Skip to content

armgit / rea

Public Imported from GitHub
0

Reverse engineer anything with agents, from app behavior down to native binaries.

docs(website): write the Touhou reconstruction essay

N0zoM1z07df205060655
diff --git a/website/AGENTS.md b/website/AGENTS.md
index a737f196f..e97181ec6 100644
--- a/website/AGENTS.md
+++ b/website/AGENTS.md
@@ -15,8 +15,10 @@ for asset notes and the source of case-study claims.
- Show what REA returns and how the agent uses it. Preserve source attribution,
target identity and verification scope.
- Blog articles teach methods, decisions and workflows through concrete cases.
- Keep the thesis and overview figure clear, and use engine details only when
- they explain a decision. Label working outlines and mark them `noindex`.
+ Lead with the takeaway. Essays can develop a personal argument through
+ experience; figures are optional and should add an explanation the prose
+ needs. Use engine details only when they explain a decision. Label working
+ outlines and mark them `noindex`.
- Keep personal paths, account data, credentials and raw captures outside the
public assets. Use generic inputs and label shortened display paths.
- Follow the same clear writing standard in search titles and descriptions.
diff --git a/website/README.md b/website/README.md
index 3c8caa5c7..bd3ad1f10 100644
--- a/website/README.md
+++ b/website/README.md
@@ -43,7 +43,7 @@ HTML files. No URL list needs to be maintained.
- `public/first-investigation/index.html`: a guided Notes export investigation, from setup to a checked CSV prediction.
- `public/showcase/index.html`: the case-study index.
- `public/blog/index.html`: articles about reconstruction methods, ports, mods and reverse engineering.
-- `public/blog/touhou-reconstruction/index.html`: the working English outline for a Touhou reconstruction article, with TH08, TH095 and TH04 timelines.
+- `public/blog/touhou-reconstruction/index.html`: a personal essay about agent-driven reconstruction, knowledge that compounds across TH08 and TH095, and the workflow's applicability to PC-98 TH04.
- `public/showcase/aegis/index.html`: Aegis's Android login-code calculation, with an adjustable clock and reference checks.
- `public/showcase/dx-ball/index.html`: sound-pan investigation and project status.
- `public/showcase/notion/index.html`: Notion's Electron clipboard bridge and rich clipboard format.
@@ -123,14 +123,18 @@ and Cloudflare's [managed robots behavior](https://developers.cloudflare.com/bot
## Content
-Blog articles explain methods and decisions across a project. Use a clear thesis,
-an overview figure, a practical workflow and a few cases that test the method.
-Case studies remain focused on one inspected behavior. Keep engine details in
-supporting references unless they explain a decision the reader needs to make.
-
-The Touhou article is currently a working outline, marked `noindex`. Replace it
-with the finished article before removing that directive. Its timeline provenance
-and counting rules are in [evidence/touhou-reconstruction-method.md](evidence/touhou-reconstruction-method.md).
+Blog articles explain methods, decisions and ideas through project experience.
+Lead with a takeaway, then develop the argument through concrete examples.
+An essay can explore a change in perspective without becoming a step-by-step
+guide. Add a figure when it contributes a useful explanation; an opening flow
+is optional. Case studies remain focused on one inspected behavior. Keep engine
+details in supporting references unless they explain a relevant decision.
+
+The Touhou article is a full English draft for local review. Its provenance,
+quoted README passage and milestone counting rules are in
+[evidence/touhou-reconstruction-method.md](evidence/touhou-reconstruction-method.md).
+The completed text has normal search metadata and is included in the generated
+sitemap. Its visible review label can be removed when the draft is finalized.
Keep the copy direct and specific. Explain the task and the result before listing
tool names. Setup commands and runtime requirements should match the released
diff --git a/website/evidence/touhou-reconstruction-method.md b/website/evidence/touhou-reconstruction-method.md
index 111e33fa9..a891d4227 100644
--- a/website/evidence/touhou-reconstruction-method.md
+++ b/website/evidence/touhou-reconstruction-method.md
@@ -1,6 +1,6 @@
# Touhou reconstruction method: evidence notes
-Checked on 10 October 2026. This document supports the working article outline
+Checked on 10 October 2026. This document supports the full English essay
at `public/blog/touhou-reconstruction/index.html`.
## Method sources
@@ -13,7 +13,7 @@ and [cross-game knowledge](https://github.com/N0zoM1z0/touhou-reconstruction-fac
documents for the underlying method: autonomous investigation, bounded changes,
empirical feedback, durable checkpoints and reusable lessons.
-The proposed industrial-age analogy explains that method. The Factory's
+The industrial-age analogy explains that method. The Factory's
software architecture also uses the Abstract Factory pattern; the article's
analogy should focus on the working process rather than that implementation.
@@ -47,6 +47,9 @@ rate or an executable-wide completion percentage.
## Public history snapshot
+This is the initial 10 October research snapshot. The later TH04 phase update
+below uses its own pinned checkpoint; it does not change these earlier counts.
+
| Project | Default-branch commits | Commits whose subject begins `gpt-web:` | Public head |
| ------- | ---------------------: | --------------------------------------: | ------------------------------------------ |
| TH08 | 1,117 | 400 | `d2a00f4acb4a13916e07a00bd13f714961714628` |
@@ -119,8 +122,54 @@ perfectly from the beginning.
## Editorial scope
-The working article teaches goals, task design, feedback loops, project memory,
-handoffs and cross-game reuse. Engine APIs, compiler flags, addresses and ledger
-schemas belong in linked supporting documentation. Planned port/mod examples
-must be backed by their own source and checks before becoming finished claims.
-The outline is labeled and `noindex`; the shared sitemap therefore excludes it.
+The article follows the author's change in perspective: previous manual RE,
+starting agent-driven work in August 2026, TH08's continuation, TH095's reuse of
+experience, and TH04's application of the method to another architecture. The
+manual-RE background and August starting point are author-provided experience.
+The industrial-era conclusion and the interpretation of the two contribution
+models are the author's argument, supported by the described project record.
+They are not a measured agent-versus-human performance ratio.
+
+Engine APIs, compiler flags, addresses and ledger schemas remain in references.
