@diegopetrucci/pi-extensions 0.1.14 → 0.1.15

package/README.md

@@ -4,6 +4,7 @@ A collection of [pi](https://github.com/earendil-works/pi-mono) agent extensions
 
 - [`minimal-footer`](./extensions/minimal-footer): Replaces pi's built-in footer with a minimal configurable two-line layout: branch/repo on the first line, context/model on the second, optional `DUMB ZONE`, plus OpenAI Codex 5-hour and 7-day usage when available.
 - [`oracle`](./extensions/oracle): Adds an Amp-style read-only oracle tool that auto-selects the strongest reasoning model on the current provider/subscription, covers pi’s built-in providers with hardcoded rankings, sets reasoning to xhigh by default, and shows live status while running.
+- [`context-cap`](./extensions/context-cap): Caps effective model context windows at 200k tokens by default so pi auto-compacts earlier; toggle temporarily with `/context-cap`.
 - [`permission-gate`](./extensions/permission-gate): Prompts for confirmation before dangerous bash commands like `rm -rf`, `sudo`, and `chmod 777`.
 - [`confirm-destructive`](./extensions/confirm-destructive): Confirms before destructive session actions like clear, switch, and fork.
 - [`notify`](./extensions/notify): Sends configurable terminal, desktop, bell, and sound notifications when pi finishes and is ready for input.
@@ -21,13 +22,15 @@ pi install npm:@diegopetrucci/pi-extensions
 Or pin the GitHub package to this release:
 
 ```bash
-pi install git:github.com/diegopetrucci/pi-extensions@v0.1.14
+pi install git:github.com/diegopetrucci/pi-extensions@v0.1.15
 ```
 
 Or a specific extension:
 
 ```bash
 pi install npm:@diegopetrucci/pi-oracle
+# or
+pi install npm:@diegopetrucci/pi-context-cap
 ```
 
 Then reload pi:

package/extensions/context-cap/README.md

@@ -0,0 +1,60 @@
+# context-cap
+
+A pi extension that treats large-context models as having an effective 200k-token context window, so pi's built-in auto-compaction starts earlier.
+
+By default, pi auto-compacts when:
+
+```text
+contextTokens > model.contextWindow - reserveTokens
+```
+
+This extension changes the active model's in-memory `contextWindow` to:
+
+```text
+min(originalContextWindow, 200000)
+```
+
+With pi's default `reserveTokens` of 16,384, models larger than 200k will proactively compact around 183,616 tokens.
+
+## Commands
+
+```text
+/context-cap status
+/context-cap off
+/context-cap on
+/context-cap toggle
+```
+
+The extension starts enabled by default. Disabling is temporary for the current extension runtime/session; after `/reload`, `/new`, `/resume`, or `/fork`, the extension starts enabled again.
+
+## Install
+
+### Standalone npm package
+
+```bash
+pi install npm:@diegopetrucci/pi-context-cap
+```
+
+### Collection package
+
+```bash
+pi install npm:@diegopetrucci/pi-extensions
+```
+
+### GitHub package
+
+```bash
+pi install git:github.com/diegopetrucci/pi-extensions
+```
+
+Then reload pi:
+
+```text
+/reload
+```
+
+## Notes
+
+- This extension mutates pi's in-memory model metadata only. It does not edit `models.json`.
+- The cap affects pi logic that reads `model.contextWindow`, including auto-compaction thresholding and UI context-window display.
+- Because pi also uses `model.contextWindow` for some overflow detection, a request that succeeds above 200k tokens on a larger model may be treated as overflow and retried after compaction. Use `/context-cap off` if you need the full model window temporarily.

