@estebanforge/pi-antigravity-bridge 1.6.0 → 1.6.2

package/CHANGELOG.md

@@ -2,6 +2,18 @@
 
 All notable changes to this project will be documented in this file.
 
+## [1.6.2] - 2026-09-19
+
+### Fixed
+
+- **Compatibility with host 0.86.0's normalized provider stream input.** Stream entry points now receive a branded `TranscriptContext` (`{ messages }` only): the `context.systemPrompt` / `context.tools` fields are gone, and only `normalizeContext()` produces the new type. G10 fresh-conversation delivery read `context.systemPrompt`, so every fresh agy conversation silently lost the system prompt block (operating instructions, AGENTS.md project context, tool-priority note); the prompt now comes from the transcript via `getCurrentSystemPrompt(context.messages)`. `emitToolUse` arguments are restricted to `Record<string, JsonValue>` per the tightened `ToolCall.arguments` contract (driver args arrive JSON-decoded off the wire, so the two call-site casts are sound). Test fixtures in `provider-{sysprompt,streaming,late-result,escalation}.test.ts` and the `scripts/test-provider.ts` smoke context wrap raw objects with `normalizeContext()`. Dev pins `@earendil-works/*` raised `^0.85.0` to `^0.86.0`; `tsc` clean, 360 tests green.
+
+## [1.6.1] - 2026-09-18
+
+### Added
+
+- **`thinking` / `effort` params on AskAntigravity (synonyms).** Delegation calls had no direct tier knob: the only lever was tier sugar inside `model` ("flash high"), so a caller told "peer review on high thinking" fell through to the configured default (medium) with no signal anything was dropped. The tool now takes `thinking` (pi vocabulary: minimal|low|medium|high|xhigh|max, clamped to agy's low|medium|high; unknown values fall back to low) plus `effort` as the same knob under agy's own name, so calling models reach for either word. An explicit level beats a tier embedded in `model` and the config default; passing both with different values errors; fixed-thinking families (Claude, GPT-OSS) ignore it because agy rejects `--effort` for them. `toAgyEffort` moved from `src/provider.ts` to `src/models.ts` so provider turns and the delegation tool share one clamp.
+
 ## [1.6.0] - 2026-09-14
 
 ### Added

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@estebanforge/pi-antigravity-bridge",
-  "version": "1.6.0",
+  "version": "1.6.2",
   "description": "Gemini provider for Pi on the Antigravity ACP server (official Google ACP) or the stream-json agy CLI. antigravity/* models in Pi's /model picker, no-patch MCP bridge: agy runs Pi's tools. ToS safe to use.",
   "keywords": [
     "pi-package",
@@ -59,10 +59,10 @@
     }
   },
   "devDependencies": {
-    "@earendil-works/pi-ai": "^0.85.0",
-    "@earendil-works/pi-coding-agent": "^0.85.0",
-    "@earendil-works/pi-server": "^0.85.0",
-    "@earendil-works/pi-tui": "^0.85.0",
+    "@earendil-works/pi-ai": "^0.86.0",
+    "@earendil-works/pi-coding-agent": "^0.86.0",
+    "@earendil-works/pi-server": "^0.86.0",
+    "@earendil-works/pi-tui": "^0.86.0",
     "@types/node": "^22.0.0",
     "tsx": "^4.19.0",
     "typebox": "^1.1.38",

package/src/ask-tool.ts

@@ -17,7 +17,7 @@ import * as path from "node:path";
 import type { ExtensionAPI } from "@earendil-works/pi-coding-agent";
 import { buildSessionContext, getAgentDir, keyHint } from "@earendil-works/pi-coding-agent";
 import { Text } from "@earendil-works/pi-tui";
-import { contentText } from "@earendil-works/pi-ai";
+import { contentText, type ThinkingLevel } from "@earendil-works/pi-ai";
 import { Type } from "typebox";
 import {
 	CONVERSATIONS_DIR,
@@ -26,7 +26,7 @@ import {
 } from "./discovery.js";
 import { loadConfig, type AgyMode, type ThinkingTier } from "./config.js";
 import { acquireBridgeSuppression } from "./mcp-registration.js";
-import { spawnAgyModelsRaw } from "./models.js";
+import { AGY_EFFORT_ORDER, spawnAgyModelsRaw, toAgyEffort } from "./models.js";
 
 // --- Constants -------------------------------------------------------------
 
@@ -80,7 +80,12 @@ EXECUTION MODES (param: mode):
 - **plan**: agy reviews and plans without writing. Use for cross-review and read-only tasks.
 - **accept-edits** (default): agy applies edits directly inside the workspace.
 