+The full draft replaces the outline and removes its `noindex` directive; the
+generated sitemap includes the article. Its visible draft label identifies
+the current local review state. No production publication was requested.
+
+## GensokyoClub quotation and policy
+
+The public README was read with `gh api` on 10 October 2026. Its latest
+README-changing commit is
+[`e874b98e`](https://github.com/GensokyoClub/th08/blob/e874b98e210b3be1ed21c5e03d54b883eac5069c/README.md#important-notice),
+dated 6 September. The notice announces a decision effective 5 September and
+describes concerns about AI decompilations, ports, attribution and their effect
+on the maintainers. Its stated plan is private development until completion.
+
+The article quotes 23 words from the passage's opening clause, followed by an
+ellipsis; it paraphrases the remaining point about the psychological toll.
+The excerpt was checked against the retrieved original, with whitespace
+normalized for display. The linked notice supplies its context.
+
+The current public head is
+[`f345c7de`](https://github.com/GensokyoClub/th08/tree/f345c7dee07b849f2ff93a0029ab5ede126bf12c).
+Its [contribution policy](https://github.com/GensokyoClub/th08/blob/f345c7dee07b849f2ff93a0029ab5ede126bf12c/CODE_OF_CONDUCT.md)
+excludes PRs created primarily using AI. The article paraphrases this policy
+and discusses the different contribution models. It preserves GensokyoClub's
+credit, describes the author's own manual and agent workflows, and does not
+assign motives beyond what the public notice states.
+
+## TH04's current DOS and 64-bit phases
+
+The latest public main checkpoint checked for this revision is
+[`bb9faca5`](https://github.com/N0zoM1z0/th04/blob/bb9faca54e2eb3c0d6c757c75d59ac29d2cd5e93/docs/RE_HANDOFF.md),
+dated 10 October 2026. Its handoff records the standalone DOS products,
+repaired normal and invincible variants, and the user's manual Windows
+observations of complete Normal routes, endings and saves. It identifies
+resumed x64 work as the current phase on the separate `port/modern-64` branch.
+That branch was publicly present at `1ffc53a61e1e82409adb40c7658d4ee2f3a21916`
+when inspected with `gh`.
+
+The essay's first-person DOS testing statement describes those maintainer
+observations, also confirmed by the author in this revision request. It is
+not attributed to a new automated website test or a fresh runtime replay.
+The 64-bit port is described as in development. The earlier four-product build
+and function-count milestones retain their historical scopes.
diff --git a/website/figures.md b/website/figures.md
index 231b41afe..534d8ac24 100644
--- a/website/figures.md
+++ b/website/figures.md
@@ -94,13 +94,12 @@ APK's selected Java and an explanatory summary.
APK, methods, source attribution and reconstruction checks. Keep the block
formula, default settings, slider outputs and reference cases consistent.
-## Touhou reconstruction article outline
+## Touhou reconstruction essay
-The proposed opening figure is a semantic HTML flow using the shared components:
-inspect/reconstruct → compile/compare/run → code/evidence/next task. Its caption
-names the human, agent, analysis and verification roles. A failed check returns
-the batch to investigation; recorded lessons feed later work. The project
-timelines use HTML dates and pinned public milestones, documented in
+The essay opens with a takeaway and develops its argument in prose. It has no
+opening flow figure. Project dates and the attributed README excerpt support
+the narrative; detailed milestone references are available in a native
+disclosure. Their provenance is documented in
[evidence/touhou-reconstruction-method.md](evidence/touhou-reconstruction-method.md).
## Original image generation prompts
diff --git a/website/public/assets/styles.css b/website/public/assets/styles.css
index a30f1e87d..bf52622c9 100644
--- a/website/public/assets/styles.css
+++ b/website/public/assets/styles.css
@@ -680,28 +680,26 @@ pre code {
.article .stats dd {
color: var(--muted);
}
-.blog-thesis {
+.blog-takeaway {
max-width: 730px;
- padding-left: 1rem;
+ margin-top: 1.8rem;
+ padding-left: 1.25rem;
border-left: 2px solid var(--accent);
+}
+.blog-takeaway .eyebrow {
color: var(--accent);
}
-.blog-timeline {
- padding: 0;
- margin-block: 1.5rem;
- list-style: none;
+.blog-takeaway p:last-child {
+ margin-bottom: 0;
}
-.blog-timeline li {
- display: grid;
- grid-template-columns: 4.5rem minmax(0, 1fr);
- gap: 1rem;
- padding-block: 0.65rem;
- border-bottom: 1px solid var(--line);
+.blog-article blockquote {
+ margin: 1.6rem 0;
+ padding-left: 1.25rem;
+ border-left: 2px solid var(--line);
+ font-size: 1.05rem;
}
-.blog-timeline time {
- color: var(--accent);
- font-weight: 600;
- font-variant-numeric: tabular-nums;
+.blog-article blockquote p {
+ color: var(--ink);
}
.caption {
font-size: 0.875rem;
diff --git a/website/public/blog/index.html b/website/public/blog/index.html
index 47196ea76..0a6148544 100644
--- a/website/public/blog/index.html
+++ b/website/public/blog/index.html
@@ -122,18 +122,20 @@
</div>
<div class="guide-index">
<section class="guide-entry">
- <p class="eyebrow">Game reconstruction<br />Article outline</p>
+ <p class="eyebrow">Game reconstruction<br />Essay</p>
<div>
- <h2>Reconstructing Touhou with AI agents</h2>
+ <h2>The industrial revolution in reverse engineering</h2>
<p>
- A repeatable workflow, illustrated by TH08, TH095 and TH04: build
- on existing work, carry lessons into a new game, and adapt the
- method to an older platform.
+ What manual RE and reconstructing TH08, TH095 and TH04 taught me
+ about agents, verification and knowledge that compounds across
+ projects.