package/extensions/context-cap/index.ts

@@ -0,0 +1,179 @@
+import type { ExtensionAPI, ExtensionContext } from "@earendil-works/pi-coding-agent";
+import type { Model, Api } from "@earendil-works/pi-ai";
+
+const DEFAULT_MAX_CONTEXT_WINDOW = 200_000;
+type AnyModel = Model<Api> | Model<any>;
+
+const originalContextWindows = new WeakMap<AnyModel, number>();
+const touchedModels = new Set<AnyModel>();
+
+type ApplyResult = {
+	changed: boolean;
+	key: string;
+	original: number;
+	effective: number;
+};
+
+function modelKey(model: AnyModel): string {
+	return `${model.provider}/${model.id}`;
+}
+
+function getOriginalContextWindow(model: AnyModel): number {
+	const existing = originalContextWindows.get(model);
+	if (typeof existing === "number") return existing;
+
+	originalContextWindows.set(model, model.contextWindow);
+	touchedModels.add(model);
+	return model.contextWindow;
+}
+
+function getEffectiveContextWindow(model: AnyModel): number {
+	return Math.min(getOriginalContextWindow(model), DEFAULT_MAX_CONTEXT_WINDOW);
+}
+
+function applyContextCap(model: AnyModel | undefined): ApplyResult | undefined {
+	if (!model) return undefined;
+
+	const key = modelKey(model);
+	const original = getOriginalContextWindow(model);
+	const effective = Math.min(original, DEFAULT_MAX_CONTEXT_WINDOW);
+	const changed = model.contextWindow !== effective;
+	model.contextWindow = effective;
+
+	return { changed, key, original, effective };
+}
+
+function restoreContextWindow(model: AnyModel | undefined): boolean {
+	if (!model) return false;
+
+	const original = originalContextWindows.get(model);
+	if (typeof original !== "number" || model.contextWindow === original) return false;
+
+	model.contextWindow = original;
+	return true;
+}
+
+function forEachRegistryModel(ctx: ExtensionContext, callback: (model: AnyModel) => void): void {
+	try {
+		for (const model of ctx.modelRegistry.getAll()) {
+			callback(model);
+		}
+	} catch {
+		// Best effort only. The active ctx.model is handled separately by callers.
+	}
+}
+
+function applyContextCapToSession(ctx: ExtensionContext): number {
+	let changed = 0;
+
+	forEachRegistryModel(ctx, (model) => {
+		if (applyContextCap(model)?.changed) changed++;
+	});
+
+	if (applyContextCap(ctx.model)?.changed) changed++;
+	return changed;
+}
+
+function restoreContextCapForSession(ctx: ExtensionContext): number {
+	let changed = 0;
+
+	forEachRegistryModel(ctx, (model) => {
+		if (restoreContextWindow(model)) changed++;
+	});
+
+	if (restoreContextWindow(ctx.model)) changed++;
+	return changed;
+}
+
+function formatTokens(tokens: number): string {
+	return tokens >= 1000 ? `${Math.round(tokens / 1000)}k` : String(tokens);
+}
+
+export default function contextCapExtension(pi: ExtensionAPI) {
+	let enabled = true;
+
+	pi.on("session_start", async (_event, ctx) => {
+		if (enabled) applyContextCapToSession(ctx);
+	});
+
+	pi.on("model_select", async (event, ctx) => {
+		if (!enabled) return;
+
+		const result = applyContextCap(event.model);
+		if (!result || !ctx.hasUI) return;
+
+		ctx.ui.setStatus(
+			"context-cap",
+			result.original > DEFAULT_MAX_CONTEXT_WINDOW
+				? `ctx cap ${formatTokens(result.effective)}/${formatTokens(result.original)}`
+				: undefined,
+		);
+	});
+
+	pi.on("session_shutdown", async (_event, ctx) => {
+		restoreContextCapForSession(ctx);
+		for (const model of touchedModels) restoreContextWindow(model);
+	});
+
+	pi.registerCommand("context-cap", {
+		description: "Toggle the 200k effective context-window cap for auto-compaction",
+		getArgumentCompletions: (prefix) => {
+			const commands = ["on", "off", "toggle", "status"];
+			const matches = commands.filter((command) => command.startsWith(prefix.trim()));
+			return matches.length > 0 ? matches.map((value) => ({ value, label: value })) : null;
+		},
+		handler: async (args, ctx) => {
+			const action = args.trim().toLowerCase() || "toggle";
+
+			if (action === "on" || action === "enable") {
+				enabled = true;
+				const changed = applyContextCapToSession(ctx);
+				ctx.ui.setStatus("context-cap", `ctx cap ${formatTokens(DEFAULT_MAX_CONTEXT_WINDOW)}`);
+				ctx.ui.notify(`Context cap enabled (${changed} model window(s) capped/restored).`, "info");
+				return;
+			}
+
+			if (action === "off" || action === "disable") {
+				enabled = false;
+				const changed = restoreContextCapForSession(ctx);
+				ctx.ui.setStatus("context-cap", undefined);
+				ctx.ui.notify(`Context cap disabled for this extension session (${changed} model window(s) restored).`, "info");
+				return;
+			}
+
+			if (action === "toggle") {
+				if (enabled) {
+					enabled = false;
+					const changed = restoreContextCapForSession(ctx);
+					ctx.ui.setStatus("context-cap", undefined);
+					ctx.ui.notify(`Context cap disabled for this extension session (${changed} model window(s) restored).`, "info");
+				} else {
+					enabled = true;
+					const changed = applyContextCapToSession(ctx);
+					ctx.ui.setStatus("context-cap", `ctx cap ${formatTokens(DEFAULT_MAX_CONTEXT_WINDOW)}`);
+					ctx.ui.notify(`Context cap enabled (${changed} model window(s) capped/restored).`, "info");
+				}
+				return;
+			}
+
+			if (action === "status") {
+				const model = ctx.model;
+				const status = enabled ? "enabled" : "disabled";
+				if (!model) {
+					ctx.ui.notify(`Context cap is ${status}. No model selected.`, "info");
+					return;
+				}
+
+				const original = getOriginalContextWindow(model);
+				const effective = enabled ? getEffectiveContextWindow(model) : model.contextWindow;
+				ctx.ui.notify(
+					`Context cap is ${status}. Current model: ${modelKey(model)} (${formatTokens(effective)}/${formatTokens(original)} effective/original).`,
+					"info",
+				);
+				return;
+			}
+
+			ctx.ui.notify("Usage: /context-cap on | off | toggle | status", "warning");
+		},
+	});
+}

