snapcompact cannot run locally: ${this.#model.id} is text-on

Error message

snapcompact cannot run locally: ${this.#model.id} is text-only.

What it means

Thrown when snapcompact is selected but the active model's declared inputs do not include `image`. snapcompact archives history into dense bitmap images the model must read back, so it fundamentally requires a vision-capable model; with a text-only model it cannot run and the library fails fast after emitting a warning notice.

Source

Thrown at packages/coding-agent/src/session/session-maintenance.ts:865

			// Focus instructions require an LLM summary, so the preference resolver
			// only selects snapcompact for an undirected manual compaction.
			const wantsSnapcompact = compactionPrep.kind !== "fromHook" && selectedMethod === "snapcompact";
			const snapcompactReady = wantsSnapcompact;
			const snapcompactShapeSetting = this.#host.settings.get("snapcompact.shape");
			let snapcompactShape: snapcompact.Shape | undefined;
			// Claude refuses inputs that reproduce its own reasoning as text
			// ("reasoning_extraction"), and the snapcompact archive is replayed as
			// text into every later request; drop `¶think:` sections for
			// Anthropic-dialect targets (issue #6093).
			const snapcompactIncludeThinking = preferredDialect(this.#model.id) !== "anthropic";
			if (wantsSnapcompact && !this.#model.input.includes("image")) {
				this.#host.emitNotice(
					"warning",
					`snapcompact needs a vision-capable model (${this.#model.id} is text-only)`,
					"compaction",
				);
				throw new Error(`snapcompact cannot run locally: ${this.#model.id} is text-only.`);
			} else if (snapcompactReady) {
				const text = snapcompact.serializeConversation(
					convertToLlm(preparation.messagesToSummarize.concat(preparation.turnPrefixMessages)),
					{ includeThinking: snapcompactIncludeThinking },
				);
				const probeText = snapcompact.renderabilityProbeText(
					text,
					preparation.previousPreserveData,
					preparation.previousSummary,
				);
				snapcompactShape = snapcompact.resolveShapeForText(probeText, this.#model, snapcompactShapeSetting);
				const renderScan = snapcompact.scanRenderability(probeText, { shape: snapcompactShape });
				if (!renderScan.isSafe) {
					const percent = (renderScan.unrenderableRatio * 100).toFixed(1);
					this.#host.emitNotice(
						"warning",
						`snapcompact disabled: unsupported characters for selected snapcompact font (${percent}%).`,
						"compaction",

View on GitHub (pinned to 9690622007)

Solutions

Select a vision-capable model (input includes "image") before running snapcompact.
Use `/compact soft` or `/compact remote` instead, which summarize as text.
If the model actually supports images, fix the model metadata (catalog/discovery) so `input` lists "image".

Example fix

// before
await session.compact({ mode: "snapcompact" }); // text-only model
// after
await session.setModel("google/gemini-2.5-flash"); // vision-capable
await session.compact({ mode: "snapcompact" });

Defensive patterns

Strategy: validation

Validate before calling

const model = session.getModel();
if (mode === "snapcompact" && model && !model.input.includes("image")) {
  throw new Error(`snapcompact needs a vision-capable model (${model.id} is text-only)`);
}

Type guard

function isVisionCapable(model: { input: string[] }): boolean {
  return model.input.includes("image");
}

Try / catch

try {
  await session.compact({ mode: "snapcompact" });
} catch (err) {
  if (err instanceof Error && err.message.includes("snapcompact cannot run locally")) {
    return session.compact({ mode: "soft" }); // text-summary fallback
  }
  throw err;
}

Prevention

Check `model.input` for "image" before choosing snapcompact.
Pick a vision-capable default model if snapcompact is your standard flow.
Distrust modality metadata for new models — verify vision support with a probe.

When it happens

Trigger: Running `/compact snapcompact` (or the snapcompact method via methodOrder) while the selected model has `input` metadata lacking "image" — e.g. a text-only model like an older or reasoning-only variant.

Common situations: Switching to a text-only model then using a saved snapcompact default; discovery metadata misreporting modality so the user believes the model supports images; hardcoded snapcompact in scripts after switching providers.

Related errors

AI-assisted analysis of can1357/oh-my-pi@9690622007 (2026-08-31). Data as JSON: /api/errors/361091a20f9c16e6. Report an issue: GitHub.