{# Reproducibility receipt + footnotes — closes the report. Notion-style callout with the exact scan fingerprint so a stakeholder can paste the command into a terminal and rerun the scan locally. Footnotes list the OWASP / CSA / ASI references the findings cited. #}
Reproducibility {{ package_version }}

Every value below is signed into scan.json. Copy the command to rerun the exact configuration; the evidence fingerprint is what an auditor can cross-check.

REPRODUCIBILITY

scan_id
{{ scan_id }}
seed
{{ rng_seed }}
guardian
{{ package_version }}
aivss
{{ aivss_formula_version }}
probes
{{ probe_library_version }}
target
{{ target_ref }}
evidence
{{ evidence_fingerprint }}
uv run agent-guardian scan \
  --target {{ target_ref }} \
  --tier {{ tier_label }} \
  --seed {{ rng_seed }}

References

  1. ¹ OWASP LLM01 — Prompt Injection. owasp.org · LLM Top 10
  2. ² OWASP LLM06 — Sensitive Information Disclosure.
  3. ³ MITRE ATLAS — Adversarial Threat Landscape for AI Systems. atlas.mitre.org
  4. Cloud Security Alliance — MAESTRO threat model.
  5. Agent Guardian Adversarial Surface Index v1 — internal policy catalogue.