Human-controlled AI delivery · Built for production

Build with AI speed.
Ship with human-grade evidence.

ASEWAVE is a human-controlled AI software delivery method for solo developers and small teams building production systems where every claim must be verified, replayable, and backed by evidence.

Human-owned Adversarial verification Evidence-first Replayable audit trail Private engine
THE OPERATING MODEL Two AI agents that don't trust each other. HUMAN GATEKEEPER phase prompt approved spec cross-check REVIEWER / PROMPT AUTHOR Spec AI Writes phase prompts Reviews investigations Challenges claims Approves every gate EXECUTOR AGENT AG / Claude Code Implements & commits Captures artefacts Produces evidence EVIDENCE VERIFIER File hashes · Test output · No duplicate MD5s · Phase closes on human review Neither agent approves its own output. No phase closes without human sign-off.
Why ASEWAVE exists

AI agents can write code fast. They can also write confident fiction fast.

ASEWAVE exists because AI-generated progress should not be accepted as truth just because it is fluent. The method forces work through investigation, implementation, evidence capture, disk verification, and independent review before a phase closes.

The goal is not autonomous AI development. The goal is controlled AI acceleration with a replayable audit trail.

The method

Four principles. One audit trail.

Wi24RD develops XDRagon Monitor using ASEWAVE — a disciplined, human-owned, AI-assisted method where every phase starts with an investigation and ends with a walkthrough tied to verifiable evidence.

A
Augmented
AI increases development volume. Human ownership keeps direction, scope and judgment grounded.
SE
Solo Engineering
One developer owns the product, prioritization, architecture and every destructive decision.
AV
Adversarial Verification
Independent verification checks claims against code, commits and test evidence before changes are accepted.
+E
Evidence
Commit hashes, screenshots, test output and walkthroughs — not narrative claims alone.
PHASE RHYTHM Every phase starts with investigation. Every phase ends with evidence. PHASE A INVESTIGATE — Read the codebase — Verify assumptions — Produce investigation report No code written yet H HUMAN GATE PHASE B IMPLEMENT — Execute approved spec — Commit & capture artefacts — Save hashes + test output AG agent only — no scope changes H HUMAN GATE PHASE C VERIFY — Evidence Verifier checks claims — Walkthrough reviewed by human — Phase closes on human approval Nothing closes on narrative alone Phase Boundary Re-Read Protocol — At every gate, the executor re-reads the original prompt verbatim to prevent LLM drift.
Operating model

One human owns direction. Separate AI roles produce and challenge the work.

AI agents assist with specification, implementation and verification — but their claims must be checked against the actual codebase. Scope, prioritization and every destructive decision stay with the human.

Reviewer / Prompt Author

Turns intent into structured phase prompts. Reviews investigations, challenges claims not backed by evidence. Does not write production code.

Executor Agent

Verifies assumptions against the real repository, implements after approval, captures artefacts, and produces walkthrough evidence.

Human Gatekeeper

Holds scope, priorities, release decisions and every destructive operation. No phase closes without human approval.

ASEWAVE operating model diagram — interaction between the human, Cowork-Claude, and executor AI across phase A, B and C boundaries
ASEWAVE Operating Model — Phase A → B → C with human gate points
The four roles

Separation of concerns at every level.

ASEWAVE is role-based, not tool-specific. Specific products and models may change. The role separation should not — a single agent reviewing its own work optimises for confidence, not correctness.

01

Human Gatekeeper

Owns scope, priorities, architecture direction, commercial direction, destructive decisions, and release decisions. Reviews prompts before execution and outputs before merge.

Judgment density, not typing speed
02

Reviewer / Prompt Author

Converts intent into structured phase prompts. Defines investigation gates. Reviews Phase A before approving Phase B. Checks Phase C walkthroughs against requirements and evidence. Does not write production code.

Structure, review, verify — not execute
03

Executor Agent

Reads the prompt, inspects the actual repository, verifies assumptions, and stops after investigation. After human approval: implements, commits, runs tests, captures artefacts, and produces walkthrough evidence.

Implement, capture, evidence — not approve
04

Evidence Verifier

Checks that claims match primary evidence: file existence, non-empty artefacts, unique hashes, test output presence, schema verification, git history alignment, and documentation drift. The production verifier logic is part of the private engine.

