Declare provider settings, and give the context figure a denominator

Two things a session could not say, and one it was saying wrongly.

**Every provider setting is reachable.** `-np 1`, the MTP draft depth, the
tool set, the sampling parameters -- most were hardcoded to what measured
best on this machine, which is right as a default and wrong as a constant:
the next machine has a different GPU and a different core count, and
nobody running this app can edit the source. `DriverKind::params` now
declares what a provider takes -- key, label, shape, what blank means, and
whether a change waits for a restart -- and the phone renders whatever
arrives, on the spawn form and in the session settings dialog. Adding a
setting to a driver is one entry in that table and no app change.
`POST /sessions/{id}/params` takes the whole map, so an absent key is the
instruction to unset; the sampling half applies at once and the session is
told in words which of the rest are waiting for a restart.

`tools` is one of them, because it is the biggest lever on a tight
context: the seven built-in definitions are ~1,300 tokens of every prompt
(2,191 against 887 with none). `"none"` omits the flag rather than passing
it on, since `--tools none` is `unknown tool "none"` and a server that
exits.

**The context figure has a denominator.** `Event::ContextWindow` carries
it, read from `llama-server`'s `/props` once the model is up -- the
measurement rather than the request, since a session that named no context
size gets the model's own. Neither coding CLI states its window, so those
keep the bare figure: "2,042" and "2,042 / 8,192" are deliberately
different-looking, and a missing ceiling is never drawn as a proportion of
an assumed one.

**And the numerator was wrong**, by the length of the last reply: it was
the prompt alone, so a five-word answer reported 2,042 against a slot
holding 2,355. It is the turn's total now, which matches `llama-server`'s
own `n_tokens` to within a token.

Two defects the review found, both of which would have shipped: changing
settings on a *stopped* session reported "no process running, so it can't
take new settings", when a stopped session is exactly when you would set
them for the next start; and `GET /tools` answers **403** rather than an
empty list on a server started without `--tools`, so reading it as a
failure made the no-tools session one that never started.

Verified against real models: settings spawned and changed live, the
restart note, a session with two tools and one with none, and the counter
checked against the server's own slot occupancy each time.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

server/src/session/llama/mod.rs

@@ -65,6 +65,32 @@ pub use tools::{DEFAULT_MODE as DEFAULT_PERMISSION_MODE, MODES as PERMISSION_MOD
 use mcp::McpServer;
 use tools::Tools;
 
+/// The sampling half of a session's settings, in the wire's own names.
+///
+/// One function because two callers need the identical mapping: the spawn, and
+/// a later change through [`Driver::set_params`]. A value that will not parse
+/// as a number is left out rather than passed through -- `llama-server` would
+/// refuse the whole request for it, which would look like the session breaking
+/// rather than like one field being wrong.
+fn sampling_from(
+    params: &std::collections::BTreeMap<String, String>,
+) -> serde_json::Map<String, Value> {
+    let mut sampling = serde_json::Map::new();
+    for (key, field) in [
+        ("temperature", "temperature"),
+        ("topP", "top_p"),
+        ("topK", "top_k"),
+        ("maxTokens", "max_tokens"),
+    ] {
+        if let Some(raw) = params.get(key)
+            && let Ok(number) = raw.parse::<f64>()
+        {
+            sampling.insert(field.to_string(), json!(number));
+        }
+    }
+    sampling
+}
+
 /// Exa's own hosted MCP server, which is what a llama session searches the web
 /// with. The address llama.cpp's web UI offers under "Exa" in its recommended
 /// servers, so a session here reaches the same thing that UI does.
@@ -76,6 +102,11 @@ pub const EXA_MCP_URL: &str = "https://mcp.exa.ai/mcp";
 /// is the escape for a machine where drafting turns out not to pay.
 const SPECULATIVE: &str = "speculative";
 
+/// The spawn parameter naming which built-in tools a session gets, as
+/// `llama-server`'s own comma-separated list. Absent is all of them, and
+/// `"none"` is the way to ask for a session that only talks.
+const TOOLS: &str = "tools";
+
 /// How many times one message may go round the call-a-tool loop.
 ///
 /// A bound rather than a budget: a small model that has decided to read the
@@ -221,8 +252,10 @@ struct Shared {
     /// Where this session's process record lives, so [`Driver::stop`] can find
     /// the server it has to end.
     session_dir: PathBuf,
-    /// Sampling settings chosen at spawn, sent with every request.
-    sampling: serde_json::Map<String, Value>,
+    /// Sampling settings, sent with every request. Behind a lock because they
+    /// are changeable while the session runs: they ride on the next request,
+    /// so unlike the server's own flags there is nothing to reload.
+    sampling: Mutex<serde_json::Map<String, Value>>,
     /// The session's working directory, which is where its tools act. `None`
     /// leaves that to `llama-server`, which is the honest answer rather than a
     /// guess at one.
@@ -289,26 +322,14 @@ impl LlamaDriver {
             "a llama.cpp session needs a model -- one of the downloaded ones, by its key",
         )?;
 