</p>
<a class="text-link" href="touhou-reconstruction/"
- >Read the article outline</a
+ >Read the article</a
>
- <p class="page-meta">By N0zoM1z0 · Working outline</p>
+ <p class="page-meta">
+ By N0zoM1z0 · 10 October 2026 · Draft for review
+ </p>
</div>
</section>
</div>
diff --git a/website/public/blog/touhou-reconstruction/index.html b/website/public/blog/touhou-reconstruction/index.html
index a4f450ab2..82660a44e 100644
--- a/website/public/blog/touhou-reconstruction/index.html
+++ b/website/public/blog/touhou-reconstruction/index.html
@@ -3,10 +3,10 @@
<head>
<meta charset="utf-8" />
<meta name="viewport" content="width=device-width, initial-scale=1" />
- <title>Reconstructing Touhou with AI Agents — REA</title>
+ <title>Reverse Engineering’s Industrial Revolution — REA</title>
<meta
name="description"
- content="An article outline about repeatable game reconstruction: project memory, small verified changes and lessons from TH08, TH095 and PC-98 TH04."
+ content="N0zoM1z0 reflects on manual reverse engineering, Touhou reconstruction with AI agents, and how knowledge compounds across TH08, TH095 and PC-98 TH04."
/>
<link
rel="canonical"
@@ -16,11 +16,11 @@
<meta property="og:site_name" content="REA" />
<meta
property="og:title"
- content="Reconstructing Touhou with AI Agents — REA"
+ content="Reverse Engineering’s Industrial Revolution — REA"
/>
<meta
property="og:description"
- content="An article outline about repeatable game reconstruction: project memory, small verified changes and lessons from TH08, TH095 and PC-98 TH04."
+ content="N0zoM1z0 reflects on manual reverse engineering, Touhou reconstruction with AI agents, and how knowledge compounds across TH08, TH095 and PC-98 TH04."
/>
<meta
property="og:url"
@@ -40,11 +40,11 @@
<meta name="twitter:card" content="summary_large_image" />
<meta
name="twitter:title"
- content="Reconstructing Touhou with AI Agents — REA"
+ content="Reverse Engineering’s Industrial Revolution — REA"
/>
<meta
name="twitter:description"
- content="An article outline about repeatable game reconstruction: project memory, small verified changes and lessons from TH08, TH095 and PC-98 TH04."
+ content="N0zoM1z0 reflects on manual reverse engineering, Touhou reconstruction with AI agents, and how knowledge compounds across TH08, TH095 and PC-98 TH04."
/>
<meta
name="twitter:image"
@@ -54,7 +54,6 @@
name="twitter:image:alt"
content="REA — Find out how software works. Reverse engineering with your coding agent."
/>
- <meta name="robots" content="noindex" />
<meta name="theme-color" content="#ffffff" />
<link rel="icon" href="../../assets/favicon.svg" type="image/svg+xml" />
<link rel="stylesheet" href="../../assets/styles.css" />
@@ -124,457 +123,506 @@
<a href="../../">Home</a> / <a href="../">Blog</a> / Touhou
reconstruction
</p>
- <p class="eyebrow">Game reconstruction · Article outline</p>
- <h1>Reconstructing Touhou with AI agents.</h1>
- <p class="lede">
- From individual expertise to a repeatable workflow. Lessons from
- continuing TH08, starting TH095, and rebuilding TH04 for PC-98.
+ <p class="eyebrow">Essay · Touhou reconstruction</p>
+ <h1>The industrial revolution in reverse engineering.</h1>
+ <p class="page-meta">
+ By N0zoM1z0 · 10 October 2026 · Draft for review
</p>
- <p class="page-meta">By N0zoM1z0 · Working outline · 10 October 2026</p>
- <p class="blog-thesis">
- <strong
- >The unit of progress is a checked change that the next agent can
- build on.</strong
- >
- </p>
- </div>
- <figure class="workflow-figure">
- <p class="figure-title">
- The factory idea, in one reconstruction batch
- </p>
- <ol class="flow-steps">
- <li>
- <span class="flow-label">01 · One question</span
- ><strong class="flow-node">Inspect → reconstruct</strong
- ><span class="flow-note"
- >The agent uses analysis tools such as REA to collect evidence,
- then writes source for one behavior or subsystem.</span
+ <div class="blog-takeaway">
+ <p class="eyebrow">Takeaway</p>
+ <p>
+ <strong
+ >Agent-driven reverse engineering changes how the work can
+ accumulate.</strong
>
- </li>
- <li>
- <span class="flow-label">02 · A tested change</span
- ><strong class="flow-node">Compile → compare → run</strong
- ><span class="flow-note"
- >Project checks test the change against the original code and the
- reconstructed program.</span
- >
- </li>
- <li>
- <span class="flow-label">03 · A saved checkpoint</span
- ><strong class="flow-node">Code + evidence + next task</strong
- ><span class="flow-note"
- >The repository keeps the result and the lesson. A fresh agent can
- continue from this state.</span
- >
- </li>
- </ol>
- <p class="flow-result">
- A failed check sends the batch back to investigation. A useful lesson
- improves the next batch—and the next game.
- </p>
- <figcaption>
- The human sets the goal, priorities and release decisions. The
- investigation remains flexible; the checks and handoff are repeatable.
- This is the proposed opening figure for the article.
- </figcaption>
- </figure>
+ An investigation can become a checked result, a failed experiment
+ can become a reusable test, and one project's experience can become
+ the next project's starting point.
+ </p>
+ <p>
+ Reconstructing TH08, TH095 and TH04 convinced me that RE is entering
+ its industrial age.
+ </p>
+ </div>
+ </div>
<div class="article-layout">
<nav class="contents" aria-label="On this page">
- <p>Article outline</p>
+ <p>In this article</p>
<ul>
- <li><a href="#goal">1. Choose the goal</a></li>
- <li><a href="#industrial-shift">2. The industrial shift</a></li>
- <li><a href="#th08">3. Continue TH08</a></li>
- <li><a href="#th095">4. Start TH095</a></li>
- <li><a href="#th04">5. Adapt to PC-98</a></li>
- <li><a href="#workflow">6. The working loop</a></li>
- <li><a href="#ports-and-mods">7. Ports and mods</a></li>
- <li><a href="#start">8. Your first batch</a></li>
+ <li><a href="#august">August 2026</a></li>
+ <li><a href="#th08">TH08: the continuation</a></li>
+ <li><a href="#different-approaches">Two approaches</a></li>
+ <li><a href="#th095">TH095: compounding</a></li>
+ <li><a href="#th04">TH04: across architectures</a></li>
+ <li><a href="#industrial-shift">The industrial shift</a></li>
+ <li><a href="#possibilities">What becomes possible</a></li>
</ul>
</nav>
- <article
- class="article"
- aria-label="Touhou reconstruction article outline"
- >
- <section class="article-section" id="goal">
- <h2>1. Decide what you want to build</h2>
- <p>
- Open with a practical question: do you want to understand a game,
- rebuild it, move it to another platform, or change how it plays?