Check claims — not generate them
Evidence discipline

Not just "it works." What changed, why, how it was tested.

The result is an audit trail by construction — what changed, why it changed, how it was tested, and what evidence supports the claim. Every walkthrough leads with disk reality that an external reviewer can replay independently.

Capture first. Artefacts are saved to disk before any prose is written.
Verify on disk. ls -la and md5sum outputs confirm existence and uniqueness.
§C.0 block mandatory. Walkthrough section C.0 contains verbatim terminal output — no exceptions.
14 red-flag checks. Automated review catches fabrication patterns before a phase closes.
§C.0 — Artefact verification
$ ls -la evidence/phase-c/
total 312
-rw-r--r--  1 ag  staff  142832  flow-login.png
-rw-r--r--  1 ag  staff    4218  pytest-output.txt
-rw-r--r--  1 ag  staff    1094  api-healthcheck.log

$ md5sum evidence/phase-c/*
7f9c2a14...  flow-login.png
3aa1dc88...  pytest-output.txt
91bd047f...  api-healthcheck.log

$ python -m pytest tests/ -v 2>&1 | tail -5
tests/test_auth.py::test_login PASSED
tests/test_api.py::test_health PASSED
✓ 47 passed in 3.42s
✓ zero duplicate MD5 hashes
✓ artefacts exist before prose
⚑ phase closes only on human review
Principles

What ASEWAVE is — and is not.

The methodology is built around controlled augmentation, adversarial separation, and mechanical proof. Understanding its limits is as important as understanding its strengths.

Human decision points at every gate

AI agents execute the human's strategy. No agent commits, closes a phase, or makes scope decisions autonomously.

Not autonomous AI development

ASEWAVE is not an agentic loop where AI runs until it decides it's done. The human green-lights every phase boundary.

Adversarial separation by design

Spec-author and executor are different agents with different roles. Neither trusts the other's claims without verification.

Not "AI-assisted coding"

That phrase means a human types with AI suggestions. ASEWAVE inverts it: AI types code, the human directs. The discipline is in the orchestration.

Replayable audit trail by construction

Commits, walkthroughs, screenshots, hashes and test outputs back every phase. An external reviewer can verify independently.

Not faster than a senior team

It scales one developer to mid-team velocity, not enterprise velocity. The human review bandwidth is deliberately preserved, not removed.

Fit

When ASEWAVE applies.

Best fit

  • Solo developer or two-person team building production software
  • Domain where audit trails matter — security tooling, regulated industries, infrastructure
  • Evidence-based release management is non-negotiable
  • Multi-month build where context preservation across sessions is critical
  • Mixed AI tier access — token-budgeted reviewer + larger executor

Less good fit

  • Pure prototyping where shipping ugly fast beats shipping correctly
  • Single-file scripts or one-off automations
  • Teams with many engineers and a preference for human-only review
  • Projects with no need for replayable audit trails
Public method

The method is public.

This site describes the public ASEWAVE principles: human ownership, adversarial separation, phase-gated delivery, evidence-first verification, and replayable audit trails.

The public method is enough to understand and apply ASEWAVE in spirit.

Private engine

The engine is private.

The operational engine remains private while it is refined in active XDRagon development. That private layer includes prompt contracts, verifier logic, memory structure, phase templates, red-flag checks, handoff conventions, routing rationale, and recovery patterns.

The private engine is what makes the workflow repeatable, fast, and commercially deployable at production depth.

You don't need the engine to start. Copy the whitepaper and paste it into any AI — ChatGPT, Claude, Gemini, Copilot, or similar — and ask it to design a phase prompt structure for your specific tools. The reviewer and executor roles can be filled by whatever you already have access to: a browser-based AI as the brain, a VSCode-based coding agent as the worker.

Current status
In production use

ASEWAVE is currently used in the production development of XDRagon. The methodology is documented publicly for reference, discussion, and attribution.

The private ASEWAVE Engine is not currently released. Commercial access, partnerships, or reference implementations may be considered later — but there is no public support, training, or implementation package at this time.

Common questions

What developers and teams ask about ASEWAVE.

