@getmagical/cli 0.8.0 → 0.9.1

package/README.md

@@ -7,6 +7,12 @@ Magical CLI for the shared MCP operation surface.
 - Shared operational commands come from `packages/mcp-registry`.
 - Do not add shared commands directly in this package or in `services/mcp/operations`.
 
+New agents created with `automations create` or added with `automations update`
+use the Recommended model pool when `config.model` is omitted. Explicit model
+selections and existing agents keep their current models. If no Recommended pool
+is configured for the agent type and environment, the save returns an actionable
+error rather than silently pinning a model.
+
 ## Usage
 
 ```bash
@@ -22,6 +28,49 @@ mgcl automations run-draft --draft-automation-id ... --agent-id ... --idempotenc
 mgcl automations publish --draft-automation-id ...   # non-production only: mgcl refuses against the production API, and the API rejects it too
 ```
 
+## Typed classification
+
+Use `mgcl classify` for focused typed judgments over context you already have.
+The request uses TypeSafe-shaped JSON with one shared `state` and a map of
+named `questions`. Version 1 supports `noul`, `choice`, and `score` questions.
+The backend uses `jev-latest` by default. Add `model` to the request to select
+an explicit supported model; the response reports the model actually used.
+Instructions, choice descriptions, rubric levels, and score legends may be
+structured JSON and are passed through without narrowing.
+The backend applies a 30-second total deadline, including bounded transient
+retries. Access is employee-only through normal CLI credentials or an allowed
+Magical runtime identity.
+
+```bash
+mgcl classify --input-file request.json
+mgcl classify --input-json '{"state":"Please refund the duplicate charge.","questions":{"refund_requested":{"type":"noul","instructions":"Does the message explicitly request a refund?"}}}'
+cat request.json | mgcl classify --input-file -
+```
+
+Pass exactly one input source. Questions in one batch are independent and
+cannot consume sibling answers. Split dependent judgments into separate calls.
+`noul` is a probability rather than a boolean; choice results retain the full
+option distribution; scores may be fractional. Confidence represents the
+returned distribution, not correctness, so include an `insufficient_evidence`
+choice when the supplied context may not establish an answer.
+
+The command evaluates only supplied text and JSON. It does not fetch URLs,
+files named in the request, run records, or media. Output is always the full
+parseable JSON response, including model, usage, probability distributions,
+score legends, and additive fields. A valid negative or uncertain answer exits
+successfully. Invalid input and backend or provider failures use the normal
+nonzero CLI error path and write diagnostics to stderr.
+
+## Publish checks after saving
+
+`automations create`, `automations update`, and `workbench patch apply` return
+`publishChecks` for the saved version and branch. Draft saves remain allowed.
+Inspect each check's `issues` (including agent/command targets and metadata),
+fix checks with `blocking: true`, and save again before completing the task.
+Nonblocking failures are advisory. `status: unknown` means the save succeeded
+but readiness could not be checked; retry before treating the task as complete.
+Passing health checks does not bypass publication permissions or review.
+
 ## Deployment targets
 
 Use the global `--env us|uk|stg` option to select a complete Magical API and
@@ -498,9 +547,9 @@ Automation settings follow the same shape (`automation:config.systemPrompt`).
 Branch creation defaults to main. Use `--source-branch-id` and optionally
 `--source-version-id` to fork from another branch or an older version in that
 branch's history. Names must use lowercase letters or numbers separated by
-single dashes, such as `release-2`, and must be 100 characters or fewer.
-Spaces, underscores, slashes, backslashes, punctuation, uppercase letters,
-leading or trailing dashes, repeated dashes, and over-length names are
+single dashes or slashes, such as `release-2` or `otto/branch-fix`, and must be
+100 characters or fewer. Spaces, underscores, backslashes, other punctuation,
+uppercase letters, leading or trailing separators, adjacent separators, and over-length names are
 rejected before an API request. Branch IDs, not names, identify targets, and
 every automation or branch ID must be a UUID; malformed IDs are rejected
 before an API request.