StoryHelm - Command Your Series (Scott)

StoryHelm is manuscript intelligence for indie series authors.

[StoryHelm - Command Your Series (Scott) @StoryHelm]

@ian_finlay No. In the Software Factory, the reviewers run on the same model as the builder. What makes them independent is that each one starts with a fresh context.

What do you do?

[StoryHelm - Command Your Series (Scott) @StoryHelm]

I let Claude write most of the code for StoryHelm. Then I spent a week building the thing that stops it from hurting me.

StoryHelm reads an entire book series and flags the contradictions the author missed. My own repo needed the same treatment.

Here is what went wrong first, all in the StoryHelm repo:

1. A commit swept 33 files from a different Claude Code session into the wrong branch.
2. Two Claude sessions both numbered the same database migration, so we had two heads.
3. A cache flush wiped 40 minutes of in-flight manuscript analysis.
4. A bulk find-and-replace silently overwrote the StoryHelm sidebar.

I had rules for every one of these. Written rules. In a file Claude reads.

It did not matter.

A rule is advice, not a wall.

So I built two things.

1. A Coding Harness.
Claude Code hooks refuse the dangerous command before it runs: force-push, shared stash, bulk delete, flushing the cache. One gate script runs only the checks a change needs, on the committed code, and answers with an exit code. The gate reports tests that were already red on main. The change does not take the blame.

2. A Software Factory.
I write a spec. One Claude agent plans it, a second Claude agent tries to break the plan, and I approve it. A third builds it in its own worktree. Three Claude reviewers check it. One of them copies only the new tests onto main and proves they fail without the change. Then it stops. I decide whether it lands.

Two decisions per change.

Everything else is machinery.

What surprised me:

-The reviewers found real bugs in the Software Factory itself, five blocking ones in a single round.
-A review caught tests that passed with the code deliberately broken. Green does not mean the test can fail.
-A report is a claim. An exit code is a check. I only trust the second.
-It is slower. A one-test fix took 38 minutes. I only use it where a mistake is expensive.

StoryHelm runs 61 Claude agents across 7 workflows on authors' manuscripts.

A mistake there is expensive.

My advice for anyone building with Claude Code: do not write better rules.

Move the rules into the tool boundary.

What is the worst thing an AI agent has done to your repo?

#buildinpublic #ClaudeCode #StoryHelm #AIengineering

[StoryHelm - Command Your Series (Scott) @StoryHelm]

@chenzhangreal @chenzhangreal - you know it. 💪💪💪

[StoryHelm - Command Your Series (Scott) @StoryHelm]

@SparkPitLabs Great question. It's the checkout/switch rule. It doesn't block the commit itself. It blocks the step before it: another session moving the shared tree onto its own branch.

[StoryHelm - Command Your Series (Scott) @StoryHelm]

@jeffbullas Yes, at age 53, I have been building StoryHelm for the last 14 months (https://t.co/UuZi3lDnlT). I am also having the time of my life, such a rush!

[StoryHelm - Command Your Series (Scott) @StoryHelm]

@DarrelFrater I would

[StoryHelm - Command Your Series (Scott) @StoryHelm]

In A Study in Scarlet (1887), Dr. Watson is "struck on the shoulder by a Jezail bullet." Three years later, in The Sign of the Four, he names that first book, and a few paragraphs on he is "nursing my wounded leg. I had a Jezail bullet through it." Arthur Conan Doyle moved his own narrator's war wound, and Holmes fans still argue about it.

StoryHelm read both public-domain novels. Nobody asked it about Watson. It flagged the shoulder-to-leg drift on its own and set the two accounts side by side, for the author to decide.

That's not a knock on Doyle. He wrote sixty Holmes stories over forty years. If you're writing a series, every scar, name and promise from book one is still yours to keep straight by book five.

StoryHelm reads your whole series, keeps one shared canon, and flags the details that drift between books. It reads. It never writes. You decide what stays.

Start with the free Series Continuity Self-Audit (no manuscript upload required):

https://t.co/DWzQX1w1RL

[StoryHelm - Command Your Series (Scott) @StoryHelm]

A hand-built series bible is a second job nobody has time for. By Book 2, it's stale.

StoryHelm builds the canon from the pages you already wrote. It reads your series. It never writes your story.

https://t.co/UuZi3lCPwl #IndieAuthor #AmWriting

[StoryHelm - Command Your Series (Scott) @StoryHelm]

The Series Tax. Three books in, you'll spend about 113 hours just keeping your story straight. That's our estimate.

StoryHelm reads your whole series and keeps one shared canon. It reads. It never writes.

https://t.co/UuZi3lCPwl #amwriting #indieauthor #selfpublishing

[StoryHelm - Command Your Series (Scott) @StoryHelm]

@DarrelFrater It is now. Sorry about that.

[StoryHelm - Command Your Series (Scott) @StoryHelm]

Your series doesn’t fall apart because you forgot how to write.

It gets harder because every book adds weight:

- More characters to track
- More promises to carry
- More world rules to honor
- More timelines to reconcile
- More reader expectations to protect

That’s the hidden cost of writing a multi-book fiction series.

By Book 3, 5, or 9, the problem isn’t imagination. It’s memory under pressure.

Common pain points indie series authors know too well:

Continuity errors across books
A scar changes sides. A city moves. A character knows something they never learned.

The reread tax
You step away for 3 months and lose a week rereading your own work before you can draft again.

Character and voice drift
A sharp-tongued lead softens without reason. A side character starts sounding like someone else.

Dropped threads and forgotten setups
You planted the relic, the prophecy, the betrayal, the missing sibling. Then the series kept moving.

World-rule inconsistencies
Magic, technology, politics, geography, and history start bending in ways readers will notice.

Scattered notes and story bibles
Canon lives in Scrivener folders, spreadsheets, old drafts, margin comments, and your head.

Pre-publication anxiety
The book is ready to upload, but one question won’t leave: what will readers catch that I missed?

StoryHelm is built for that exact pressure.

It’s an AI manuscript-intelligence platform for long-form fiction and multi-book series. It reads and analyzes the full series, then turns scattered manuscripts into a living canon and Series Atlas.

StoryHelm helps you:

- Surface contradictions across books
- Track characters, aliases, relationships, and appearances
- Identify silent risks like dropped threads and uncarried promises
- Preserve world rules, timelines, and series logic
- Reduce reread time after long gaps
- Triage revision work before readers or editors find the seams
- Keep authorial control where it belongs: with you

It does not write your prose.
It does not replace your judgment.
It does not decide what your story should become.

It gives you the series-wide memory and editorial visibility serious indie authors need when the story is too large to hold in one head.

You write the story. StoryHelm helps make sure it holds together.

Explore StoryHelm at https://t.co/RebQVxxN0E.

[StoryHelm - Command Your Series (Scott) @StoryHelm]

A reader is keeping a spreadsheet of your series.

StoryHelm checks your books against each other, so you see the slip before they do.

Free Series Continuity Self-Audit, no manuscript upload:

https://t.co/UuZi3lDnlT

[jack friks @jackfriks]

it worked btw

[StoryHelm - Command Your Series (Scott) @StoryHelm]

Kill him in book two and book four falls apart. StoryHelm shows you the damage before you write a word of it.