+1 888 373 0322ContactAUTHORIZED SECURITY VALIDATION
Reaper is available for assets the customer owns or is authorized to test. Active work begins only after the engagement documents targets, permissions, methods, operating windows, stop conditions and evidence handling — and every finding leaves evidence a defender can reproduce and a reviewer can re-check.
Passive review of what a surface already reveals to anyone — DNS, TLS, headers, robots, sitemap and exposed routes — without bypassing a single access control. This runs before authorization because it touches nothing it was not already shown; it maps the attack surface the way an adversary first sees it, and nothing here is active testing.
Within written scope: whether authorization holds, whether tenant isolation actually isolates, whether object access respects ownership, whether rate limits bind, and whether endpoints behave safely under abuse. The question is never whether a request succeeds — it is whether a request that should be refused is refused, proven with a reproducible case rather than asserted.
Client-provided or explicitly scoped source, dependency, secret, environment and build-pipeline review. The surface that leaks most quietly is the one nobody tests: a committed secret, an over-broad token, a build step that trusts its inputs. Each exposure is demonstrated against the actual repository state, not inferred from a checklist.
Prompt, tool, webhook, memory, source-leakage and approval-gate testing inside an authorized environment. An agent is evaluated as an actuator: the finding is what it can be made to do outside its jurisdiction — a tool it should not reach, an instruction smuggled through untrusted input, an approval gate that can be stepped around. As agents gain authority to act, this is the surface that matters most.
Named cloud exposure, routing, services and public misconfiguration within the approved target set. Third-party infrastructure stays out of bounds unless separately permitted in writing. Posture is observed and evaluated inside the boundary — what is reachable, what is exposed, what is misconfigured — and the boundary is honored in both directions.
Every finding carries its reproduction conditions, a severity, a containment or fix path, and the evidence a reviewer needs to confirm it independently. Observation is kept separable from interpretation, so a defender can walk from the raw result to the conclusion without taking the examiner's word for a step. Retest is bounded to the defined finding and its agreed remediation.
The hardest-scrutinized part of any validation offering is not the technique — it is the authority. Three rules govern every engagement before a single active probe runs.
A public request for a test confers no authorization. Reaper is available for assets the customer owns or is explicitly authorized to test, and active work begins only after the engagement records the targets, permissions, methods, operating windows, stop conditions and evidence handling. Anything not expressly included is denied by default.
If a target, a method, a timing window, an authority or an impact limit becomes ambiguous mid-engagement, active work stops until it is clarified in writing. A stop condition is honored the moment it is met. Ambiguity is treated as a halt, not as room to improvise.
The underlying evidence tooling operates read-only by construction — it scans whitelisted paths, generates receipts, and classifies files while confirming that no deploy, delete, network call or secret access occurred. Validation produces evidence; it is engineered not to change the system it is examining beyond what the authorization permits.
On authorization, an engagement runs as an ordered sequence from the authorization gate to retest closure. Each phase produces evidence a defender can use and a reviewer can re-check, so the work is accountable rather than performed on trust.
The surface is mapped, then gated against the written authorization. Reachability is never read as consent; no active work proceeds without recorded scope and a named restoring authority.
A finding is proven within the impact limits of the authorization and recorded so another operator can reproduce it, placed against a known adversary-technique matrix so a defender can act on it. The artifact carries method, not opinion.
Each finding carries its fix path and the evidence behind it — receipts, hashes, logs and replayable artifacts — so remediation starts from reproducible proof rather than a narrative.
A fix is confirmed by a scoped retest comparing before-and-after state, with the actor and the next review recorded. A residual live path keeps the engagement open; a ticket status is not closure.
The engagement does not authorize credential theft, persistence, credential stuffing, out-of-scope extraction or any testing beyond the written Rules of Engagement.
Discuss a security validation engagement