- </p>
- <ul>
- <li>
- <strong>Reconstruction:</strong> recover source and check it
- against the original program.
- </li>
- <li>
- <strong>Port:</strong> make that source work with a new
- platform’s rendering, input, audio and files.
- </li>
- <li>
- <strong>Mod:</strong> understand a behavior, change it
- deliberately, then test the result.
- </li>
- </ul>
- <p class="caption">
- Planned figure: one recovered codebase branches into a faithful
- build, a port and a mod. Each branch names its own success check.
+ <article class="article" aria-label="Touhou reconstruction essay">
+ <section class="article-section" id="august">
+ <h2>What changed in August 2026</h2>
+ <p>
+ Before this reconstruction work, I had done a lot of manual
+ reverse engineering. The familiar loop was to inspect
+ instructions, follow references, form an explanation, test it, and
+ revise it. Progress depended on how much of that loop I could
+ carry out myself, and how well I could keep the growing picture in
+ my head.
+ </p>
+ <p>
+ In August 2026, I began exploring agent-driven RE seriously. I
+ wanted to see how far an agent could go when it had access to the
+ program, ordinary development tools, and feedback from real
+ checks. Touhou reconstruction became the place where I could
+ pursue that question through a substantial project.
+ </p>
+ <p>
+ The early progress was startling. Tasks that I would have budgeted
+ around my own time and attention could keep moving while the agent
+ investigated, tried source formulations, compiled them, and
+ examined the result. My work increasingly involved choosing the
+ direction, improving the tools and checks, and deciding what the
+ accumulated evidence meant.
+ </p>
+ <p>
+ That changed which projects seemed reasonable to attempt. A large
+ executable began to look like a collection of investigations we
+ could organize and keep advancing. Recovering source, making it
+ readable, and eventually using it for a port became parts of an
+ achievable project.
+ </p>
+ <p>
+ The numbers first got my attention. The more interesting question
+ was what made that progress possible—and whether the process could
+ improve as it worked.
</p>
</section>
- <section class="article-section" id="industrial-shift">
- <h2>2. Turn expertise into a repeatable process</h2>
- <p>
- The industrial-age analogy: agents do the work, tools let them
- inspect the program, checks test the result, and the repository
- preserves what they learned.
- </p>
- <ul>
- <li>
- <strong>Make the task small enough to check.</strong> One
- behavior or connected subsystem has a clear input, output and
- test.
- </li>
- <li>
- <strong>Make feedback fast.</strong> A compile error, mismatch
- or runtime failure tells the agent what to investigate next.
- </li>
- <li>
- <strong>Make experience reusable.</strong> A lesson becomes a
- script, test, evidence note or handoff that survives the
- conversation.
- </li>
- </ul>
+ <section class="article-section" id="th08">
+ <h2>TH08: building on human work</h2>
+ <p>
+ <a href="https://github.com/N0zoM1z0/th08">TH08</a>, Imperishable
+ Night, began for me as a continuation of
+ <a href="https://github.com/GensokyoClub/th08"
+ >GensokyoClub's reconstruction</a
+ >. Their public work already contained source, names, build
+ knowledge and a contribution history. It gave the continuation a
+ foundation. That foundation matters when discussing everything
+ that followed.
+ </p>
+ <p>
+ I preserved the imported history through the public 10 August
+ checkpoint. The independent continuation began on 13 August. By 19
+ August, the project's ledger recorded source for all 1,107
+ identified game functions. A playable Linux reconstruction port
+ was committed on 24 August, roughly eleven days after the
+ continuation began. A Web edition followed, then a native Linux
+ 64-bit release on 30 August.
+ </p>
+ <p>
+ Those dates made the change in pace tangible. What interested me
+ most was that the work extended beyond producing plausible C++. It
+ included comparisons, build repairs, recovery of meaningful names
+ and types, and the work needed to turn reconstructed pieces into a
+ program people could run.
+ </p>
+ <p>
+ This also exposed the importance of feedback. A function can look
+ correct and compile correctly while its place in the whole program
+ is wrong. Two functions can accidentally use separate copies of
+ state that should be shared. A port can hide a problem that
+ appears when the historical build runs. Each failure adds
+ something useful if we can reproduce it and keep the check.
+ </p>
+ <p>
+ Over time, more of the project lived outside any one conversation:
+ source, evidence, comparison scripts, regression tests, and
+ focused notes about what to do next. A fresh agent could inspect
+ that state and continue. The repository was becoming the project's
+ working memory.
+ </p>
<p>
<strong
- >Standardize the checks and the handoff; leave the investigation
- flexible.</strong
+ >A successful batch gave us both recovered code and a better
+ environment for the next batch.</strong
>
- REA fits at the inspection step: it gives the agent code,
- references and observations. Source reconstruction and project
- validation complete the loop.
</p>
</section>
- <section class="article-section" id="th08">
- <h2>3. TH08: continue the work people already did</h2>
- <p>
- Start with the public GensokyoClub reconstruction: its code,
- names, build knowledge and contribution history. Show how a
- continuation turns that foundation into an explicit baseline and a
- sequence of checked tasks.
- </p>
- <ul>
- <li>
- Preserve credit and inspect what the inherited source actually
- establishes.
- </li>
- <li>
- Record the target, existing checks, missing source and next
- useful work.
- </li>
- <li>
- Use the path to Linux and Web to introduce the value of
- readable, maintained source.
