The scoreboard
Public numbers from both platforms. Most engagements are under NDA, so specifics are shared privately on request.
HackerOne
Verify on HackerOne- 61
- Bugs exterminated
- 6.45 / 7.0
- Signal
- 23.41
- Impact
- Clear
- Background check
YesWeHack
Verify on YesWeHack- 113
- Bugs exterminated
- 23.8
- Impact
- #115
- League table
- 2,332
- Points
I exterminate all common infestations
Every website, app and API has them. The only question is who finds them first: me, or somebody far less polite.
-
SQL and code injection
Structural pestEats straight through your data layer. Left untreated, brings the whole application down.
name=' OR 1=1 -- -
Broken access control
Door and lock specialistYour locks look fine, but half of them don't check who's holding the key. I test every door.
GET /admin 200 OK -
IDOR
Gets into other people's dataChange one number in a request, get another user's data. Astonishingly common; very quick to catch.
id=41id=42 -
Information disclosure
Leaks and seepageSecrets, internals and credentials seeping out of the foundations, then chained into worse.
GET /.env 200 OK -
Cross-site scripting
Swarms: stored, reflected, DOMOne unescaped field and it's in every visitor's browser. I trace each swarm to demonstrated account impact. No alert(1) and gone.
<img src=x onerror=…>
Now with a mechanical assistant
Same pair of hands on every job, now backed by AI tooling that covers more ground, faster. The machine does the legwork. A human does the hunting, the proving and the signing-off.
The machine
AI-assisted- Recon at scaleSubdomains, endpoints and forgotten assets mapped and triaged
- Reading the haystackJS bundles, API schemas and source skimmed for interesting sinks
- Variant ideasEvery confirmed bug brainstormed into its siblings
- Grunt workTooling, scripts and payload lists built on demand
The hunter
By hand- Business logicAuth flows, roles and workflows probed like a user who shouldn't be there
- Proof, not guessesEvery finding reproduced manually with a working PoC
- ChainingSmall leaks joined up into demonstrated, real impact
- Reports you can trustReviewed line by line; no hallucinated findings, ever
The machine finds the haystack. I find the needle.
The treatment plan
-
1 Survey
AI-assisted
The whole property mapped: every subdomain, endpoint and dusty corner surfaced and ranked.
-
2 Inspection
By hand
Hands-on testing of the logic, permissions and flows that scanners and models walk straight past.
-
3 Treatment
Hand and machine
Findings proven, chained and pushed to real impact, with AI hunting for variants of each.
-
4 Report
Human-verified
Clear steps to reproduce, honest severity and an obvious fix. Every word checked by me.
The Peckish guarantee
Human-verified, always.
AI makes me faster, but it doesn't make the calls. Every finding is reproduced by hand before it reaches you: no scanner spray-and-pray, no machine-written guesswork, no 40-page pile of duplicates. Application logic over noise, impact over volume, chained where possible and written up so the fix is obvious.
- HackerOne Clear Background checked
- Hand-checked Every report