-        let mut sampling = serde_json::Map::new();
-        for (key, field) in [
-            ("temperature", "temperature"),
-            ("topP", "top_p"),
-            ("topK", "top_k"),
-            ("maxTokens", "max_tokens"),
-        ] {
-            if let Some(raw) = meta.params.get(key)
-                && let Ok(number) = raw.parse::<f64>()
-            {
-                sampling.insert(field.to_string(), json!(number));
-            }
-        }
+        let sampling = sampling_from(&meta.params);
 
         let driver = Self {
             shared: Arc::new(Shared {
                 sink,
                 transcript: transcript.to_path_buf(),
                 session_dir: session_dir.to_path_buf(),
-                sampling,
+                sampling: Mutex::new(sampling),
                 cwd: meta
                     .cwd
                     .as_ref()
@@ -403,6 +424,15 @@ impl LlamaDriver {
                         "{model} loaded and answering at {endpoint} with {} tools",
                         tools.offered().map_or(0, |offered| offered.len()),
                     );
+                    // Asked now rather than carried from the spawn flags: a
+                    // session that named no context size gets the model's
+                    // own, which only the loaded server knows, and one that
+                    // named an impossible one gets whatever it settled for.
+                    // Either way this is the measurement rather than the
+                    // request.
+                    if let Some(window) = context_window(&endpoint) {
+                        shared.emit(Event::ContextWindow { tokens: window });
+                    }
                     Serving::Ready {
                         endpoint: endpoint.clone(),
                         tools: Arc::new(tools),
@@ -531,14 +561,12 @@ fn spawn_server(
         "127.0.0.1".into(),
         "--port".into(),
         forward.there.to_string(),
-        // The built-in agent tools -- read, search, edit, shell. Every one of
-        // them, because a session offered a subset is a session that says "I
-        // can't do that" about something it was installed to do, and the
-        // question of whether a particular call should happen is the
-        // permission gate's rather than a flag's. They run on the machine
-        // serving the model, which is the machine the files are on.
-        "--tools".into(),
-        "all".into(),
+        // The built-in agent tools -- read, search, edit, shell. All of them
+        // unless the session says otherwise, because whether a particular call
+        // should happen is the permission gate's question rather than a flag's.
+        // They run on the machine serving the model, which is the machine the
+        // files are on.
+        //
         // One slot, not the four `llama-server` picks on its own. A session is
         // one conversation making one request at a time -- the driver holds a
         // second message until the turn ends -- so the other three are context
@@ -553,6 +581,27 @@ fn spawn_server(
         "-np".into(),
         "1".into(),
     ];
+    // Settable because it is not free: the definitions of all seven are around
+    // 2,000 tokens of every prompt -- measured at 2,191 against 1,322 for two
+    // of them -- which on a small context window is a quarter of it spent
+    // before anything is said.
+    //
+    // "none" omits the flag rather than passing it on: `--tools none` is
+    // `tools setup failed: unknown tool "none"` and a server that exits, since
+    // the argument is a list of tool names and no-tools is what having no flag
+    // means.
+    match meta.params.get(TOOLS).map(|chosen| chosen.trim()) {
+        Some("none") => {}
+        chosen => {
+            args.push("--tools".into());
+            args.push(
+                chosen
+                    .filter(|c| !c.is_empty())
+                    .unwrap_or("all")
+                    .to_string(),
+            );
+        }
+    }
     // A model that carries a multi-token-prediction head drafts with it, which
     // is most of a 50% speed-up for free -- the tensors are in the file
     // whether or not they are used, and without the flag `llama-server` says
@@ -573,6 +622,11 @@ fn spawn_server(
         ("contextSize", "-c"),
         ("gpuLayers", "-ngl"),
         ("threads", "-t"),
+        // How far ahead the draft head guesses. Not defaulted here: 2 measured
+        // 7% faster than llama.cpp's 3 on this machine's GPU, once, which is
+        // a reason to make the knob reachable and not a reason to move it for
+        // everybody.
+        ("specDraftNMax", "--spec-draft-n-max"),
     ] {
         if let Some(value) = meta.params.get(key) {
             args.push(flag.to_string());
@@ -854,7 +908,10 @@ fn converse(
         if shared.cancel.load(Ordering::SeqCst) {
             return Ok(());
         }
-        let reply = generate(endpoint, &messages, tools, &shared.sampling, shared)?;
+        // Read per call rather than per turn, so a sampling change made while
+        // a long turn is running reaches the rest of it.
+        let sampling = shared.sampling.lock().unwrap().clone();
+        let reply = generate(endpoint, &messages, tools, &sampling, shared)?;
         let calls = reply.calls;
         messages.push(Message {
             tool_calls: calls.iter().map(Call::wire).collect(),
@@ -1046,6 +1103,49 @@ impl Driver for LlamaDriver {
     // is called, and the rename has already happened where the name lives.
     fn set_title(&self, _title: &str) {}
 
