@callstack/apex 0.3.0 → 0.3.2

package/README.md

@@ -148,15 +148,43 @@ their own backup first, so an undo is itself reversible. Apex CLI records each b
 | OpenCode | `~/.config/opencode/opencode.json` or `.jsonc` | Adds `provider.callstack.ai`, OpenAI-compatible transport, model and environment-key reference; preserves the default model. A config already in OpenCode 2's `providers` format gets the v2 entry instead, and you add the key with `/connect`. |
 | Codex 0.134.0+ | `~/.codex/callstack_ai.config.toml` | Adds a self-contained Responses API provider/profile; leaves base config and default model untouched. |
 | Claude Code | `~/.claude/settings.json` | Merges attribution flag only; use `apex run claude` for gateway credentials and model selection. |
-| Pi | `~/.pi/agent/models.json` | Adds `providers.callstack` with Chat Completions, Apex and `$CALLSTACK_AUTH_TOKEN`; preserves other models. Requires a Pi version supporting `$VAR` key interpolation. |
+| Pi | `~/.pi/agent/models.json` and `settings.json` | Adds `providers.callstack` with Chat Completions, Apex and `$CALLSTACK_AUTH_TOKEN`; saves `medium` as Apex's per-model thinking default for direct launches. Preserves other models and their defaults. Requires a Pi version supporting `$VAR` key interpolation and `modelThinkingLevels`. |
 | Cursor | Guided setup | Prints endpoint, API-key and custom-model steps; does not modify private editor storage. |
 | VS Code / Copilot | Guided setup | Detects VS Code, not whether Copilot is installed; prints custom-endpoint steps and model JSON, retaining the editor-generated secret reference. |
-| Vercel AI SDK / Eve | Guided setup | Detected from the `package.json` in the directory you run Apex CLI from (`ai`, `@ai-sdk/openai` or `eve`); prints the connector snippet, which reads the key from `CALLSTACK_AUTH_TOKEN`, and the Eve context-window setting. Nothing is written into the project. |
-
-OpenCode, Pi, Codex and the Copilot model JSON also declare what Apex can do, in each tool's own
-fields: a 262,144-token context window, 16,384 output tokens, tool calling, image input, and the
-`none`/`low`/`medium`/`xhigh` reasoning efforts (default `medium`). Cursor and Claude Code have no
-such fields for a custom model.
+| Vercel AI SDK / Eve | Guided setup | Detected from the `package.json` in the directory you run Apex CLI from (`ai`, `@ai-sdk/openai` or `eve`); prints the connector snippet, which reads the key from `CALLSTACK_AUTH_TOKEN`. Nothing is written into the project. |
+
+Apex configs declare tool calling, image input, and the supported
+`none`/`low`/`medium`/`xhigh` reasoning efforts (default `medium`) where supported.
+OpenCode, Pi, Copilot, Claude's launcher, and the AI SDK guidance retain a 32,768-token output
+budget. OpenCode, Pi, Codex, Copilot, and Claude's launcher declare the model's native
+262,144-token context window using each app's documented fields. Input limits and compaction
+thresholds use the harness defaults; Apex CLI does not set a separate input budget or compaction
+reserve.
+Pi's per-model default applies to fresh direct launches and model selection; `apex run pi` also
+selects `medium` explicitly. Resumed Pi sessions retain their saved thinking level.
+
+The gateway model ID, authentication, `medium` default, and `off` → `none` translation remain
+Callstack-specific.
+
+| Harness | Context field | Output field | Official reference |
+| --- | --- | --- | --- |
+| OpenCode | `limit.context` | `limit.output` | [Custom providers](https://opencode.ai/docs/providers/) |
+| Pi | `contextWindow` | `maxTokens` | [Custom models](https://pi.dev/docs/latest/models) |
+| Codex | `model_context_window` | No output-limit config key | [OpenAI configuration reference](https://developers.openai.com/codex/config-reference/) |
+| Claude Code launcher | `CLAUDE_CODE_MAX_CONTEXT_TOKENS` | `CLAUDE_CODE_MAX_OUTPUT_TOKENS` | [Environment variables](https://code.claude.com/docs/en/env-vars) |
+| VS Code / Copilot | `contextWindow` | `maxOutputTokens` | [Custom endpoint models](https://code.visualstudio.com/docs/agent-customization/language-models) |
+
+The Claude launcher sets context/output limits only in the child process and does not set compaction
+window or percentage controls. Pi retains its native compaction settings. The AI SDK
+snippet sets the output budget without inventing a context-window option.
