baro-ai 0.43.3 → 0.44.1

package/README.md

@@ -69,7 +69,31 @@ flowchart LR
     F --> PR([Pull Request])
 ```
 
-Every story is one **Claude Code subprocess** (or one Mozaik-native OpenAI session) — auth inherits from your existing setup, no API key plumbing.
+Every story is one CLI subprocess — Claude Code, OpenAI Codex CLI, or a Mozaik-native OpenAI Responses session, depending on `--llm`. Auth inherits from whichever CLI you already have signed in, no API key plumbing.
+
+## Three LLM backends, one DAG
+
+```bash
+baro --llm claude  "Your goal"   # default — Claude Code on Anthropic Max subscription
+baro --llm codex   "Your goal"   # OpenAI Codex CLI on ChatGPT Pro/Plus subscription
+baro --llm openai  "Your goal"   # Mozaik-native OpenAI Responses (per-call API billing)
+baro --llm hybrid  "Your goal"   # Claude on Architect/Planner/Surgeon, Codex on Story/Critic
+```
+
+Same orchestration. Same DAG. Same prompts. The only thing that moves is which provider every agent talks to. `--llm hybrid` is the new default-recommendation for serious runs — Claude where the upstream plan matters, Codex for the parallel story+critic work that dominates the budget.
+
+Each phase has its own override flag if you want to mix it yourself:
+
+```bash
+baro --architect-llm claude \
+     --planner-llm  claude \
+     --story-llm    codex  \
+     --critic-llm   codex  \
+     --surgeon-llm  claude \
+     "Your goal"
+```
+
+Full breakdown at [docs.baro.rs/llm-providers](https://docs.baro.rs/llm-providers) — provider economics, per-phase routing, the side-by-side benchmark across three real tasks: [**I tested Claude Code vs OpenAI Codex in my parallel agent setup. Then I built a hybrid.**](https://jigjoy.ai/blog/claude-code-vs-codex-baro)
 
 ## Recent real run
 
@@ -82,7 +106,7 @@ Every story is one **Claude Code subprocess** (or one Mozaik-native OpenAI sessi
 | **Architect** | One Opus call before planning — emits a `DecisionDocument` that pins every cross-cutting design decision (file paths, schemas, API shapes, library choices) so 30 parallel agents don't each invent their own |
 | **Planner** | Decomposes the goal into a story DAG, with the DecisionDocument already pinned |
 | **Conductor** | State machine that drives the run by reacting to bus events |
-| **StoryAgent** | One Claude Code subprocess per story; multi-turn loop until story completes |
+| **StoryAgent** | One CLI subprocess per story (Claude Code / Codex / OpenAI Responses, picked by `--llm` or `--story-llm`); multi-turn loop until story completes |
 | **Critic** | Per-turn evaluator (Haiku). On fail verdict, injects corrective feedback as the agent's next turn |
 | **Sentry** | Flags overlapping Edit/Write tool calls across concurrent stories |
 | **Librarian** | Indexes one agent's Read/Grep findings so siblings don't redo the exploration |
@@ -96,16 +120,22 @@ Bus is open. CI deployers, Slack notifiers, ticket triggers — all new particip
 ```bash
 npm install -g baro-ai
 
-# Full run (default — Architect + Planner + parallel Story Agents)
+# Full run (default — Claude on every phase via Claude Code CLI)
 baro "Migrate the hardcoded category data to a backend dictionary"
 
 # Trivial goal — skip Architect + Critic + Surgeon, single story
 baro --quick "fix the typo on line 42 of README.md"
 
-# Route every phase through GPT-5.5 instead of Claude
+# Codex everywhere (ChatGPT Pro/Plus subscription, ~3-11× cheaper per run than Claude)
+baro --llm codex "Refactor the database layer"
+
+# Per-phase routing — Claude upstream (tight plans), Codex downstream (cheap writes)
+baro --llm hybrid "Add WebSocket support across api and frontend"
+
+# Route every phase through GPT-5.5 (Mozaik-native OpenAI API)
 OPENAI_API_KEY=sk-... baro --llm openai "Refactor the database layer"
 
-# Limit parallelism (Anthropic plan tiers cap concurrency)
+# Limit parallelism (plan-tier concurrency caps)
 baro --parallel 3 "Add unit tests for the auth module"
 
 # Dry-run first, execute later
@@ -134,7 +164,11 @@ For a deeper side-by-side on a real refactor, see [baro vs Claude Code `/goal`](
 
