@tuned-tensor/cli 0.13.0-beta.1 → 0.13.1-beta.0

package/README.md

@@ -8,19 +8,57 @@
 
 ## Cloud sunset
 
-This release unregisters the hosted API commands (`tt auth`, `tt push`,
-`tt balance`, `tt topup`, `tt cloud`, and the rest of the managed tree). The
-modules remain in the repository so they can be restored later. `tt local …`
-is kept as a hidden alias; new scripts should call `tt run`, `tt doctor`,
-`tt runs list`, and so on directly.
+This release unregisters the hosted API commands (`tt auth`, `tt publish`,
+`tt push`, `tt balance`, `tt topup`, `tt cloud`, and the rest of the managed
+tree). The modules remain in the repository so they can be restored later.
+`tt local …` is kept as a hidden alias; new scripts should call `tt run`,
+`tt doctor`, `tt runs list`, and so on directly.
+
+## Migrating from 0.11
+
+`0.13` is the local-only CLI. If you are coming from `0.11.0`:
+
+- Call `tt run`, `tt doctor`, and `tt serve` at the root. `tt local …` still
+  works as a hidden alias.
+- Hosted account commands (`tt auth`, `tt publish`, `tt push`, and the rest)
+  are unregistered.
+- `Qwen/Qwen3.5-2B` is pinned to Hugging Face snapshot
+  `15852e8c16360a2fea060d615a32b45270f8a8fc`. Prefetch, train, eval, and serve
+  reject any other revision.
+- `tt serve active` fails unless an adapter is activated. Activation requires
+  a `generalRegression` suite in `local-runner.json`.
+- `tt info` reports `status: local`.
 
 ## Install
 
+On Linux or macOS:
+
 ```bash
-npm install -g @tuned-tensor/cli
+curl -fsSL https://tunedtensor.com/install.sh | sh
 tt --version
 ```
 
+The installer bootstraps Node.js 22.19+ if needed, installs
+`@tuned-tensor/cli` with npm `--ignore-scripts`, and falls back to
+`~/.local` when the global npm prefix is not writable. Inspect it first
+with `curl -fsSL https://tunedtensor.com/install.sh | less`. Pin a
+dist-tag or version on `sh`, not `curl`:
+
+```bash
+curl -fsSL https://tunedtensor.com/install.sh | TT_VERSION=beta sh
+curl -fsSL https://tunedtensor.com/install.sh | TT_VERSION=0.13.0 sh
+```
+
+Or install the package directly:
+
+```bash
+npm install -g --ignore-scripts @tuned-tensor/cli
+tt --version
+```
+
+Uninstall with `npm uninstall -g @tuned-tensor/cli` (add
+`--prefix ~/.local` if the curl installer used that prefix).
+
 Node.js 22.19.0 or newer is required. Local training additionally needs
 [`uv`](https://docs.astral.sh/uv/) and a supported NVIDIA CUDA host; the locked
 Python runner ships with the npm package and is prepared on first use.
@@ -41,33 +79,41 @@ The interactive agent harness and conversation state run on your laptop.
 Inference runs through the provider/model you select, which may be a
 local endpoint or a remote provider.
 
-`tt` reuses provider authentication and custom model definitions from
-`~/.pi/agent/auth.json` and `~/.pi/agent/models.json`. Authenticate a provider
-through its normal login flow; `tt` deliberately has no provider-secret flags.
-Then inspect and select a provider/model:
+Run `tt` to open one terminal for conversation and commands. Workflow commands
+such as `runs list` and `doctor` work immediately. Chat waits until you choose
+a provider and model from inside the shell:
 
-```bash
-tt agent models --all
-tt agent configure --provider anthropic --model claude-sonnet-4-5 --thinking high
-tt agent status
-```
+```text
+tt v0.13.1-beta.0
+agent not configured · workflow model base
+ctrl+c stop/clear · ctrl+d exit · /help commands · tab complete
 
-`--thinking` accepts `off`, `minimal`, `low`, `medium`, `high`, `xhigh`, or
-`max`.
-Selection metadata can be overridden per process with
-`TUNED_TENSOR_AGENT_PROVIDER`, `TUNED_TENSOR_AGENT_MODEL`, and
-`TUNED_TENSOR_AGENT_THINKING`. These values are not credentials.
+Use /model to choose a provider and model. Workflow commands work now.
+› /model
+› /login openai
+› /model openai
+› /model openai/gpt-5.6-sol
+```
 
-After configuration, run `tt` to open one terminal for conversation and
-commands:
+Use `/login` to save a provider API key. The shell asks which provider, then
+prompts for the key with hidden input and stores it under TT's local agent
+config. `/login <provider>` skips the provider prompt.
+`tt` does not accept provider secrets as flags and does not read
+`~/.pi/agent/`. `/model` and `/login` show OpenAI and OpenRouter.
+Other catalog providers remain available if you type their id.
+Unauthenticated entries are marked `auth required`. Local endpoints such as
+Ollama still need a placeholder `apiKey` in `models.json`.
 