+
+Run `apex init` again after updating the CLI to preview upgrades to an existing Apex setup; pass
+`--apply` to save them. Apex context/output metadata and reasoning mappings are updated in place. Older OpenCode input
+limits and Codex compaction overrides are removed so each harness uses its own compaction defaults.
+Codex's obsolete `model_max_output_tokens` key from older Apex versions is removed, and its dedicated
+profile becomes self-contained. Unrelated settings, models, credentials and comments are preserved;
+the ordinary backups and `apex undo` also cover upgrades. `none` remains supported: the gateway
+translates it to thinking off.
 
 Respects `XDG_CONFIG_HOME`, `CODEX_HOME`, `CLAUDE_CONFIG_DIR`, `PI_CODING_AGENT_DIR`,
 `XDG_STATE_HOME`/`APEX_STATE_DIR`, and Windows `APPDATA`. Project-specific settings and custom
@@ -201,9 +229,11 @@ Apex CLI will not break what you already have.
 How that works underneath:
 
 - Previews print paths and key/value changes, never secrets.
-- All selected configurations are parsed before any write. Malformed files, conflicting Codex
-  profile keys and symlinked files or directories are refused, and one bad assistant blocks the
+- All selected configurations are parsed before any write. Malformed files and symlinked files or
+  directories are refused, and one bad assistant blocks the
   whole batch instead of half-applying it.
+- The dedicated Codex Apex profile's owned keys are upgraded rather than rejected as conflicts;
+  the base `config.toml` and other profiles remain untouched.
 - JSON/JSONC edits preserve comments and unrelated keys. The Callstack endpoint is set to the
   documented one. A key the assistant already has (in OpenCode's own key store, or in Pi's
   `models.json`, where Apex CLI 0.2 put them) is kept, so upgrading never leaves an assistant

package/dist/cli.js

@@ -2212,26 +2212,123 @@ function editJson(text, changes, path) {
 	parseJson(next, path);
 	return next;
 }
-function addToml(text, desired, path) {
-	let existing;
+function updateToml(text, desired, path, obsolete = []) {
+	const before = text ?? "";
 	try {
-		existing = parse$1(text ?? "");
+		parse$1(before);
 	} catch {
 		throw new Error(`Invalid TOML in ${path}; nothing overwritten.`);
 	}
-	const missing = {};
-	for (const [key, value] of Object.entries(desired)) if (!Object.hasOwn(existing, key)) missing[key] = value;
-	else if (!isDeepStrictEqual(existing[key], value)) throw new Error(`Conflicting ${key} in ${path}. Review it manually; nothing overwritten.`);
-	if (!Object.keys(missing).length) return text ?? "";
-	const scalars = Object.fromEntries(Object.entries(missing).filter(([, value]) => !object(value)));
-	const tables = Object.fromEntries(Object.entries(missing).filter(([, value]) => object(value)));
-	const parts = [];
-	if (Object.keys(scalars).length) parts.push(stringify(scalars));
-	if ((text ?? "").trim()) parts.push(text);
-	if (Object.keys(tables).length) parts.push(stringify(tables));
-	const next = `${parts.map((part) => part.replace(/\s+$/, "")).join("\n\n")}\n`;
-	parse$1(next);
-	return next;
+	const eol = before.includes("\r\n") ? "\r\n" : "\n";
+	const keyText = (key) => /^[\w-]+$/.test(key) ? key : JSON.stringify(key);
+	const lines = before.split(/\r?\n/);
+	const valueAt = (value, keys) => keys.reduce((parent, key) => parent?.[key], value);
+	const merge = (old, next) => object(next) ? Object.fromEntries(Object.entries({
+		...object(old) ? old : {},
+		...next
+	}).map(([key, value]) => [key, Object.hasOwn(next, key) ? merge(old?.[key], next[key]) : value])) : next;
