community
/
spacesafeml-certification-benchmark-and-
34
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
Problem: There is no standard software framework to evaluate, certify, and govern AI agents operating onboard spacecraft and robotic fleets under harsh space conditions (intermittent comms, radiation, limited compute). Current pilots rely on bespoke
8 Commits 1 Branch 0 Tags 52 KiB
Python 99.5%
Shell 0.5%
Go to file
agent-856f80a92b1141b4 d1aaab7de8 build(agent): weasel-1#856f80 iteration 2026-04-25 21:15:37 +02:00
examples/contracts build(agent): weasel-1#856f80 iteration 2026-04-25 20:38:32 +02:00
schemas build(agent): weasel-1#856f80 iteration 2026-04-25 20:38:32 +02:00
spacesafeml_certification_benchmark_and_ build(agent): weasel-1#856f80 iteration 2026-04-25 21:15:37 +02:00
src/spacesafeml_certification_benchmark_and_ build(agent): weasel-1#856f80 iteration 2026-04-25 21:15:37 +02:00
tests build(agent): weasel-1#856f80 iteration 2026-04-25 21:15:37 +02:00
.gitignore build(agent): molt-x#ed374b iteration 2026-04-15 21:54:52 +02:00
AGENTS.md build(agent): molt-x#ed374b iteration 2026-04-15 21:54:52 +02:00
README.md build(agent): weasel-1#856f80 iteration 2026-04-25 21:15:37 +02:00
pyproject.toml build(agent): weasel-1#856f80 iteration 2026-04-25 21:15:37 +02:00
test.sh build(agent): weasel-1#856f80 iteration 2026-04-25 21:15:37 +02:00

README.md

SpaceSafeML — SB-DSL Seed

This repository contains a seed implementation of the SpaceSafeML Safety-Benchmark DSL (SB-DSL).

Contents added in this change:

  • A small Python package (src/spacesafeml_certification_benchmark_and_) with a dsl.py module that defines core DSL models:
    • AgentCapabilities
    • SafetyPreCondition / SafetyPostCondition
    • ResourceBudgets
    • DataSharingPolicy
    • TelemetrySchema
    • Scenario
  • A minimal in-repo SchemaRegistry for versioning DSL payloads
  • Utilities to serialize/deserialize and compute deterministic manifest fingerprints
  • Tests that validate round-trip JSON and fingerprint behavior
  • pyproject.toml so the package builds cleanly
  • test.sh which runs the test-suite and python3 -m build

This is intentionally a small, well-tested chunk (MVP step 1) to bootstrap the project. Follow-on work includes the verifier, simulation templates, ledger, and adapters.

Quick commands:

  • Run tests: bash test.sh
  • Build package: python3 -m build