@tryarcanist/cli 0.1.254 → 0.1.255

package/README.md

@@ -189,7 +189,6 @@ arcanist sessions create your-org/your-repo "finish this PR" --continue-pr https
 printf "add tests" | arcanist sessions create your-org/your-repo --prompt-stdin --json
 printf "add tests" | arcanist sessions create your-org/your-repo --prompt-stdin --wait
 arcanist sessions create your-org/your-repo - --model gpt-5.5
-arcanist sessions create your-org/your-repo "port to claude" --backend claude_code --model claude-opus-4-8
 arcanist sessions create your-org/your-repo "refactor auth" --reasoning-effort xhigh
 arcanist sessions create your-org/your-repo "use the saved login" --browser-identity identity-id
 arcanist sessions create your-org/your-repo "start signed out" --browser-identity none
@@ -198,7 +197,7 @@ arcanist sessions create your-org/your-repo "review the trace" --uploaded-file t
 arcanist sessions create your-org/your-repo "retry-safe create" --idempotency-key 1f0e6f1a-...
 ```
 
-`--backend` picks the agent runtime backend: `codex` (default) or `claude_code`. `--model` must be valid for the chosen backend and the CLI rejects a mismatch before any network call; run `arcanist models list` for the selectable models per backend. When `--model` is omitted the backend default is used. Non-codex backends require the matching provider credential configured in Settings.
+`--model` pins the model for the session; the CLI rejects an unknown model before any network call. Run `arcanist models list` for the selectable models. When `--model` is omitted the default is used. Claude models run on the Codex runtime and require the Anthropic provider credential configured in Settings.
 
 `--wait` blocks until the created prompt finishes, prints the resulting PR/branch line in human-readable mode when it is already available, and exits non-zero if the prompt finishes with `status: failed`, making it suitable for cron or other schedulers that alert on command failure. JSON mode waits quietly and prints the create payload after the prompt completes successfully. `--poll-interval <ms>` tunes the completion check frequency.
 
@@ -226,7 +225,7 @@ The CLI derives separate session and prompt idempotency keys from the provided v
 
 `--onboarding` creates an onboarding session: the agent authors the repo's `.arcanist.json` (`appRuntime` runtime contract plus optional legacy `verify` metadata), a root `REVIEW.md` with repo-specific review checks (skipped only when the repo yields none), `.arcanist/` runtime files as needed, and `ARCANIST.md` only when the repo has no usable project doc, proves what it can inside its sandbox, and opens the setup PR ready for review. Zeus discovers and observes review commands itself; the live server-side `repo_review_config.verify_config_json` surface supplies test-coverage lens overrides. The onboarding behavior is driven by the bridge's canonical onboarding playbook, not the prompt, so `--onboarding` needs no prompt: `arcanist sessions create <repo> --onboarding --wait`. Any prompt supplied alongside `--onboarding` is ignored. Team use; not part of the external API surface.
 
-JSON mode returns `{sessionId, sessionUrl?, repoUrl, model?, agentRuntimeBackend?, reasoningEffort?, autoVerify?, baseBranch?, startBranch?, continuePrUrl?, continueMode?, onboarding?, promptId?}`. `sessionUrl` is emitted only when the server returns it; `agentRuntimeBackend` only when `--backend` is passed. With `--wait`, JSON mode also includes best-effort result fields when available: `prUrl?`, `publishedBranch?`, and `lastBranch?`.
+JSON mode returns `{sessionId, sessionUrl?, repoUrl, model?, reasoningEffort?, autoVerify?, baseBranch?, startBranch?, continuePrUrl?, continueMode?, onboarding?, promptId?}`. With `--wait`, JSON mode also includes best-effort result fields when available: `prUrl?`, `publishedBranch?`, and `lastBranch?`.
 
 ### `arcanist sessions qa <pr-url>`
 
@@ -235,7 +234,6 @@ Starts a QA verification session for an existing GitHub pull request.
 ```bash
 arcanist sessions qa https://github.com/your-org/your-repo/pull/123
 arcanist sessions qa https://github.com/your-org/your-repo/pull/123 --model gpt-5.4
-arcanist sessions qa https://github.com/your-org/your-repo/pull/123 --backend claude_code --model claude-opus-4-8
 arcanist sessions qa https://github.com/your-org/your-repo/pull/123 --reasoning-effort xhigh
 arcanist sessions qa https://github.com/your-org/your-repo/pull/123 --wait
 arcanist sessions qa https://github.com/your-org/your-repo/pull/123 --idempotency-key 1f0e6f1a-...
@@ -245,9 +243,8 @@ The PR URL must be `https://github.com/<owner>/<repo>/pull/<number>`.
 The CLI derives the repo from that URL, creates a QA session with `qa: true` and `targetPrUrl`, then enqueues the verifier prompt (skipped when the server signals `promptAlreadyEnqueued`).
 Inspect progress with the session URL or the session events stream.
 