ASEWAVE (Augmented Solo Engineering With Adversarial Verification + Evidence) is a human-controlled AI software delivery methodology. It enables solo developers and small teams to build production-grade software at multi-engineer pace by pairing one human gatekeeper with two adversarial AI agents — one that implements and one that independently verifies. Every phase must produce evidence before it closes.

ASEWAVE structures every unit of work into four phases: Investigation, Implementation, Evidence Capture, and Verification. The human sets direction and owns phase gates. An Executor AI agent builds the work. A separate Verifier AI agent independently challenges every claim. No phase closes until evidence — terminal output, file hashes, test results — exists on disk and survives independent review.

ASEWAVE is for solo developers, tech leads, and CTOs building production systems with AI assistance where reliability, auditability, or compliance matters. It fits domains like security tooling, regulated industries, and infrastructure — where AI-generated progress cannot be accepted without independent verification.

Vibe-coding accepts AI output if it looks correct. ASEWAVE requires that every AI-generated claim is verified against primary evidence before a phase closes. The human never delegates judgment — only execution. Adversarial separation means the AI that builds the work is never the same AI that approves it.

A phase gate in ASEWAVE is a mandatory checkpoint between work phases. It only opens when specific evidence exists: file existence confirmed, output hashes unique, tests passed, terminal logs captured, and a structured walkthrough completed. The gate prevents AI-fabricated progress from passing as real delivery.

Adversarial verification means the AI role that builds code is structurally separated from the AI role that reviews and challenges it. In ASEWAVE, the Executor Agent implements; the Verifier Agent independently checks claims against disk evidence, git history, and documentation. Neither AI approves its own work.

Yes. The ASEWAVE methodology is model-agnostic. The public method can be applied with any AI coding assistant. The private ASEWAVE Engine — which implements the full prompt contracts, verifier logic, and phase templates — is currently closed and not publicly released.

To get started with your own setup: copy the whitepaper at asewave.org/whitepaper.html and paste it into any AI assistant — ChatGPT, Claude, Gemini, Copilot, or similar — and ask it to design a concrete phase prompt structure for your specific tools. Tell it which AI you'll use as reviewer and which as executor. The method works with any pairing: a browser or desktop AI as the brain, any VSCode-based coding agent as the worker.

The ASEWAVE methodology is publicly documented and free to use with attribution. The operational engine — prompt contracts, verifier logic, memory structure, phase templates, and red-flag checks — is private and not currently released. See asewave.org for the full public method and whitepaper.

Phase A (Investigation) typically takes 10–30 minutes of AI time and 5–10 minutes of human review. Phase B (Implementation) scales with task complexity — a focused feature or fix may take 20–60 minutes. Phase C (Verification) usually takes 15–30 minutes of AI time plus a structured human walkthrough review. The human is not waiting for AI to finish — they are reviewing, deciding, and gating. The discipline cost is in the review time, not the elapsed clock.

The executor agent must never expand scope beyond the approved phase prompt, approve its own output, close a phase unilaterally, or delete or overwrite evidence files. It must also not proceed to Phase B before Phase A has been reviewed and approved by the human, and must not declare a task complete without running and capturing real test output. Any action that bypasses the human gate is a protocol violation — regardless of whether the result looks correct.

Language models in long sessions can subtly reinterpret their original instructions — optimising for what seems reasonable rather than what was specified. The Phase Boundary Re-Read Protocol counters this by requiring the executor agent to re-read the original phase prompt verbatim at the start of each new phase, before any other context. This hard-anchors execution to the approved specification rather than to accumulated conversation context, which may have drifted through clarifications, digressions, or extended tool use.

Contact

Partnerships and serious inquiries.

ASEWAVE is a methodology, not a product — there is no support channel. For questions about the method, copy the whitepaper into ChatGPT or Claude and ask directly; it works well. This form is for genuine partnerships, professional collaboration, or matters that truly require direct contact.

Method question? Copy the whitepaper into ChatGPT or Claude — it can walk you through the full setup. No account or contact needed.

Submissions are private. No account required.

Message received.

Wi24rd-Com reviews every submission personally and will reply to your email if you left one.

ASEWAVE — Augmented Solo Engineering With Adversarial Verification + Evidence
Coined by wi24rd.com during the XDRagon development process.
The methodology was refined and documented with AI-assisted review, including Claude as a methodology collaborator.
Attribution appreciated where the term is used.