-COMPACT OUTPUT (param: digest): when true, the prompt is prefixed to request compact digests instead of full file contents. Defaults on for plan, off for accept-edits.`;
+COMPACT OUTPUT (param: digest): when true, the prompt is prefixed to request compact digests instead of full file contents. Defaults on for plan, off for accept-edits.
+
+THINKING LEVEL (params: thinking, effort - SYNONYMS for one knob):
+- pi calls it thinking, agy calls it effort. Same thing. Pass ONE of the two.
+- Values (pi vocabulary): minimal|low|medium|high|xhigh|max. Clamped to agy's low|medium|high; unknown values fall back to low. "peer review on high thinking" -> thinking: "high".
+- An explicit level beats a tier embedded in model ("flash high") and the configured default. Omit both for the configured default.`;
 
 // --- Types -----------------------------------------------------------------
 
@@ -176,6 +181,10 @@ export function resolveModel(
 	input: string,
 	entries: ModelEntry[],
 	defaultThinking: ThinkingTier,
+	/** Explicit thinking param (thinking/effort). Beats a tier embedded in
+	 *  the alias ("flash high") and the configured default; clamped to the
+	 *  family's real tiers, ignored for fixed-thinking families. */
+	preferredTier?: ThinkingTier,
 ): ResolvedModel | null {
 	const lower = input.toLowerCase().trim();
 
@@ -232,6 +241,7 @@ export function resolveModel(
 	if (familyTiers.size === 0) return toResolved(candidates[0].full, null);
 
 	const preferred =
+		preferredTier ??
 		tier ??
 		(familyTiers.has(defaultThinking) ? defaultThinking : FAMILY_DEFAULT_TIER[family]);
 	const chosenTier = nearestTier([...familyTiers], preferred);
@@ -293,6 +303,18 @@ export async function registerAskAntigravityTool(
 						"Model alias or exact id. Friendly: 'flash', 'pro', 'gemini'. Add a tier: 'flash high'. Pin a version: '3.5 flash'. Exact: 'gemini-3.6-flash-medium'. Omit for the configured default.",
 				}),
 			),
+			thinking: Type.Optional(
+				Type.String({
+					description:
+						"Thinking level (= agy effort tier). pi vocabulary: minimal|low|medium|high|xhigh|max, clamped to agy's low|medium|high (unknown values fall back to low). Overrides a tier embedded in `model`. Omit for the configured default.",
+				}),
+			),
+			effort: Type.Optional(
+				Type.String({
+					description:
+						"Alias for `thinking` (agy's own name for the same knob). Pass ONE of the two; different values on both is an error.",
+				}),
+			),
 			mode: Type.Optional(
 				Type.Union([Type.Literal("plan"), Type.Literal("accept-edits")], {
 					description:
@@ -327,8 +349,14 @@ export async function registerAskAntigravityTool(
 			// row identifies what will actually run, not just explicit args.
 			const cfg = loadConfig();
 			const requestedModel = (args.model as string | undefined)?.trim() || cfg.defaultModel;
+			const thinkingArg = (args.thinking as string | undefined) ?? (args.effort as string | undefined);
 			const resolved =
-				resolveModel(requestedModel, entries, cfg.defaultThinking) ?? { model: requestedModel };
+				resolveModel(
+					requestedModel,
+					entries,
+					cfg.defaultThinking,
+					thinkingArg ? toAgyEffort(thinkingArg as ThinkingLevel, AGY_EFFORT_ORDER) : undefined,
+				) ?? { model: requestedModel };
 			const thinking: ThinkingTier = resolved.effort ?? cfg.defaultThinking;
 			const mode: AgyMode = (args.mode as AgyMode | undefined) ?? "accept-edits";
 			const useDigest = typeof args.digest === "boolean" ? args.digest : mode === "plan";
@@ -415,8 +443,27 @@ export async function registerAskAntigravityTool(
 					details: emptyDetails(requestedModel),
 				};
 			}
+			if (
+				typeof params.thinking === "string" &&
+				typeof params.effort === "string" &&
+				params.thinking !== params.effort
+			) {
+				return {
+					content: [
+						{
+							type: "text",
+							text: "thinking and effort are synonyms for the same knob - pass one, not both with different values.",
+						},
+					],
+					details: emptyDetails(requestedModel),
+				};
+			}
+			const thinkingArg = (params.thinking as string | undefined) ?? (params.effort as string | undefined);
+			const preferredTier = thinkingArg
+				? toAgyEffort(thinkingArg as ThinkingLevel, AGY_EFFORT_ORDER)
+				: undefined;
 			const resolved =
-				resolveModel(requestedModel, entries, config.defaultThinking) ?? {
+				resolveModel(requestedModel, entries, config.defaultThinking, preferredTier) ?? {
 					model: requestedModel,
 				};
 

