Canon Rules

Canon rules are the source of truth your story is checked against. Define them by hand, or let PlotLens derive them from your manuscript and confirm them on review.

What canon means in PlotLens

Canon is your story’s ground truth: the facts, relationships, and world constraints that everything else must stay consistent with. In PlotLens, canon is expressed as rules — structured statements like “Alice has blue eyes” or “Magic cannot create food” that the validator uses to flag contradictions in new content.

Rules carry a type, a severity, an epistemic status, and a review status. The validator only checks against rules that are accepted and active, so you stay in control of what counts as canon.

When to use this

Reach for canon rules when you want to:

• Pin down a fact you keep accidentally contradicting (eye color, ages, places of birth).
• Lock a worldbuilding constraint so drafts that break it get flagged automatically.
• Curate the rules PlotLens auto-derived from your manuscripts before they enter validation.
• Roll a rule back to an earlier wording after an edit you didn’t mean to keep.

Rule types

PlotLens has five rule types. Pick the one that matches the shape of the fact you’re encoding.

Type	What it captures	Example
Attribute	A scalar property of a single entity.	”Alice has blue eyes.”
Relationship	A directed link between two entities.	”John is Mary’s brother.”
Timeline	A temporal constraint between events.	”The coronation happens before the siege.”
World rule	A systemic constraint on the story world.	”Magic cannot create food.”
Fact	A standalone assertion not tied to a specific entity.	”The capital is Eldoria.”

How rules get created

Rules enter your project two ways.

Manual. You open the Canon Rules page and create a rule from scratch. You give it a name, description, type, severity (critical, high, medium, or low), and the constraint itself. Manual rules are accepted by default and active immediately.

Auto-derived. After PlotLens extracts entities from a document, a background job proposes rules based on what it found:

• Entity attributes become attribute rules (scalar values only).
• Entity relationships with confidence of 0.5 or higher become relationship rules.
• Concept entities with descriptions become world rules.

Auto-derived rules land with a pending review status and a snippet of the source text from your manuscript so you can see exactly why PlotLens proposed them. They do not affect validation until you accept them.

Epistemic metadata

Every rule has an epistemic status that tells the validator how seriously to treat it. This is how you encode that something is canon-in-the-world without it being canon-as-narrated, and it is how you suppress false-positive conflicts.

Status	Meaning	Validator behavior
fact	Established truth in the story world.	Checked at full severity.
rumor	Hearsay, not confirmed.	Downgraded to low severity.
speculation	A character’s guess or theory.	Downgraded to low severity.
lie	A character is lying on the page.	Excluded from validation.
dream	Dream, vision, or hallucination content.	Excluded from validation.
retcon	Superseded by a later authoritative version.	Excluded from validation.
red_herring	A misdirect placed on purpose.	Excluded from validation.
unreliable_pov	Stated by a narrator you can’t trust.

If validation keeps flagging something you wrote on purpose — a character’s lie, a deliberate misdirect, a dream sequence — change the rule’s epistemic status instead of disabling the rule.

How to create a rule

1. Open your project and go to Canon in the left sidebar.
2. Click New rule in the top right.
3. Give the rule a short name and a one-sentence description.
4. Pick a rule type (attribute, relationship, timeline, world rule, or fact).
5. Set the severity (critical, high, medium, or low). Severity controls how loud the validator is when something contradicts this rule.
6. Optionally set an epistemic status. The default is fact.
7. Fill in the rule body — the entity reference, the relationship target, the timeline pair, or the free-form constraint, depending on the type.
8. Click Save. The rule is active immediately and will be used the next time you run validation.

Reviewing auto-derived rules

When you upload a new document, PlotLens proposes rules in the background. They wait for you in the Pending filter on the Canon Rules page.

1. Open Canon and switch the review filter to Pending.
2. Click any rule to open the side panel. You’ll see the rule itself plus the source text it was derived from.
3. Accept the rule to add it to active canon, or Reject it to discard it.
4. To move quickly, select multiple rules and use the bulk Accept or Reject controls.

Accepted rules become active canon. Rejected rules are dismissed and won’t be re-proposed for the same source.

Versioning and revert

Every edit to a rule creates a new version. PlotLens keeps the full history.

• Open a rule and click Version history to see every change, who made it, and when.
• Click any prior version to view a side-by-side diff against the current version.
• Click Revert to this version to restore an earlier wording. Revert creates a new version on top of history rather than deleting anything, so you can always undo your undo.

Plan availability

Canon rules are a Plus feature. Custom rules — both manually created and the auto-derived review flow — are gated to Plus and above.

Plan	Custom rule limit
Free	Not available
Lite	Not available
Plus	100 rules per project
Pro	Unlimited
Small Team	Unlimited
Studio, Production, Enterprise	Unlimited

If you’re on Free or Lite and try to create a rule, you’ll see an upgrade prompt pointing at the Plus plan. Multi-project canon (sharing a rule set across projects) requires Pro or higher.

See Billing & plans for the full plan matrix.

Limits & edge cases

• Plus rule cap. Plus tops out at 100 custom rules per project. Auto-derived rules count toward this limit once accepted, not while pending. If you hit the cap, reject low-value pending rules or upgrade to Pro for unlimited.
• Low-confidence relationships. Auto-inference skips relationships with confidence under 0.5. If a relationship you expect never appears as a pending rule, the extractor wasn’t sure enough — create it manually.
• Source document deleted. If you delete a document that produced a rule, the rule itself stays. The source-text quote in the review panel may go stale.
• Rejected rules and re-inference. Rejecting a pending rule is sticky for that exact source. If you re-import a meaningfully different version of the document, you may see a similar rule proposed again — that’s expected.
• Source-text length. Auto-derived rules cap their source-text quote at roughly 2 KB. The full passage stays in the source document.
• Team permissions. On team-owned projects, editing rules requires the Editor role or higher. Viewers can read canon but not change it.

Common pitfalls

• Treating every auto-derived rule as gospel. Pending rules are proposals, not canon. Skim the source text before accepting — extraction is good, not perfect.
• Disabling rules to suppress false positives. If a rule is correct but the content you wrote is intentionally non-canon (a lie, a dream, a red herring), change the rule’s epistemic status rather than disabling the rule. Disabling hides the rule from validation entirely; epistemic status keeps the model honest.
• Setting everything to critical severity. If everything is critical, nothing is. Reserve critical for facts that would break the story if contradicted.
• Creating duplicate rules. Before adding a manual rule, search the canon list — you may already have a pending rule covering the same fact.
• Forgetting to review pending rules. Pending rules don’t affect validation. If your story checks feel quiet, check the Pending filter for unreviewed canon.

Related

• Entity extraction — where auto-derived rules come from
• Validation — how rules become contradiction findings
• Timeline — temporal canon and timeline rules
• Billing & plans — Plus tier and rule limits