-`--backend` picks the agent runtime backend: `codex` (default) or `claude_code`.
-`--model` must be valid for the chosen backend, and the CLI rejects mismatches before any network call.
-When `--model` is omitted, the backend default is used.
+`--model` pins the model for the QA session, and the CLI rejects an unknown model before any network call.
+When `--model` is omitted, the default is used.
 
 `--wait` blocks until the enqueued QA prompt finishes and exits non-zero if the prompt fails.
 JSON mode waits quietly and prints the create payload after the prompt completes successfully.
@@ -257,7 +254,7 @@ JSON mode waits quietly and prints the create payload after the prompt completes
 The CLI derives separate session and prompt idempotency keys from the provided value.
 If session creation succeeds but prompt enqueue fails, retry the same command with the same `--idempotency-key`.
 
-JSON mode returns `{sessionId, sessionUrl?, repoUrl, targetPrUrl, model?, agentRuntimeBackend?, reasoningEffort?, promptId?}`.
+JSON mode returns `{sessionId, sessionUrl?, repoUrl, targetPrUrl, model?, reasoningEffort?, promptId?}`.
 `promptId` is absent when the coordinator already enqueued the verifier prompt (no duplicate send). Active-verifier and per-PR run-limit conflicts both use exit code `4`; active-verifier errors include the existing session handle when the server returns it.
 
 ### `arcanist sessions send <session-id> [prompt]`
@@ -438,7 +435,7 @@ printf "summarize new regressions" | arcanist automations create your-org/your-r
 arcanist automations create https://github.com/your-org/your-repo - --cron "0 14 * * 1-5" --name "Weekday summary"
 ```
 
-Use `--model <model>` to pin the model for sessions started by the automation. Backend is derived from the model; non-codex backends such as Claude are limited to Arcanist team businesses. When `--model` is omitted the codex backend default is used.
+Use `--model <model>` to pin the model for sessions started by the automation. When `--model` is omitted the Codex default is used.
 
 Use `--slack-team-id <team-id>` together with `--slack-channel-id <channel-id>` to deliver each completed automation digest into a Slack channel. Both flags are required together; the target workspace must already be connected to Arcanist and the bot must already be in the channel.
 

package/dist/index.js

@@ -13136,7 +13136,7 @@ codex.command("use <state>").description("Turn using your saved Codex subscripti
 codex.command("status").description("Show Codex subscription availability and whether auth is saved").action((options, command) => codexStatusCommand(options, command));
 codex.command("logout").description("Deactivate the selector and remove the stored Codex subscription auth for your user").action((options, command) => codexLogoutCommand(options, command));
 var sessions = program.command("sessions").description("Session commands");
-program.command("review <pr-url>").description("Run a Zeus review for a GitHub pull request").option("--focus <text>", "Focus the review on a specific concern").option("--model <model>", "Pin the reviewer model (Arcanist members only); defaults to automatic routing").option("--wait", "Wait for the review and print its verdict and findings").option("--poll-interval <ms>", "Polling interval in milliseconds while waiting (default 10000)").addHelpText(
+program.command("review <pr-url>").description("Run a Zeus review for a GitHub pull request").option("--focus <text>", "Focus the review on a specific concern").option("--model <model>", "Pin the reviewer model (Arcanist members only); defaults to the Codex reviewer").option("--wait", "Wait for the review and print its verdict and findings").option("--poll-interval <ms>", "Polling interval in milliseconds while waiting (default 10000)").addHelpText(
   "after",
   `
 Examples:
@@ -13144,9 +13144,8 @@ Examples:
   arcanist review https://github.com/org/repo/pull/123 --focus "Check auth" --wait
   arcanist review https://github.com/org/repo/pull/123 --model gpt-5.5 --wait
 
-When the PR came from an Arcanist session, Zeus defaults to the runtime backend opposite
-the author's as a cross-check; otherwise it uses the default Codex reviewer. --model pins
-a specific reviewer model instead (members only).
+Zeus reviews with a pinned Codex reviewer model. --model pins a different reviewer model
+instead (members only).
 Zeus review is currently available to Arcanist members only.
 `
 ).action((prUrl, options, command) => reviewCommand(prUrl, options, command));

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@tryarcanist/cli",
-  "version": "0.1.254",
+  "version": "0.1.255",
   "description": "CLI for Arcanist — create and manage coding agent sessions",
   "type": "module",
   "bin": {