+	const inline = (value) => value instanceof Date ? stringify({ value }).trim().replace(/^value\s*=\s*/, "") : object(value) ? `{ ${Object.entries(value).map(([key, item]) => `${JSON.stringify(key)} = ${inline(item)}`).join(", ")} }` : Array.isArray(value) ? `[${value.map(inline).join(", ")}]` : stringify({ value }).trim().replace(/^value\s*=\s*/, "");
+	const statements = [];
+	let section = [];
+	for (let i = 0; i < lines.length; i++) {
+		if (!lines[i].trim() || lines[i].trimStart().startsWith("#")) {
+			statements.push({
+				text: lines[i],
+				section
+			});
+			continue;
+		}
+		let block = lines[i], parsed;
+		for (;;) try {
+			parsed = parse$1(block);
+			break;
+		} catch {
+			if (++i >= lines.length) throw new Error(`Cannot safely update TOML in ${path}.`);
+			block += eol + lines[i];
+		}
+		if (block.trimStart().startsWith("[")) {
+			section = [];
+			let cursor = parsed;
+			while (object(cursor) && Object.keys(cursor).length === 1) {
+				const key = Object.keys(cursor)[0];
+				section.push(key);
+				cursor = cursor[key];
+			}
+			statements.push({
+				text: block,
+				section,
+				header: true,
+				array: Array.isArray(cursor)
+			});
+			continue;
+		}
+		let equal = 0, keyQuote = null;
+		for (; equal < block.length; equal++) {
+			const char = block[equal];
+			if (keyQuote === "\"" && char === "\\") {
+				equal++;
+				continue;
+			}
+			if (keyQuote) {
+				if (char === keyQuote) keyQuote = null;
+			} else if (char === "\"" || char === "'") keyQuote = char;
+			else if (char === "=") break;
+		}
+		const rawKey = block.slice(0, equal).trim();
+		let cursor = parse$1(`${rawKey} = 0`), keys = [];
+		while (object(cursor)) {
+			const key = Object.keys(cursor)[0];
+			keys.push(key);
+			cursor = cursor[key];
+		}
+		const full = [...section, ...keys];
+		if (obsolete.includes(full.join("."))) continue;
+		const next = valueAt(desired, full);
+		if (next !== void 0 && !isDeepStrictEqual(valueAt(parsed, keys), merge(valueAt(parsed, keys), next))) {
+			let comment = "", quote = null;
+			if (!block.includes(eol)) for (let j = equal + 1; j < block.length; j++) {
+				const char = block[j];
+				if (quote === "\"" && char === "\\") {
+					j++;
+					continue;
+				}
+				if (quote) {
+					if (char === quote) quote = null;
+				} else if (char === "\"" || char === "'") quote = char;
+				else if (char === "#") {
+					comment = ` ${block.slice(j)}`;
+					break;
+				}
+			}
+			block = `${block.slice(0, equal + 1)} ${inline(merge(valueAt(parsed, keys), next))}${comment}`;
+		}
+		statements.push({
+			text: block,
+			section
+		});
+	}
+	const leaves = (value, prefix = []) => Object.entries(value).flatMap(([key, item]) => object(item) ? leaves(item, [...prefix, key]) : [[[...prefix, key], item]]);
+	for (const [keys, value] of leaves(desired)) {
+		const current = parse$1(statements.map((item) => item.text).join(eol));
+		if (isDeepStrictEqual(valueAt(current, keys), value)) continue;
+		let header = -1, prefix = [];
+		statements.forEach((item, index) => {
+			if (item.header && !item.array && item.section.length < keys.length && item.section.every((key, i) => key === keys[i]) && item.section.length > prefix.length) {
+				header = index;
+				prefix = item.section;
+			}
+		});
+		let end = statements.findIndex((item, index) => index > header && item.header);
+		if (end === -1) end = statements.length;
+		statements.splice(end, 0, {
+			text: `${keys.slice(prefix.length).map(keyText).join(".")} = ${inline(value)}`,