- </li>
- </ul>
- <ol class="blog-timeline">
- <li>
- <time datetime="2026-08-13">13 Aug</time
- ><span>Independent continuation begins.</span>
- </li>
- <li>
- <time datetime="2026-08-19">19 Aug</time
- ><span
- >Source recorded for all 1,107 identified game
- functions.</span
- >
- </li>
- <li>
- <time datetime="2026-08-24">24 Aug</time
- ><span
- >Playable Linux port committed, about 11 days after the
- continuation began.</span
- >
- </li>
- <li>
- <time datetime="2026-08-30">30 Aug</time
- ><span>Native Linux 64-bit release published.</span>
- </li>
- </ol>
- <p>
- <strong>Lesson:</strong> a good workflow makes existing expertise
- easier to extend. The starting foundation remains part of the
- result.
+ <section class="article-section" id="different-approaches">
+ <h2>Two different ideas of reconstruction</h2>
+ <p>
+ There is a real disagreement around this work, and GensokyoClub's
+ public README expresses it directly. Its notice announces that
+ further development will happen privately until completion. One
+ passage reads:
+ </p>
+ <blockquote>
+ <p>
+ The rise of grifters (AI decompilations and ports) in this space
+ taking from our work paints a bad image for future decompilation
+ efforts…
+ </p>
+ </blockquote>
+ <p>
+ The notice also describes the psychological toll on the
+ maintainers. Their contribution policy excludes pull requests
+ produced primarily with AI. These are people putting their free
+ time into difficult work, and their published work helped make my
+ continuation possible. That effort deserves respect. The
+ disagreement I want to discuss concerns how reconstruction is done
+ and how contributions are evaluated.
+ </p>
+ <p>
+ In the manual RE workflow I knew, expertise and execution were
+ closely connected. The person developing an understanding of a
+ function also carried out much of the investigation and
+ reconstruction. Personal knowledge, careful authorship and trust
+ in contributors were central to the project.
+ </p>
+ <p>
+ Manual projects already preserve knowledge in code, build tools
+ and tests. What changed for me was having an agent use that
+ accumulated knowledge to investigate, implement and test the next
+ change.
+ </p>
+ <p>
+ My continuation takes an agent-led approach. Agents carry out the
+ new engineering; the human sets direction and release decisions.
+ Understanding has to be retained in source and evidence, while
+ comparisons, builds and runtime experiments test the results. A
+ contribution can be useful even when an agent performed most of
+ the work, provided its reasoning and result can be examined.
+ </p>
+ <p>
+ This creates a difficult transition for communities built around
+ the earlier model. Open source makes existing work available for
+ continuation, and agents can change the pace of that continuation
+ dramatically. Years of careful work may become the starting point
+ for something that moves much faster. Recognition, provenance and
+ the standards for accepting new work become more important in that
+ situation.
+ </p>
+ <p>
+ The industrial analogy helps me make sense of the disagreement. A
+ craft contains skill, judgment and accumulated knowledge. When
+ machinery changes how execution happens, those skills find new
+ roles in designing the process, detecting defects and deciding
+ what should be produced. Different communities can reasonably
+ choose different ways to organize that work.
+ </p>
+ <p>
+ My choice is to develop this continuation openly, preserve its
+ sources and history, and make the new work reviewable. The
+ technical question I want to pursue is how far agent-driven
+ reconstruction can go with strong evidence and a process that
+ keeps improving.
</p>
<p class="caption">
- <a href="https://github.com/N0zoM1z0/th08">TH08 repository</a> ·
+ Source: GensokyoClub's
+ <a
+ href="https://github.com/GensokyoClub/th08/blob/e874b98e210b3be1ed21c5e03d54b883eac5069c/README.md#important-notice"
+ >README notice</a
+ >, checked on 10 October 2026, and its
+ <a
+ href="https://github.com/GensokyoClub/th08/blob/f345c7dee07b849f2ff93a0029ab5ede126bf12c/CODE_OF_CONDUCT.md"
+ >contribution policy</a
+ >. The quotation is a shortened excerpt. TH08's
<a
href="https://github.com/N0zoM1z0/th08/blob/d2a00f4acb4a13916e07a00bd13f714961714628/README.md#credits-and-provenance"
- >History and credits</a
+ >credits and provenance</a
>
+ record the continuation boundary.
</p>
</section>
<section class="article-section" id="th095">
- <h2>4. TH095: start a new game with an established method</h2>
- <p>
- The new repository starts with zero confirmed authored functions.
- It already has experience from earlier reconstructions: target
- checks, task boundaries, comparison tooling and a way to retain
- evidence.
- </p>
- <ul>
- <li>
- Bring up the build and comparison loop early, then work through
- bounded tasks.
- </li>
- <li>
- Use the timeline to show what “fast” means for source recovery,
- exact comparisons and a playable build.
- </li>
- <li>
- Explain one integration failure in plain language: individually
- correct functions can still use the wrong shared state when
- assembled into a program.
- </li>
- </ul>
- <ol class="blog-timeline">
- <li>
- <time datetime="2026-08-29">29 Aug</time
- ><span
- >Repository initialized; authored source count starts at
- zero.</span
- >
- </li>
- <li>
- <time datetime="2026-09-07">7 Sep</time
- ><span
- >697 / 697 identified game functions have source, about 9 days
- later.</span
- >
- </li>
- <li>
- <time datetime="2026-09-08">8 Sep</time
- ><span
- >696 / 697 identified game functions accepted as exact.</span
- >
- </li>
- <li>
- <time datetime="2026-09-10">10 Sep</time
- ><span
- >Windows i386 reconstruction marked playable, about 12 days
- after initialization.</span
- >
- </li>
- </ol>
- <p>
- <strong>Lesson:</strong> the next game starts with more than a
- prompt. It starts with a working process and the previous game’s
- lessons.
+ <h2>TH095: experience starts compounding</h2>
+ <p>
+ <a href="https://github.com/N0zoM1z0/th095">TH095</a>, Shoot the
+ Bullet, made the accumulated experience much easier to see. Its
+ new repository began on 29 August with zero confirmed game
+ functions in the reconstruction ledger. It already had a way to
+ work: identify the target, bring up analysis and comparison tools,
+ keep tasks manageable, preserve evidence, and leave checkpoints
+ another session could use.