package/src/models.ts

@@ -21,7 +21,7 @@ import { spawn } from "node:child_process";
 import fs from "node:fs";
 import os from "node:os";
 import path from "node:path";
-import type { Api, Model, ThinkingLevelMap } from "@earendil-works/pi-ai";
+import type { Api, Model, ThinkingLevel, ThinkingLevelMap } from "@earendil-works/pi-ai";
 
 const DISCOVERY_TIMEOUT_MS = 8_000;
 
@@ -31,6 +31,47 @@ export type AgyEffort = "low" | "medium" | "high";
 /** low < medium < high, for sorting/clamping. */
 const EFFORT_RANK: Record<AgyEffort, number> = { low: 0, medium: 1, high: 2 };
 
+/** pi thinking-effort order mirrors agy's, for clamping. */
+export const AGY_EFFORT_ORDER: readonly AgyEffort[] = ["low", "medium", "high"];
+
+/** Map pi's thinking level onto one of the tiers `efforts` the base actually
+ *  supports, clamping to the nearest available. agy rejects an effort tier a
+ *  base doesn't list (e.g. medium on Pro), so we never emit one. Shared by the
+ *  provider turns AND the AskAntigravity delegation tool (thinking/effort
+ *  params): one vocabulary, one clamp. When pi sends no level we fall to the
+ *  first available tier. */
+export function toAgyEffort(
+	reasoning: ThinkingLevel | undefined,
+	efforts: readonly AgyEffort[],
+): AgyEffort {
+	let candidate: AgyEffort;
+	switch (reasoning) {
+		case "minimal":
+		case "low":
+			candidate = "low";
+			break;
+		case "medium":
+			candidate = "medium";
+			break;
+		case "high":
+		case "xhigh":
+		case "max":
+			candidate = "high";
+			break;
+		default:
+			candidate = efforts[0] ?? "low";
+	}
+	if (efforts.includes(candidate)) return candidate;
+	const i = AGY_EFFORT_ORDER.indexOf(candidate);
+	for (let j = i; j < AGY_EFFORT_ORDER.length; j++) {
+		if (efforts.includes(AGY_EFFORT_ORDER[j])) return AGY_EFFORT_ORDER[j];
+	}
+	for (let j = i - 1; j >= 0; j--) {
+		if (efforts.includes(AGY_EFFORT_ORDER[j])) return AGY_EFFORT_ORDER[j];
+	}
+	return efforts[0] ?? "low";
+}
+
 /** Split an agy slug into (base, tier). tier is null when the slug has no
  *  -high/-medium/-low suffix (claude-sonnet-4-6, gpt-oss-120b-medium's "medium"
  *  IS its suffix here, claude-opus-4-6-thinking is not a tier). */