+			section: prefix
+		});
+	}
+	const after = statements.map((item) => item.text).join(eol);
+	const result = parse$1(after);
+	for (const [keys, value] of leaves(desired)) if (!isDeepStrictEqual(valueAt(result, keys), value)) throw new Error(`Cannot safely update ${keys.join(".")} in ${path}.`);
+	return after === before ? before : `${after.replace(/\s+$/, "")}${eol}`;
 }
 function sha256(text) {
 	return createHash("sha256").update(text, "utf8").digest("hex");
@@ -2311,8 +2408,7 @@ const RUNNABLE = [
 	"pi"
 ];
 const CONTEXT_WINDOW = 262144;
-const MAX_INPUT = 24e4;
-const MAX_OUTPUT = 16384;
+const MAX_OUTPUT = 32768;
 const EFFORTS = [
 	"none",
 	"low",
@@ -2330,7 +2426,6 @@ const OPENCODE_MODEL = {
 	},
 	limit: {
 		context: CONTEXT_WINDOW,
-		input: MAX_INPUT,
 		output: MAX_OUTPUT
 	},
 	options: { reasoningEffort: "medium" },
@@ -2349,7 +2444,6 @@ const OPENCODE_V2_PROVIDER = {
 		},
 		limit: {
 			context: CONTEXT_WINDOW,
-			input: MAX_INPUT,
 			output: MAX_OUTPUT
 		},
 		settings: { reasoningEffort: "medium" },
@@ -2372,7 +2466,8 @@ const PI_MODEL = {
 		xhigh: "xhigh",
 		max: null
 	},
-	contextWindow: CONTEXT_WINDOW
+	contextWindow: CONTEXT_WINDOW,
+	maxTokens: MAX_OUTPUT
 };
 async function exists(path) {
 	try {
@@ -2497,7 +2592,34 @@ async function planAssistant(assistant) {
 			const storedKey = await openCodeHasKey(assistant.authFile);
 			await json(hasJsonc ? "opencode.jsonc" : "opencode.json", (existing) => {
 				v2 = Object.hasOwn(existing, "providers");
-				if (v2) return [[["providers", "callstack.ai"], OPENCODE_V2_PROVIDER]];
+				if (v2) return [
+					[[
+						"providers",
+						"callstack.ai",
+						"name"
+					], OPENCODE_V2_PROVIDER.name],
+					[[
+						"providers",
+						"callstack.ai",
+						"package"
+					], OPENCODE_V2_PROVIDER.package],
+					[[
+						"providers",
+						"callstack.ai",
+						"settings",
+						"baseURL"
+					], BASE_URL],
+					...Object.entries(OPENCODE_V2_PROVIDER.models[MODEL]).map(([key, value]) => [[
+						"providers",
+						"callstack.ai",
+						"models",
+						MODEL,
+						key
+					], key === "settings" ? {
+						...existing.providers?.["callstack.ai"]?.models?.[MODEL]?.settings,
+						...value
+					} : value])
+				];
 				return [
 					[[
 						"provider",
@@ -2515,18 +2637,22 @@ async function planAssistant(assistant) {
 						"options",
 						"baseURL"
 					], BASE_URL],
-					...storedKey ? [] : [[[
+					...storedKey || existing.provider?.["callstack.ai"]?.options?.apiKey ? [] : [[[
 						"provider",
 						"callstack.ai",
 						"options",
 						"apiKey"
 					], "{env:CALLSTACK_AUTH_TOKEN}"]],
-					[[
+					...Object.entries(OPENCODE_MODEL).map(([key, value]) => [[
 						"provider",
 						"callstack.ai",
 						"models",
-						MODEL
-					], OPENCODE_MODEL]
+						MODEL,
+						key
+					], key === "options" ? {
+						...existing.provider?.["callstack.ai"]?.models?.[MODEL]?.options,
+						...value
+					} : value])
 				];
 			});
 			if (v2) changes[0].steps = ["OpenCode 2 stores provider keys itself: run /connect in OpenCode, pick callstack.ai → \"Manually enter API Key\", and paste your Callstack key."];
@@ -2557,9 +2683,14 @@ async function planAssistant(assistant) {
 						"providers",
 						"callstack",
 						"models"
-					], models.some((model) => model.id === "callstack/Apex") ? models.map((model) => model.id === "callstack/Apex" ? PI_MODEL : model) : [...models, PI_MODEL]]
+					], models.some((model) => model.id === "callstack/Apex") ? models.map((model) => model.id === "callstack/Apex" ? {