</p>
- <p class="caption">
- <a href="https://github.com/N0zoM1z0/th095">TH095 repository</a> ·
- <a
- href="https://github.com/N0zoM1z0/th095/blob/3442dcf29d9e0b5bef384a49bd0c9ad32a711826/docs/WHOLE_BUILD_TODO.md"
- >Build and runtime checkpoint</a
+ <p>
+ By 7 September, all 697 identified game functions had source. By 8
+ September, 696 of them were accepted as exact comparisons. The
+ whole program linked on 9 September, and the Windows i386
+ reconstruction was marked playable on 10 September, about twelve
+ days after initialization.
+ </p>
+ <p>
+ I found this more exciting than the first project's speed. TH095
+ began with the process and experience developed during TH08 and
+ the other reconstruction work. We could approach a new target with
+ better questions, better checks and fewer assumptions left
+ implicit.
+ </p>
+ <p>
+ For example, matching individual functions had taught us to pay
+ attention to the assembled program early. Does the source build
+ together? Do different parts of the game use the same state? Does
+ the reconstructed executable survive the transitions we need to
+ exercise? Those questions became part of how the next project was
+ organized.
+ </p>
+ <p>
+ This is what I mean by knowledge compounding. A lesson that stays
+ in my head helps while I am present. A lesson retained as a test,
+ a script or a precise evidence note can help every later session.
+ A failure becomes valuable when the next agent can recognize it
+ without repeating the whole investigation.
+ </p>
+ <p>
+ <strong
+ >The method becomes part of the starting material for the next
+ game.</strong
>
+ As that material improves, we can spend more attention on what is
+ new about the target.
+ </p>
+ <p>
+ This compounds across people as well as projects. Someone joining
+ later can rerun a check, inspect the reason for a source choice,
+ or continue from a recorded unknown. The project can make use of
+ experience without requiring everyone to reconstruct its entire
+ history from scratch.
</p>
</section>
<section class="article-section" id="th04">
- <h2>5. TH04: keep the method, rethink the assumptions</h2>
- <p>
- PC-98 changes the practical problem. The game spans four DOS
- programs and depends on older hardware behavior. Existing ReC98
- work supplies valuable knowledge; the new reconstruction still
- needs its own checks.
- </p>
- <ul>
- <li>
- Map the whole product: startup, menus, gameplay, endings and the
- resident component.
- </li>
- <li>
- Show why one excellent function percentage can leave important
- parts of the game untested.
- </li>
- <li>
- Adapt the comparison and runtime checks to the platform. Feed
- discoveries back into reusable tests.
- </li>
- </ul>
- <ol class="blog-timeline">
- <li>
- <time datetime="2026-09-06">6 Sep</time
- ><span>Reconstruction repository bootstrapped.</span>
- </li>
- <li>
- <time datetime="2026-09-26">26 Sep</time
- ><span
- >661 / 663 reviewed game functions accepted across four
- programs.</span
- >
- </li>
- <li>
- <time datetime="2026-10-02">2 Oct</time
- ><span
- >Four standalone DOS products built, about 26 days after
- bootstrap.</span
- >
- </li>
- </ol>
- <p>
- <strong>Lesson:</strong> reuse the process, then verify the
- assumptions that the new platform changes.
+ <h2>TH04: the workflow survives a different architecture</h2>
+ <p>
+ <a href="https://github.com/N0zoM1z0/th04">TH04</a>, Lotus Land
+ Story, took the work into the PC-98 DOS era. We moved from 32-bit
+ Windows games to a 16-bit environment with four cooperating
+ programs and hardware behavior that had to be understood on its
+ own terms. Existing
+ <a href="https://github.com/nmlgc/ReC98">ReC98 work</a> provided
+ valuable knowledge and source material here too.
</p>
- <p class="caption">
- <a href="https://github.com/N0zoM1z0/th04">TH04 repository</a> ·
- <a href="../../showcase/th04/">A smaller TH04 investigation</a>
+ <p>
+ The DOS reconstruction is now working, including complete Normal
+ routes, endings and saves from my manual testing. The current work
+ is a native 64-bit port. These are distinct phases: first
+ establish a working reconstruction on the original platform, then
+ carry its behavior into a new environment.
</p>
- </section>
- <section class="article-section" id="workflow">
- <h2>6. Run small experiments and keep useful checkpoints</h2>
- <p>
- Bring the three stories together into a working loop a reader can
- use.
- </p>
- <ol>
- <li>
- <strong>Establish a baseline.</strong> Identify the exact target
- and rerun the checks that matter to this task.
- </li>
- <li>
- <strong>Pick one question.</strong> Choose a behavior or
- subsystem with a result you can inspect.
- </li>
- <li>
- <strong>Gather evidence.</strong> Follow the code that reads or
- changes a value, and the functions that call it; name what is
- known and what still needs testing.
- </li>
- <li>
- <strong>Reconstruct and check.</strong> Compare the affected
- code, build the program, and exercise relevant behavior.
- </li>
- <li>
- <strong>Commit and hand off.</strong> Keep source, evidence,
- failed hypotheses and the next task in the repository.
- </li>
- </ol>
- <p>
- Explain how this loop continues across fresh agent sessions. Turn
- a recurring failure into a test so the next game benefits from it.
+ <p>
+ TH04's architecture changed the evidence we needed and the tools
+ used to collect it. It did not require inventing a different way
+ to organize the work. We could still isolate a question, inspect
+ the original behavior, reconstruct it, compare the result and
+ retain the lesson.
+ </p>
+ <p>
+ The difficult parts became concrete experiments. Which part of
+ startup establishes a piece of state? What happens when gameplay
+ hands control to an ending? Which hardware effect must the
+ reconstructed code preserve? With observations and a useful check,
+ an agent could work through these questions in the same iterative
+ way as it did on the Windows titles.
+ </p>
+ <p>
+ That is why TH04 matters to the larger argument. The process
+ generalized across a substantial change in platform. Architecture
+ determined the subject of the investigation; it did not become a
+ fundamental barrier to the workflow.
+ </p>
+ <p>
+ The 64-bit port now gives that work another use. We can revisit
+ the recovered behavior, separate it from old platform services and
+ test new implementations against what we learned. The
+ reconstruction supplies both source and a growing body of
+ knowledge about how the game should behave.
