AI-AGENT.CAM

Independent Cybersecurity Research

Vulnerability research • Bitcoin & Lightning • Protocol audit • Responsible disclosure

Security research on the systems that move money: cryptocurrency chains, wallets, and protocols — plus AI tooling built for security researchers. Every engagement is scoped, authorized, and evidence-driven.

About

AI-AGENT.CAM logo

Jacob F. Cudney

Bug-bounty handle: aragondragon • Ontario, Canada

I'm an independent security researcher specializing in cryptocurrency protocol and wallet security. My work focuses on finding real, demonstrable vulnerabilities in blockchain clients, wallet software, and the protocols that connect them — and proving each finding with reproducible evidence before it is reported.

I'm a credited contributor to Bitcoin Core: a fee-rate calculation flaw I reported was fixed upstream, with the fix crediting me by name. My research practice is built on responsible disclosure — findings go to the maintainers first, with working proof-of-concept evidence, and are made public only through the project's own process.

Alongside manual research, I develop AI-assisted tooling and reusable research skills that make audits faster and more systematic — the same tooling I bring to client engagements.

  • Cryptocurrency chain & protocol auditing
  • Wallet security review (software & signing flows)
  • Bitcoin & Lightning security research
  • AI tooling development for security research
  • Responsible disclosure, always

Proof — The Public Record

Much of my work is private, under disclosure, or credited through programs that don't publish researcher names. What follows is only what has been made public — verifiable at the source.

Bitcoin Core • Merged fix

Bitcoin Core PR #36712

A fee-rate (CFeeRate) flaw I identified and reported was fixed in Bitcoin Core. The fix was merged on November 2, 2026, and the pull request credits Jacob F Cudney (aragondragon) by name.

View the pull request on GitHub →

Scroll • Closed as completed

Scroll zkvm-prover Issue #261

A commitment-binding issue I raised against Scroll's zkvm-prover was closed as completed by the Scroll team, with thanks, confirming that production binds the commitments as the report described.

View the issue on GitHub →

Open source • Public research

openvm-halo2-verifier-l0

A public research repository from my zero-knowledge / verifier security work, published under my GitHub account for anyone to inspect.

View the repository on GitHub →

Services & Specializations

Any legitimate, scoped research engagement is possible. You provide the scope and the target — the repo, codebase, app, or website — and I provide methodical, evidence-backed research.

Chain & Protocol Auditing

Security auditing of cryptocurrency chains and protocols: consensus-adjacent logic, payment and fee handling, state transitions, and the edge cases where implementations diverge from specification — the class of flaw behind my Bitcoin Core credit.

Wallet Security Review

Auditing of wallet software and signing flows: key handling at the code level, transaction construction, address and change handling, update mechanisms, and the UI-to-signer path where users get misled.

Security Code Review

Focused or full review of a repo, codebase, application, or website you own or are authorized to have tested — vulnerability hunting with reproducible proof for every finding, not scanner noise.

AI Tools & Skills for Security Research

Development of AI-assisted tooling and reusable research skills: custom checkers, audit workflows, and agent-driven research pipelines tailored to your codebase or bug-bounty program.

Reverse Engineering — On Request, In Scope Only

Reverse engineering is available on request only when the target and purpose qualify under a defined, lawful scope — for example, your own software, interoperability, or defensive analysis. If it doesn't meet scope, it doesn't happen.

Responsible Disclosure Support

Findings are delivered as clear, filing-ready reports: impact, reproduction steps, proof-of-concept evidence, and remediation guidance — handled through proper disclosure channels.

The rule: no work starts without a written scope. The scope defines the target, the authorization, what's in and out, and the deliverables. It protects you and it protects the research.

Rates

Engagements are priced by scope, not by the hour spent guessing. All rates are in Canadian dollars; USD figures are approximate. Minimum engagement: CA$2,500.

Tier 1

Focused Review

Targeted Review — from CA$4,500

Focused scope, up to ~1,000 lines

  • One target: a single repo, module, app, or site
  • Defined vulnerability classes, defined window
  • Written findings report with reproduction evidence
  • Best for: a release check or a specific concern

Tier 2

Full Audit

Wallet / Codebase Audit — from CA$12,000

1–2 weeks · Protocol / Chain Audit from CA$30,000 — quote only

  • Whole-system review: chain, protocol, or wallet
  • Threat modelling plus deep manual audit
  • Proof-of-concept for every confirmed finding
  • Remediation guidance and re-test of fixes
  • Best for: launches, major upgrades, custody-adjacent code

Tier 3

Ongoing Research Partner

Research Retainer — CA$4,000/month

Continuing coverage, scoped monthly

  • Continuing research across releases
  • Custom AI research tooling built for your codebase
  • Priority response on new attack surface
  • Best for: live protocols and active wallet products

Rate card

  • HourlyCA$225 (~US$160)
  • DayCA$1,750 (~US$1,240)
  • WeekCA$7,500 (~US$5,320)
  • Minimum engagementCA$2,500
  • AI Security Tooling Buildfrom CA$5,000 — quote only
  • Scope assessmentCA$500 — credited against the engagement if it proceeds
  • Rush engagements+40%
  • Remediation re-check after fixes25% of the original fee

No written scope, no start. No work begins without a written scope. Send the target and what you're worried about; you'll get a scoped quote with a fixed deliverable before any work begins. Items marked quote only are priced from the scope — the floor above is the starting point, not a fixed price.

Boundaries — What I Don't Do

These are plain, permanent, and not negotiable. They are what make the work above trustworthy.

No malware. Ever.

I do not create, modify, weaponize, or deploy malware of any kind — no exceptions, no "for research" carve-outs on other people's systems.

No scope, no work

No engagement starts without a defined, lawful scope and clear authorization for the target. If you can't show you're authorized to have it tested, I can't test it.

Reverse engineering only in scope

Reverse engineering is done on request only, and only when it qualifies under a written scope — your own software, interoperability, or defense. Otherwise it's declined.

I never handle your keys

I do not take custody of, handle, or ask for seed phrases or private keys — yours or your users'. Wallet research is done on code and test environments, never on live keys.

FAQ

With a scope document. Email me the target — repo, codebase, app, or website — what you're concerned about, and confirmation that you're authorized to have it tested. You'll receive a scoped quote with fixed deliverables before any work begins.

A written report: each confirmed finding with impact, reproduction steps, and proof-of-concept evidence, plus remediation guidance. Unconfirmed suspicions are labelled as such — never padded into findings.

Yes — private codebases are normal. Provide access under your scope and authorization; the work and its findings stay confidential to you.

On request, and only when it qualifies under a defined, lawful scope — your own software, interoperability, or defensive analysis. Requests outside that are declined.

No. I don't create, modify, or deploy malware of any kind, and I don't build tooling to attack systems outside an authorized scope. That boundary is permanent.

Never. Wallet research is performed on code and test environments with test keys only. Anyone who asks for your seed phrase is not doing legitimate research.

Contact

Engagements begin with a scope document. Send the target and your concerns — repo, codebase, app, or website — and confirmation that you're authorized to have it tested.

aragondragon@ai-agent.cam

Jacob F. Cudney (aragondragon) • AI-AGENT.CAM • Ontario, Canada