+						...model,
+						...PI_MODEL,
+						...model.compat ? { compat: Object.fromEntries(Object.entries(model.compat).filter(([key]) => key !== "reasoningEffortMap")) } : {}
+					} : model) : [...models, PI_MODEL]]
 				];
 			});
+			await json("settings.json", [[["modelThinkingLevels", `callstack/${MODEL}`], "medium"]]);
 			break;
 		case "claude":
 			await json("settings.json", [[["env", "CLAUDE_CODE_ATTRIBUTION_HEADER"], "0"]]);
@@ -2567,7 +2698,7 @@ async function planAssistant(assistant) {
 		case "codex": {
 			const path = join(assistant.directory, "callstack_ai.config.toml");
 			const before = await readConfig(path);
-			const after = addToml(before, {
+			const after = updateToml(before, {
 				model_provider: "callstack_ai",
 				model: MODEL,
 				model_context_window: CONTEXT_WINDOW,
@@ -2579,7 +2710,7 @@ async function planAssistant(assistant) {
 					wire_api: "responses",
 					requires_openai_auth: false
 				} }
-			}, path);
+			}, path, ["model_max_output_tokens", "model_auto_compact_token_limit"]);
 			changes.push({
 				path,
 				before,
@@ -2609,7 +2740,9 @@ function launchOptions(id, env = process.env) {
 				"--provider",
 				"callstack",
 				"--model",
-				MODEL
+				MODEL,
+				"--thinking",
+				"medium"
 			],
 			env: childEnv
 		};
@@ -2629,7 +2762,9 @@ function launchOptions(id, env = process.env) {
 				ANTHROPIC_DEFAULT_OPUS_MODEL: MODEL,
 				ANTHROPIC_DEFAULT_SONNET_MODEL: MODEL,
 				ANTHROPIC_DEFAULT_HAIKU_MODEL: MODEL,
-				CLAUDE_CODE_SUBAGENT_MODEL: MODEL
+				CLAUDE_CODE_SUBAGENT_MODEL: MODEL,
+				CLAUDE_CODE_MAX_CONTEXT_TOKENS: String(CONTEXT_WINDOW),
+				CLAUDE_CODE_MAX_OUTPUT_TOKENS: String(MAX_OUTPUT)
 			});
 			return {
 				args: ["--model", MODEL],
@@ -2663,7 +2798,11 @@ const AI_SDK_SNIPPET = [
 	`  baseURL: '${BASE_URL}',`,
 	"  apiKey: process.env.CALLSTACK_AUTH_TOKEN,",
 	"});",
-	`// apex('${MODEL}') with generateText / streamText.`,
+	"const requestOptions = {",
+	`  model: apex('${MODEL}'),`,
+	`  maxOutputTokens: ${MAX_OUTPUT},`,
+	"};",
+	"// Pass requestOptions with your prompt to generateText / streamText.",
 	`// Reasoning effort (${EFFORTS.join(", ")}):`,
 	"// providerOptions: { callstack: { reasoningEffort: 'medium' } }"
 ];
@@ -2674,11 +2813,7 @@ const MANUAL = {
 		"Keep the generated apiKey secret reference; add this object to its models array:",
 		...COPILOT_MODEL.split("\n")
 	],
-	"ai-sdk": [
-		"Apex CLI never writes your key into a project. Keep it in CALLSTACK_AUTH_TOKEN (for example in a .env that git ignores) and create the provider with:",
-		...AI_SDK_SNIPPET,
-		`Eve: set modelContextWindowTokens: ${CONTEXT_WINDOW} on defineAgent, or compaction will fail to compile.`
-	]
+	"ai-sdk": ["Apex CLI never writes your key into a project. Keep it in CALLSTACK_AUTH_TOKEN (for example in a .env that git ignores) and create the provider with:", ...AI_SDK_SNIPPET]
 };
 const MANUAL_IDS = Object.keys(MANUAL);
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@callstack/apex",
-  "version": "0.3.0",
+  "version": "0.3.2",
   "description": "Detects the AI coding assistants installed on your machine and configures them to use Callstack Apex.",
   "license": "MIT",
   "keywords": [
@@ -24,6 +24,7 @@
     "build": "tsdown",
     "pretest": "npm run build",
     "prepack": "npm run build",
+    "prepublishOnly": "npm test",
     "test": "node --test",
     "check": "for f in src/*.js bin/apex.js; do node --check \"$f\" || exit 1; done && npm run build"
   },