 Before the shell opens, `tt` performs a short, non-blocking npm version check.
 If a newer stable release is available it recommends
 `npm install -g @tuned-tensor/cli@latest`; offline or unavailable registry
 checks are ignored.
 
+After a model is selected, the banner shows it and ordinary sentences go to
+the agent:
+
 ```text
-tt v0.13.0-beta.1
+tt v0.13.1-beta.0
 agent anthropic/claude-sonnet-4-5 · workflow model base
 ctrl+c stop/clear · ctrl+d exit · /help commands · tab complete
 
@@ -100,16 +146,24 @@ is bound to the current workspace so changing directories before approval fails
 safely instead of writing somewhere else. Ambiguous write failures remain
 sealed as `outcome_unknown` for manual inspection. `/reject` never mutates.
 
-Useful shell controls include `/help`, `/status`, `/context`, `/model`, `/cd`,
-`/clear`, and `/exit`.
+Useful shell controls include `/help`, `/status`, `/context`, `/model`,
+`/login`, `/cd`, `/clear`, and `/exit`.
 The shell banner and `/context` surface the assistant's configured provider and
 model (`agent provider/model`) separately from the workflow model, so the
 model answering your prompts is always visible. `/model` shows and changes the
-laptop-local TT agent model: run it with no arguments for a short list of
-suggestions, search with `/model <query>` (only the closest matches are
-shown), or switch with `/model <provider>/<model>` (for example
-`/model anthropic/claude-sonnet-4-5`). The equivalent non-interactive commands
-are `tt agent models`, `tt agent configure`, and `tt agent status`.
+laptop-local TT agent model: run it with no arguments to list providers and the
+two suggested models (GPT-5.6 Sol and DeepSeek V4 Flash), `/model <provider>`
+to list that provider's models, search
+with `/model <query>` (only the closest matches are shown), or switch with
+`/model <provider>/<model>` (for example `/model anthropic/claude-sonnet-4-5`).
+`/login` asks which provider to authenticate, then prompts for that provider's
+API key with hidden input. `/login <provider>` skips the provider prompt.
+The equivalent non-interactive commands are `tt agent models`,
+`tt agent configure`, and `tt agent status`.
+`tt agent configure --thinking` accepts `off`, `minimal`, `low`, `medium`,
+`high`, `xhigh`, or `max`. Selection metadata can be overridden per process with
+`TUNED_TENSOR_AGENT_PROVIDER`, `TUNED_TENSOR_AGENT_MODEL`, and
+`TUNED_TENSOR_AGENT_THINKING`. These values are not credentials.
 The shell keeps normal terminal scrollback and command history only for the
 current process.
 
@@ -173,7 +227,8 @@ tt models verify local-<run-id>
 tt models serve local-<run-id> --config local-runner.json
 ```
 
-Local currently certifies text SFT with `Qwen/Qwen3.5-2B`,
+Local currently certifies text SFT with `Qwen/Qwen3.5-2B` (pinned to snapshot
+`15852e8c16360a2fea060d615a32b45270f8a8fc`),
 `nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16`, and
 `meta-models/Muse-Glimmer-30B`, using LoRA/PEFT and CUDA. The 30B/3B-active
 Nemotron path is intended for DGX Spark-class unified memory, uses activation
@@ -184,7 +239,9 @@ vision tower frozen and unused. Evaluation and serving may use CPU. Training
 artifacts, datasets, and model weights remain on the execution host.
 
 Activation is optional and requires the run to pass a configured
-`generalRegression` gate. Once activated, use
+`generalRegression` gate. Without that suite, `tt models activate` fails
+closed. `tt serve active` also fails closed if nothing is activated — it does
+not silently serve the protected base model. Once activated, use
 `tt serve active --config local-runner.json`.
 
 `tt serve base` does not automatically inject the adjacent project spec.