package/src/provider.ts

@@ -19,13 +19,15 @@
 
 import {
 	createAssistantMessageEventStream,
+	getCurrentSystemPrompt,
 	type AssistantMessage,
 	type AssistantMessageEventStream,
-	type Context,
+	type JsonValue,
 	type Message,
 	type Model,
 	type SimpleStreamOptions,
 	type ThinkingLevel,
+	type TranscriptContext,
 	type Usage,
 } from "@earendil-works/pi-ai";
 import fs from "node:fs";
@@ -33,7 +35,7 @@ import type { Api } from "@earendil-works/pi-ai";
 import type { DriverActivity, TurnDriver, TurnHandle } from "./driver-types.js";
 import { toPiUsage } from "./stream-events.js";
 import { mapAgyToolToNative } from "./native-tools.js";
-import { type AgyEffort, type AgyModelEntry } from "./models.js";
+import { toAgyEffort, type AgyModelEntry } from "./models.js";
 import { SessionStore } from "./sessions.js";
 import { loadConfig } from "./config.js";
 import { GATE_MARKER, mapNativeToShadow, stripMarkerFields } from "./approval-gate.js";
@@ -59,7 +61,7 @@ function zeroUsage(): Usage {
 /** Extract the latest user message as a flat prompt string. agy maintains its
  *  own conversation history via --conversation, so we collapse pi's structured
  *  message to text. Returns null if the last message isn't a user message. */
-function extractUserPrompt(context: Context): string | null {
+function extractUserPrompt(context: TranscriptContext): string | null {
 	const last = context.messages[context.messages.length - 1];
 	if (!last || last.role !== "user") return null;
 	const content = last.content;
@@ -75,7 +77,7 @@ function extractUserPrompt(context: Context): string | null {
 /** Image blocks of the latest user message (pi-ai ImageContent: base64 data
  *  + mimeType). The ACP engine forwards them as typed content blocks; the
  *  stream-json CLI prompt is text-only, so its driver simply ignores these. */
-function extractImages(context: Context): Array<{ data: string; mimeType: string }> {
+function extractImages(context: TranscriptContext): Array<{ data: string; mimeType: string }> {
 	const last = context.messages[context.messages.length - 1];
 	if (!last || last.role !== "user" || typeof last.content === "string") return [];
 	return last.content
@@ -100,7 +102,8 @@ const DIGEST_PREAMBLE =
 
 // --- pi system prompt (G10) ----------------------------------------------------
 //
-// pi composes context.systemPrompt every turn: its own operating instructions
+// pi normalizes the system prompt into the transcript every turn (read via
+// getCurrentSystemPrompt): its own operating instructions
 // plus every AGENTS.md/CLAUDE.md it loaded (global agent dir first, then
 // ancestors). The provider used to drop it, so agy models never saw the user's
 // machine-level or project-level instructions. agy has no system-prompt flag
@@ -338,46 +341,6 @@ export interface StreamSimpleDeps {
 	log?: (event: string, data?: unknown, level?: "debug" | "info" | "warn" | "error") => void;
 }
 