</p>
<p class="caption">
- Planned figure: a failing experiment returns to investigation; an
- accepted batch feeds both the next task and the shared knowledge
- base.
+ Project state as of 10 October 2026:
+ <a
+ href="https://github.com/N0zoM1z0/th04/blob/bb9faca54e2eb3c0d6c757c75d59ac29d2cd5e93/docs/RE_HANDOFF.md"
+ >DOS reconstruction and manual testing</a
+ >
+ ·
+ <a href="https://github.com/N0zoM1z0/th04/tree/port/modern-64"
+ >64-bit port</a
+ >. The port remains in development.
</p>
</section>
- <section class="article-section" id="ports-and-mods">
- <h2>7. Use the recovered source for ports and mods</h2>
- <p>
- Explain the payoff: source with meaningful names, clear ownership
- and reproducible checks is something you can change with
- confidence.
- </p>
- <ul>
- <li>
- Use TH08’s Linux and Web products to illustrate replacing
- platform services while preserving game logic.
- </li>
- <li>
- Show how a small mod begins with one recovered rule and a
- controlled change.
- </li>
- <li>
- Connect the desired outcome to its checks: faithful behavior, a
- working new platform, or an intentional gameplay difference.
- </li>
- </ul>
- <p class="caption">
- Planned figure: checked reconstruction → readable source →
- portable backend or a deliberate gameplay change.
+ <section class="article-section" id="industrial-shift">
+ <h2>What makes this an industrial shift</h2>
+ <p>
+ These projects changed my view of the bottleneck. As more
+ investigation and implementation can be carried out by agents, the
+ working environment becomes increasingly important. The agent
+ needs access to the target, tools it can compose, feedback it can
+ act on, and memory that survives its current session.
+ </p>
+ <p>
+ REA fits into that picture by making reverse-engineering tools
+ available through interfaces an agent can use. It can inspect
+ code, follow relationships and return evidence for the next
+ question. Reconstruction projects add their own compilers,
+ comparisons and runtime checks. Together, these give an agent a
+ way to move from an explanation to an experiment with an
+ observable result.
+ </p>
+ <p>
+ Fast execution makes good feedback more valuable. A bad assumption
+ can produce a large amount of plausible source. A useful check can
+ reject that assumption early and point to the next investigation.
+ Improving a check improves many later decisions, which is why
+ tooling and verification deserve as much attention as the
+ recovered code.
+ </p>
+ <p>
+ The work also needs continuity. A conversation can end while the
+ campaign continues. Its result should leave the repository in a
+ state another agent can understand: what changed, what was tested,
+ what remains uncertain, and where to continue. This lets a large
+ reconstruction proceed through many small experiments without
+ relying on one uninterrupted session.
+ </p>
+ <p>
+ The
+ <a
+ href="https://github.com/N0zoM1z0/touhou-reconstruction-factory"
+ >Touhou Reconstruction Factory</a
+ >
+ grew out of these experiences. It gives the projects shared ways
+ to retain evidence, run checks and carry useful lessons between
+ games. The point is to make the next investigation easier to
+ execute, easier to judge and easier to resume.
+ </p>
+ <p>
+ This is the industrial shift I see: expertise increasingly lives
+ in a working process that can be reused and improved. Human
+ judgment shapes the process. Agents supply much of the execution.
+ Verified results and accumulated lessons make the process more
+ capable over time.
</p>
</section>
- <section class="article-section" id="start">
- <h2>8. Start with one behavior you can verify</h2>
+ <section class="article-section" id="possibilities">
+ <h2>The projects we can now consider</h2>
+ <p>
+ The pace matters because it changes the cost of beginning. When I
+ imagined doing every investigation manually, I also imagined
+ committing my own attention to each one. A sufficiently large
+ project could be interesting and still remain something I would
+ never realistically attempt.
+ </p>
+ <p>
+ An agent with access, feedback and project memory changes that
+ calculation. Recovering a game, making its source readable,
+ building a port, or exploring a mod can become a sustained
+ engineering effort with a practical way to keep going. The work
+ still involves difficult questions, but there is now a process for
+ giving those questions sustained attention.
+ </p>
+ <p>
+ TH08 showed me what an agent-led continuation could achieve on top
+ of human work. TH095 showed how the experience could carry
+ forward. TH04 showed that the method could cross into a very
+ different platform. Together, they made the industrial analogy
+ feel concrete.
+ </p>
<p>
- Close with a small first task, an example agent prompt, and what
- the reader should keep: recovered source, a reproducible check and
- a clear next question.
+ I now look at an unfamiliar program and ask: what access, feedback
+ and accumulated knowledge would let an agent work on this
+ reliably?
</p>
<p>
- The article’s final prompt will walk through one bounded
- reconstruction batch. The
- <a href="../../guides/native/">native analysis guide</a> and
- <a href="../../get-started/">setup guide</a> cover the tools
- needed to begin.
+ That question opens up projects I would previously have left
+ alone. Each one can add code, checks and experience that make the
+ next more approachable. That is the future of reverse engineering
+ I want to keep exploring.
</p>
+ </section>
+ <section class="article-section" id="sources">
<details class="native-details">
- <summary>Sources and timeline counting</summary>
+ <summary>Project milestones and sources</summary>
<p>
- Dates above come from public GitHub commits and releases,
- checked on 10 October 2026. Elapsed days are calendar time
- between recorded checkpoints. Each row names its milestone:
- source presence, function comparison, build or runtime.
+ The dates describe recorded project checkpoints, checked against
+ public GitHub history on 10 October 2026. Elapsed time is
+ calendar time between commits. Source presence, exact
+ comparisons, a build and a runtime result each name a different
+ milestone.
</p>
<ul>
<li>
- <a
- href="https://github.com/N0zoM1z0/touhou-reconstruction-factory/tree/038b6bed51d273c8b3680553b6f43e1b97d19bd1"
- >Touhou Reconstruction Factory</a
- >: project memory, bounded work, empirical feedback and
- cross-game lessons.