 ## Requirements
 
-- [Claude CLI](https://docs.anthropic.com/en/docs/claude-cli) authenticated (for `--llm claude`, the default) **or** `OPENAI_API_KEY` set (for `--llm openai`)
+- At least one of:
+  - [Claude CLI](https://docs.anthropic.com/en/docs/claude-cli) authenticated (for `--llm claude`, the default)
+  - [OpenAI Codex CLI](https://github.com/openai/codex) authenticated (for `--llm codex`)
+  - `OPENAI_API_KEY` set (for `--llm openai`)
+  - Both Claude CLI **and** Codex CLI authenticated (for `--llm hybrid`)
 - Node.js 20+
 - macOS (arm64/x64), Linux (x64/arm64), Windows (x64)
 - `gh` CLI (optional, for automatic PR creation)

package/dist/cli.mjs

@@ -14467,17 +14467,26 @@ async function orchestrate(config) {
   const env = new AgenticEnvironment();
   const emitTui = config.emitTuiEvents ?? true;
   const llm = config.llm ?? "claude";
-  if (llm === "openai") {
+  const storyLlm = config.storyLlm ?? llm;
+  const criticLlm = config.criticLlm ?? llm;
+  const surgeonLlm = config.surgeonLlm ?? llm;
+  const isHybrid = (/* @__PURE__ */ new Set([storyLlm, criticLlm, surgeonLlm, llm])).size > 1;
+  if (isHybrid) {
     process.stderr.write(
-      "[orchestrate] llm=openai: Architect, Planner, Critic, Surgeon, StoryAgent all running through Mozaik's native OpenAI runner (gpt-5.x).\n"
+      `[orchestrate] hybrid routing: story=${storyLlm} critic=${criticLlm} surgeon=${surgeonLlm} (default=${llm})
+`
+    );
+  } else if (llm === "openai") {
+    process.stderr.write(
+      "[orchestrate] llm=openai: Story, Critic, Surgeon all running through Mozaik's native OpenAI runner (gpt-5.x).\n"
     );
   } else if (llm === "codex") {
     process.stderr.write(
-      "[orchestrate] llm=codex: every LLM phase shells out to `codex exec --json` (ChatGPT subscription path). Architect / Planner / Critic / Surgeon / StoryAgent all running through Codex.\n"
+      "[orchestrate] llm=codex: Story, Critic, Surgeon all shelling out to `codex exec --json` (ChatGPT subscription path).\n"
     );
   } else {
     process.stderr.write(
-      "[orchestrate] llm=claude: every LLM phase shells out to the Claude Code CLI.\n"
+      "[orchestrate] llm=claude: Story, Critic, Surgeon all shelling out to the Claude Code CLI.\n"
     );
   }
   if (config.auditLogPath) {
@@ -14518,12 +14527,12 @@ async function orchestrate(config) {
         }))
       };
     };
-    if (llm === "openai") {
+    if (surgeonLlm === "openai") {
       surgeon = new SurgeonOpenAI({
         snapshot,
         model: config.surgeonModel ?? "gpt-5.5"
       });
-    } else if (llm === "codex") {
+    } else if (surgeonLlm === "codex") {
       surgeon = new SurgeonCodex({
         snapshot,
         useLlm: config.surgeonUseLlm ?? true,
@@ -14544,12 +14553,12 @@ async function orchestrate(config) {
     const targets = new Map(
       prd.userStories.filter((s) => s.acceptance && s.acceptance.length > 0).map((s) => [s.id, s.acceptance])
     );
-    if (llm === "openai") {
+    if (criticLlm === "openai") {
       critic = new CriticOpenAI({
         targets,
         model: config.criticModel ?? "gpt-5.4-mini"
       });
-    } else if (llm === "codex") {
+    } else if (criticLlm === "codex") {
       critic = new CriticCodex({
         targets,
         model: config.criticModel
@@ -14631,7 +14640,7 @@ async function orchestrate(config) {
   conductor.join(env);
   const storyFactory = new StoryFactory({
     cwd: config.cwd,
-    llm,
+    llm: storyLlm,
     openaiModel: config.storyModel ?? "gpt-5.5",
     storyModelOverride: config.storyModel
   });
@@ -14983,6 +14992,22 @@ function parseArgs(argv) {
         args.llm = v;
         break;
       }
+      case "--story-llm":
+      case "--critic-llm":
+      case "--surgeon-llm": {
+        const v = required(argv, ++i, a);
+        if (v !== "claude" && v !== "openai" && v !== "codex") {
+          process.stderr.write(
+            `[cli] ${a} must be 'claude' | 'openai' | 'codex', got '${v}'
+`
+          );
+          process.exit(2);
+        }
+        if (a === "--story-llm") args.storyLlm = v;
+        else if (a === "--critic-llm") args.criticLlm = v;
+        else args.surgeonLlm = v;
+        break;
+      }
       default:
         process.stderr.write(`[cli] unknown flag: ${a}
 `);
@@ -15062,6 +15087,9 @@ async function main() {
     surgeonModel: args.surgeonModel,
     intraLevelDelaySecs: args.intraLevelDelaySecs,
     llm: args.llm,
+    storyLlm: args.storyLlm,
+    criticLlm: args.criticLlm,
+    surgeonLlm: args.surgeonLlm,
     storyModel: args.storyModel
   };
   if (args.llm === "openai" && !process.env.OPENAI_API_KEY) {