Recommendations with a reason.
Cipher orders paths and labs using the interests you selected during onboarding and your own recent recorded practice. Recommendation cards explain the skill or activity signal behind the choice.
VulnOS / CipherCybersecurity tutor · Beta
Context first.Answers second.
Cipher is the learning layer inside VulnOS. Today it helps you choose what to practise, gives bounded help inside a lab, and answers questions about the platform and its published material. The full product is being built to connect goals, evidence, skills, and the next useful task.
Open Cipher Beta01Available now in Beta
A nudge before an answer.
Inside supported labs, Cipher works from the current task and reviewed teaching context. Authored hints remain staged so help can move you forward without quietly completing the exercise for you.
Questions that stay with the lesson.
Use the tutor for VulnOS learning questions when a suggestion or hint is not enough. Free and Premium accounts have separate daily allowances, shown inside Learn.
Your context stays yours.
The application selects a bounded context for the signed-in learner. The model does not receive direct access to the main VulnOS database, another learner's record, private payment data, or unrestricted author files.
02The honest Beta boundary
Useful help.
Visible limits.
Cipher Beta does not watch your live terminal, control your machine, grade hidden reasoning, or guarantee that every generated sentence is correct. It cannot see work you have not submitted to VulnOS.
Use the evidence in the lab as the source of truth. Cipher can refuse unsafe, unrelated, unsupported, or account-crossing requests. A refusal is part of the product boundary, not a missing feature.
Read the Learn AI notice01
No invisible access
Cipher receives selected context through the application. It does not browse private platform systems on its own.
02
No cross-learner memory
Recommendations and learning context are scoped to the signed-in account.
03
No silent completion
Tutor replies do not submit flags, unlock later hint stages, or award completion.
03The next four weeks
Build the evidence timeline.
Unify goals, onboarding choices, published content, attempts, task outcomes, and recent practice into a private learner timeline that can be rebuilt and corrected.
Ground every teaching move.
Strengthen version-aware retrieval from reviewed author material, keep later hints hidden until eligible, and attach useful source references when the answer depends on published content.
Choose the next useful task.
Turn activity into a cautious skill state, respect prerequisites and access, explain each recommendation, and offer remediation when the evidence shows a gap.
Prove it before release.
Run privacy, deletion, prompt-injection, quality, latency, and cost checks. Release through a controlled pilot only when ordinary learning remains correct during AI delays or outages.
This is the current engineering sequence, not a promise that every long-term Cipher capability will ship within four weeks. Scope can move when security, teaching quality, or measured reliability needs more work.
01
A living skill model
Confidence tied to reproducible evidence, prerequisites, recency, and reassessment instead of a single score.
02
Adaptive practice
Remediation, independent practice, transfer tasks, and readiness checks chosen from what the learner has actually demonstrated.
03
Auditable outcomes
Practical assessments and certificates backed by versioned content, reviewed criteria, and traceable evidence.
04The full product direction
A tutor that
can explain
its next move.
The final Cipher product is designed as a learning system, not a chat box. It will connect a learner's goal, current evidence, skill prerequisites, authorised content, and assessment history before deciding what to explain or recommend.
Later stages extend this foundation into specialised offensive, defensive, cloud, AI, and simulated OT learning. Team and enterprise capabilities come after learner isolation, quality, and assessment integrity are proven.
Start with the Beta.
Keep the evidence.
Use Cipher for direction and context, then prove the result inside the lab.