Sillytavern Component Update

SillyTavern Component Update

Shared communication

Apply TW plain-language and guidance rules to user-facing work. Explain terms in context; preserve the user's chosen 新人/入门/熟练/老手 level without inferred changes. 新人 and 入门 receive detailed explanations; every level receives needed and bug explanations unless the user explicitly waives that scope. Soul and prose modes never disable this baseline. Load the shared reference for task entry, level management, or explanation decisions.

Before a write-capable update, use $consult-tavernweave-library with the sillytavern-component-update route to load A0 and the smallest matching ST guides. Keep the receipt with the component-mode scope lock; it does not authorize repacking the full card.

Keep the selected component as the unit of change. Produce importable test artifacts without forcing an unrelated card rebuild.

Freeze the delivery mode

Before writing, record exactly one mode:

• component: emit only importable regex, Tavern Helper script, or script-folder JSON plus a validation report;
• full-card: emit the changed component artifacts and an assembly handoff, then use $sillytavern-card-pipeline for composition and packaging.

If the user did not select a mode, stop at a read-only plan. Never infer full-card permission from a request to update one regex or helper script.

Read component-update-contract.md before creating a spec or output directory. Read importable-formats.md when selecting a regex or Tavern Helper script dialect.

Establish the target

1. Read the project manifest, component registry, and current packaged surface.
2. Identify the target by stable ID, not display name alone.
3. Record the source file, current hash, output kind, runtime owner, enabled state, and expected untouched fields.
4. Preserve unknown fields and prior releases unless an explicit migration contract says otherwise.
5. Verify version-sensitive field names with $sillytavern-api-reference when the installed SillyTavern or Tavern Helper version differs from the pinned evidence.

Plan and build

Create a UTF-8 JSON spec and inspect the write plan first:

node scripts/plan-component-update.mjs --spec update-spec.json --out staging
node scripts/build-importable-component.mjs --spec update-spec.json --out staging --write

The build command is dry-run unless --write is present. Accept the script's output paths only when they remain inside the declared staging directory.

For component mode, reject any plan that emits a card JSON or PNG. For full-card mode, require assembly.targetCard, assembly.componentId, and assembly.expectedUntouchedHash; the builder emits a handoff but does not package the card itself.

Validate the artifact

Run:

node scripts/validate-importable-component.mjs staging

Require:

• valid UTF-8 JSON and a recognized component kind;
• stable, unique IDs;
• compilable regex patterns and non-empty helper-script content;
• preserved enabled state, ordering, buttons, data, and extension fields;
• no whole-card output in component mode;
• semantic parity outside the selected component in full-card mode.

Static validation proves only the artifact contract. Keep import, execution, regex placement, message rendering, and interaction pending until real SillyTavern acceptance through $sillytavern-runtime-debug.

Report

State the selected mode, source hash, generated files, validation result, pipeline handoff if any, untouched scope, and remaining live-import checks.

Resources

• scripts/plan-component-update.mjs: validate a spec and print the exact write plan.
• scripts/build-importable-component.mjs: emit deterministic component JSON and an optional full-card assembly handoff.
• scripts/validate-importable-component.mjs: validate one artifact or a directory.
• references/component-update-contract.md: mode, spec, and output ownership.
• references/importable-formats.md: supported regex and helper-script shapes.