Roadmap
How the research is sequenced: making automated vulnerability discovery safe by design. Each theme below links to the write-up where one exists. Status is a plain, honest snapshot — it changes as the thesis work moves.
Foundations — safety as architecture
Safety rails, not safety warnings
Why responsible automation has to be architectural: the constraints live in the system, not in a disclaimer. Read →
Authorization as a type, not a checkbox
Modelling "allowed to act" so it cannot be forgotten or bypassed. Read →
Scope encoding: allowlists the gate owns
Scope as gate-owned data, so discovery logic can never widen its own reach. Read →
Evidence & evaluation
A reproducible vuln lab you actually trust
Authorized, self-owned fixtures so results are repeatable and honest. Read →
Measuring railed discovery against fixtures
Metrics against known-answer fixtures, not hope. Read →
The audit log as evidence
A separate, tamper-evident record so findings can be verified without re-touching a target. Read →
The cost of being safe
Measuring the real trade-off between safety and effectiveness. Read →
In progress
Verification that fails closed
Confirming a finding only when the evidence clears a confidence floor, and treating anything uncertain as not confirmed. Write-up to follow.
Continuous evaluation harness
Growing the fixture corpus and the automated checks that keep the rails honest as the system changes.
Planned
Thesis write-up
Consolidating the rails, the evaluation method, and the safety/effectiveness trade-off into the Master's thesis.
Open method notes
Publishing reusable patterns for safe automation that others can adopt, kept strictly to public-safe, non-target-specific detail.
This roadmap is a living plan, not a promise of dates. Follow along on the blog or get in touch.