hostname
proof_hash SHA-256 seal over the entire proof

Verification checks

sk verify runs 5 integrity checks:

  1. Proof seal — proof_hash matches recomputed hash of all fields
  2. stdout integrity — stdout_hash matches actual stdout content
  3. stderr integrity — stderr_hash matches actual stderr content
  4. Script integrity — script_hash matches current script on disk (if available)
  5. Required fields — all mandatory fields are present

Use cases

  • CI/CD integrity — prove your build ran exactly as claimed
  • Audit trails — tamper-proof execution records for compliance
  • Reproducibility — capture environment + output for debugging
  • Supply chain security — verify build artifacts weren't modified

Author

John David Taylor Preston — Founder-Architect Safety Kernel / Aethel Grid / Undermoon OS / Civics OS

License

MIT

About

Fail-closed governance engine for AI safety

Topics

Resources

Code of conduct

Contributing

Security policy

Stars

1 star

Watchers

0 watching

Forks

Releases

Sponsor this project

Packages

Contributors

Languages