-/** pi thinking-effort order mirrors agy's, for clamping. */
-const AGY_EFFORT_ORDER: readonly AgyEffort[] = ["low", "medium", "high"];
-
-/** Map pi's thinking level onto one of the tiers `efforts` the base actually
- *  supports, clamping to the nearest available. agy rejects an effort tier a
- *  base doesn't list (e.g. medium on Pro), so we never emit one. A base slug is
- *  invalid without --effort, so when pi sends no level we default to the
- *  middle tier (or the highest available). */
-export function toAgyEffort(
-	reasoning: ThinkingLevel | undefined,
-	efforts: readonly AgyEffort[],
-): AgyEffort {
-	let candidate: AgyEffort;
-	switch (reasoning) {
-		case "minimal":
-		case "low":
-			candidate = "low";
-			break;
-		case "medium":
-			candidate = "medium";
-			break;
-		case "high":
-		case "xhigh":
-		case "max":
-			candidate = "high";
-			break;
-		default:
-			candidate = efforts[0] ?? "low";
-	}
-	if (efforts.includes(candidate)) return candidate;
-	const i = AGY_EFFORT_ORDER.indexOf(candidate);
-	for (let j = i; j < AGY_EFFORT_ORDER.length; j++) {
-		if (efforts.includes(AGY_EFFORT_ORDER[j])) return AGY_EFFORT_ORDER[j];
-	}
-	for (let j = i - 1; j >= 0; j--) {
-		if (efforts.includes(AGY_EFFORT_ORDER[j])) return AGY_EFFORT_ORDER[j];
-	}
-	return efforts[0] ?? "low";
-}
-
 // --- G9: no-patch pi-tool round-trips -----------------------------------------
 //
 // The MCP bridge's onToolCall parks the call here instead of executing it:
@@ -961,13 +924,14 @@ function nextRtId(kind: "nat" | "wrap"): string {
 	return `${kind}-${++RT_SEQ}`;
 }
 
-/** Emit a complete toolCall block and end the pi call with toolUse. */
+/** Emit a complete toolCall block and end the pi call with toolUse.
+ *  pi 0.86 restricts ToolCall.arguments to JSON-compatible values. */
 function emitToolUse(
 	stream: AssistantMessageEventStream,
 	blocks: BlockState,
 	id: string,
 	name: string,
-	args: Record<string, unknown>,
+	args: Record<string, JsonValue>,
 ): void {
 	const partial = blocks.partial;
 	closeThinking(stream, blocks);
@@ -1090,7 +1054,7 @@ function acpEditFileArg(args: Record<string, unknown>): string | undefined {
 				emitToolUse(stream, blocks, id, mapped.tool, {
 					reasoning: `re-exec of agy ${activity.name} for display`,
 					...mapped.args,
-				});
+				} as Record<string, JsonValue>);
 				return "parked";
 			}
 			{
@@ -1107,7 +1071,9 @@ function acpEditFileArg(args: Record<string, unknown>): string | undefined {
 		case "bridge_call": {
 			// Park the pi call: real tool name + args, toolUse stopReason. pi
 			// executes; the toolResult returns on the next stream call.
-			emitToolUse(stream, blocks, activity.callId, activity.name, activity.args);
+			// Driver args are JSON-decoded off the agy wire, so the JsonValue
+			// cast is sound.
+			emitToolUse(stream, blocks, activity.callId, activity.name, activity.args as Record<string, JsonValue>);
 			return "parked";
 		}
 	}
@@ -1117,7 +1083,7 @@ function acpEditFileArg(args: Record<string, unknown>): string | undefined {
 async function runTurnDriver(
 	stream: AssistantMessageEventStream,
 	model: Model<Api>,
-	context: Context,
+	context: TranscriptContext,
 	options: SimpleStreamOptions | undefined,
 	entries: AgyModelEntry[],
 	store: SessionStore,
@@ -1219,7 +1185,9 @@ async function runTurnDriver(
 		// Fresh conversation only: agy stores the block in its own history, so
 		// re-sending it every turn would bloat each prompt and bust the cache.
 		const sysPrompt =
-			config.systemPrompt && !existing?.conversationId ? context.systemPrompt : undefined;
+			config.systemPrompt && !existing?.conversationId
+				? getCurrentSystemPrompt(context.messages) || undefined
+				: undefined;
 		const fullPrompt =
 			late.length > 0
 				? buildLateResultPrompt(late, prompt || undefined)
@@ -1298,7 +1266,7 @@ async function runTurnDriver(
  *  fallback). */
 export function createStreamSimple(
 	deps: StreamSimpleDeps,
-): (model: Model<Api>, context: Context, options?: SimpleStreamOptions) => AssistantMessageEventStream {
+): (model: Model<Api>, context: TranscriptContext, options?: SimpleStreamOptions) => AssistantMessageEventStream {
 	const { entries, store, roundTrips } = deps;
 
 	return function streamSimple(model, context, options) {