- </li>
- <li>
+ <strong>TH08:</strong>
<a
href="https://github.com/N0zoM1z0/th08/commit/001bf3e9c91cc35b79c7a0e36b3565b86f494362"
- >TH08 continuation start</a
+ >13 August continuation</a
>,
<a
href="https://github.com/N0zoM1z0/th08/blob/15caba73df6735fc44eac4783bc61b8d098ea830/docs/PROGRESS.md"
- >source checkpoint</a
+ >19 August source ledger</a
>,
<a
href="https://github.com/N0zoM1z0/th08/commit/da07b21c8cb63230142b0a66228d6aac66555f3f"
- >Linux port</a
- >
- and
+ >24 August Linux port</a
+ >,
+ <a
+ href="https://github.com/N0zoM1z0/th08/commit/529eb1a3afa05ae36467731eaccb12e3e50e132f"
+ >26 August Web edition</a
+ >, and
<a
href="https://github.com/N0zoM1z0/th08/releases/tag/v0.2.0-linux-64bit"
- >64-bit release</a
+ >30 August 64-bit Linux release</a
>.
</li>
<li>
+ <strong>TH095:</strong>
<a
href="https://github.com/N0zoM1z0/th095/blob/8c31e6f892d945423b7f7b439747a2bd915f82d8/docs/PROGRESS.md"
- >TH095 initial ledger</a
+ >29 August initial ledger</a
>,
<a
href="https://github.com/N0zoM1z0/th095/blob/0fa857fb2b2ec6237732e4d6be8d0a28069c9330/docs/PROGRESS.md"
- >source checkpoint</a
- >
- and
+ >7 September source ledger</a
+ >,
<a
href="https://github.com/N0zoM1z0/th095/blob/74baddfccb712d1311a4d6e7487a63cd1100452e/docs/PROGRESS.md"
- >exact comparison checkpoint</a
+ >8 September comparisons</a
+ >,
+ <a
+ href="https://github.com/N0zoM1z0/th095/commit/33d46a0ee48f37060d973125a1a7632a40f8d998"
+ >9 September linkage</a
+ >, and
+ <a
+ href="https://github.com/N0zoM1z0/th095/blob/3442dcf29d9e0b5bef384a49bd0c9ad32a711826/docs/WHOLE_BUILD_TODO.md"
+ >10 September playable-build record</a
>.
</li>
<li>
+ <strong>TH04:</strong>
<a
- href="https://github.com/N0zoM1z0/th04/commit/f402cb30b7e795ea4328b6de794e9eefc52c97b3"
- >TH04 bootstrap</a
- >,
+ href="https://github.com/N0zoM1z0/th04/blob/bb9faca54e2eb3c0d6c757c75d59ac29d2cd5e93/docs/RE_HANDOFF.md"
+ >10 October handoff</a
+ >
+ records the working DOS reconstruction, the maintainer's
+ complete Normal-route testing and the current 64-bit phase.
+ </li>
+ <li>
+ <strong>Method:</strong> the Factory's
<a
- href="https://github.com/N0zoM1z0/th04/blob/1cd1a2577a3dbf4c17ac7a939faed5eeb81a6ebf/docs/PROGRESS.md"
- >four-artifact ledger</a
+ href="https://github.com/N0zoM1z0/touhou-reconstruction-factory/blob/038b6bed51d273c8b3680553b6f43e1b97d19bd1/docs/agent-autonomy.md"
+ >agent autonomy</a
>
and
<a
- href="https://github.com/N0zoM1z0/th04/commit/4b6a57f8c0625da07c38258b78f5f205aae92805"
- >four-product build checkpoint</a
- >.
+ href="https://github.com/N0zoM1z0/touhou-reconstruction-factory/blob/038b6bed51d273c8b3680553b6f43e1b97d19bd1/docs/knowledge-base.md"
+ >cross-game knowledge</a
+ >
+ documents retain the working principles and lessons.
</li>
</ul>
</details>
+ <p>
+ <a class="text-link" href="../">More articles</a> ·
+ <a class="text-link" href="../../showcase/th04/"
+ >Explore a TH04 investigation</a
+ >
+ </p>
</section>
</article>
</div>
diff --git a/website/style-guide.md b/website/style-guide.md
index 94202e903..77f517784 100644
--- a/website/style-guide.md
+++ b/website/style-guide.md
@@ -146,9 +146,12 @@ purposes, and immutable commit links for checkpoint-dependent facts.
### Method articles in the Blog
-Use longer articles to explain how to approach a problem, choose a workflow and
-learn from project outcomes. Open with the reader's goal and one clear thesis.
-Use an overview figure to show the method before expanding its steps.
+Use longer articles to explain methods, choices and ideas through project
+experience. Open with a clear takeaway. A personal essay can follow how the
+author's understanding changed, using concrete episodes to develop a deeper
+argument. Keep the reasoning easy to follow; a Blog article does not need to
+become a tutorial. Figures are optional. Add one when it explains something
+useful rather than repeating a familiar opening flow.
Choose cases that teach different decisions. For example, TH08 illustrates
continuing existing work, TH095 illustrates reusing an established method in a
@@ -158,12 +161,18 @@ beside the milestone they measure, with pinned evidence links in supporting
details. Calendar time, source presence, exact comparison, build success and
runtime checks each describe a different result.
-Make an analogy concrete and use it briefly. The industrial analogy can connect
-agent execution, tools, quality checks and durable knowledge; the actual workflow
-should remain clear without the metaphor. Explain necessary terms in plain
-language and leave engine inventories in references. Finish with a bounded first
-task the reader can try. Working outlines should be visibly labeled and marked
-`noindex` until they become finished articles.
+Make an analogy concrete. The industrial analogy can connect agent execution,
+quality checks, durable knowledge and a change in how work is organized. Explain
+what changes and support the argument with experience. Keep necessary terms
+plain and leave engine inventories in references. A method guide can finish with
+a first task; an essay can finish with the implications of its argument.
+
+When discussing another project's approach, preserve its credit and contribution
+history. Attribute quotations, link their context and distinguish the author's
+interpretation from the other project's stated position. Describe the technical
+or contribution-model disagreement without assigning motives to its authors.
+Working outlines should be visibly labeled and marked `noindex` until they
+become finished articles.
## Make figures explain a relationship