AI › Module 10 › Lesson 3
Evidence Pack
Bundle Topic 32 literacy evidence from `$AI_LAB` — LLM apps, secure dev, governance, findings; no stranger access.
Visual · t32_evidence_pack
Evidence pack = mentor handoff from $AI_LAB literacy. Original Cyberlium.
Opening
One indexed pack beats scattered AI notes — and never includes jailbreak kits or stranger app test exports.
Assemble evidence pack: `$AI_LAB` RoE, lab setup proof, injection/poisoning/threat labs if brief assigned, LLM app lab, secure dev lab, governance lab, findings lab, integrity statement (YOUR apps only, authorized red team, no jailbreak weaponization, no stranger app testing, limitations documented). Directory chmod 600 — index.md lists contents UTC redactions. Mentor copy: report + diagrams + eval stubs, not bypass repos. Gap paragraph for Topic 33 Threat Intelligence — separate authorized track. Scan folder for jailbreak kits, stranger app refs, prod keys — zero tolerance.
1. Pack contents
Minimum: RoE, LLM app, secure dev, governance, findings, integrity index.
Optional: injection/poisoning module labs — no refused technique files.
Command guide
Try these commands — Pack contents
═══ TOOLS & WEBSITES ═══ Browse / read these (authorized learning only — stay in YOUR lab / program scope)
NIST AI RMF — https://www.nist.gov/itl/ai-risk-management-framework MITRE ATLAS — https://atlas.mitre.org/ CISA AI — https://www.cisa.gov/ai
═══ INSTALL ═══
Linux (Debian/Ubuntu):
Command — copy this
sudo apt install python3
macOS:
Command — copy this
brew install python3
Windows: Download https://python.org/downloads/
═══ LINUX / macOS ═══
Command — copy this
export AI_LAB=${AI_LAB:-$HOME/cyberlium-lab/t32-ai}
python3 - <<'PY'
import os
lab = os.environ.get('AI_LAB', os.path.expanduser('~/cyberlium-lab/t32-ai'))
path = os.path.join(lab, 'notes', 'capstone-evidence-index.txt')
open(path, 'w').write('
'.join([
'CAPSTONE EVIDENCE INDEX — YOUR t32-ai lab only',
'Ethics: roe.txt, no-weaponized-ai.txt, responsible-ai-disclosure.txt',
'Basics: context-and-tools-literacy, trust-boundaries-checklist',
'Injection: direct/indirect named, defenses-checklist, echo_bot.py tests',
'Poisoning: data-poisoning-named, adversarial-ml-literacy, supply chain',
'Threats: phishing/deepfake/malware detection signals (no payloads)',
'OWASP: llm-top10-index, rag-plugins, output-handling checklists',
'Secure dev: sdlc checklist, eval-checklist, allow-deny-list',
'Governance: ethics checklist, ai-red-team-named, policy-and-roe',
'Findings: writeup template, severity-triage, findings-lab-index',
'Refusals: no malware/deepfake/jailbreak stranger SaaS/stranger API scrape',
]))
print(f'Wrote {path}')
PYCommand — copy this
grep -E 'Injection:|Refusals|OWASP:' "$AI_LAB/notes/capstone-evidence-index.txt"
Primary tools to practice this lesson: grep, python3. Reference sites: NIST AI RMF (https://www.nist.gov/itl/ai-risk-management-framework); MITRE ATLAS (https://atlas.mitre.org/); CISA AI (https://www.cisa.gov/ai). Run every command in the box — install first, then the usage lines — only on YOUR lab / program scope.
2. Index fields
Artifact name, module lesson, path, UTC date, description, redactions applied.
Checklist row ↔ index row accountability.
3. Integrity
$AI_LAB only, YOUR apps discipline, no stranger access, share-safe export.
Scan before seal — remove accidental prod keys or bypass kit files.
4. What you ship: AI Security evidence pack
Bundled deliverables + index + integrity — chmod 600, report not bypass kits default.
5. What you record before the next lesson
Evidence pack path.
6. Wrong vs right: stranger SaaS vs YOUR toy LLM
Worked failure — same MSF word, opposite target. Right never needs a café Wi-Fi or classmate laptop.
Wrong
Evidence pack includes jailbreak weaponization repo and stranger app scan export.
Right
Evidence pack assembled. Next: Capstone Lab.
Mission: assemble evidence pack
1) Bundle LLM app secure dev governance findings index. 2) Integrity statements. 3) Confirm zero refused technique files. 4) Gap note for Topic 33 Threat Intelligence.
Stuck? Ask Cyberlium AI Mentor
Ask Mentor: “Minimum index rows for capstone?”
Knowledge Check
APPLY: Evidence pack includes:
Multiple choice
Knowledge Check
APPLY: True or False: Bypass kit repos OK in pack.
True or False
Knowledge Check
APPLY: Integrity line states:
Multiple choice