package/extensions/context-cap/package.json

@@ -0,0 +1,28 @@
+{
+  "name": "@diegopetrucci/pi-context-cap",
+  "version": "0.1.0",
+  "description": "A pi extension that caps effective model context windows at 200k tokens for earlier auto-compaction.",
+  "keywords": ["pi-package", "pi", "context", "compaction"],
+  "license": "MIT",
+  "repository": {
+    "type": "git",
+    "url": "git+https://github.com/diegopetrucci/pi-extensions.git",
+    "directory": "extensions/context-cap"
+  },
+  "files": [
+    "index.ts",
+    "README.md"
+  ],
+  "publishConfig": {
+    "access": "public"
+  },
+  "pi": {
+    "extensions": [
+      "index.ts"
+    ]
+  },
+  "peerDependencies": {
+    "@earendil-works/pi-ai": "*",
+    "@earendil-works/pi-coding-agent": "*"
+  }
+}

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@diegopetrucci/pi-extensions",
-  "version": "0.1.14",
-  "description": "A collection of pi extensions, including a minimal custom footer, an Amp-style oracle, a permission gate for dangerous bash commands, confirm-before-destructive session actions, and terminal notifications when pi is ready for input.",
+  "version": "0.1.15",
+  "description": "A collection of pi extensions, including a minimal custom footer, an Amp-style oracle, a 200k context cap for auto-compaction, a permission gate for dangerous bash commands, confirm-before-destructive session actions, and terminal notifications when pi is ready for input.",
   "keywords": ["pi-package", "pi", "terminal", "agent"],
   "license": "MIT",
   "repository": {
@@ -30,6 +30,7 @@
     "extensions": [
       "./extensions/minimal-footer/index.ts",
       "./extensions/oracle/index.ts",
+      "./extensions/context-cap/index.ts",
       "./extensions/permission-gate/index.ts",
       "./extensions/confirm-destructive/index.ts",
       "./extensions/notify/index.ts"