+    /// Takes new settings: the sampling half now, and says so about the rest.
+    ///
+    /// The split is what [`crate::config::ParamSpec::restart`] describes, and
+    /// it is said out loud rather than left to the screen, because the screen
+    /// can only say what a setting *usually* does -- this is the one place
+    /// that knows whether this session's server was started with the old
+    /// value. A session already stopped needs no such note: its next start
+    /// will read all of them.
+    fn set_params(&self, params: &std::collections::BTreeMap<String, String>) {
+        *self.shared.sampling.lock().unwrap() = sampling_from(params);
+        // Only the settings that actually differ from what this session's
+        // server was started with. Listing every restart-only one on every
+        // save would be a wall of text about nothing having changed.
+        let waiting: Vec<&str> = crate::config::DriverKind::LlamaCpp
+            .params()
+            .iter()
+            .filter(|spec| {
+                spec.restart && params.get(spec.key) != self.respawn.meta.params.get(spec.key)
+            })
+            .map(|spec| spec.label)
+            .collect();
+        // Nothing to say to a session with no server: its next start reads all
+        // of them, which is what the note would have been asking for.
+        let running = matches!(&*self.shared.serving.lock().unwrap(), Serving::Ready { .. });
+        if !waiting.is_empty() && running {
+            let one = waiting.len() == 1;
+            self.shared.emit(Event::Error {
+                message: format!(
+                    "{} {} saved. {} when this session's server next starts -- stop and start \
+                     the session, or change its model, to load {} now.",
+                    waiting.join(", "),
+                    if one { "is" } else { "are" },
+                    if one {
+                        "It takes effect"
+                    } else {
+                        "They take effect"
+                    },
+                    if one { "it" } else { "them" },
+                ),
+            });
+        }
+    }
+
     fn set_permission_mode(&self, mode: &str) {
         if !tools::MODES.contains(&mode) {
             self.shared.emit(Event::Error {
@@ -1436,6 +1536,23 @@ fn log_tail(session_dir: &Path) -> String {
 /// How much of that log to carry into a message somebody reads on a phone.
 const LOG_TAIL_LINES: usize = 6;
 
+/// How many tokens this server can hold, from the server itself.
+///
+/// `/props` reports the per-slot context, which is the whole of it because
+/// this driver always starts one slot -- see the `-np` argument. `None` for a
+/// server that would not answer, which draws as no ceiling rather than as a
+/// guessed one.
+fn context_window(endpoint: &str) -> Option<u64> {
+    ureq::get(format!("{endpoint}/props"))
+        .call()
+        .ok()?
+        .body_mut()
+        .read_json::<Value>()
+        .ok()?
+        .pointer("/default_generation_settings/n_ctx")
+        .and_then(Value::as_u64)
+}
+
 /// One streamed completion: posts the conversation, emits each text delta as
 /// it arrives, and assembles whatever tool calls came with it.
 ///
@@ -1491,9 +1608,16 @@ fn generate(
     // at once interleaves their fragments, each tagged with its index.
     let mut calls: Vec<Call> = Vec::new();
     let mut tokens = 0u64;
-    // The prompt side only, which is what the model is holding -- the same
-    // definition the other dialects report, so one word on the phone means one
-    // thing whichever kind of session it is.
+    // What the model is holding when this call ends: the prompt it was given
+    // plus the reply it produced, which is exactly what the next call's prompt
+    // begins with.
+    //
+    // The prompt alone was wrong, and measurably: a five-word reply reported
+    // 2,042 against a slot holding 2,355 (2026-09-19, checked against
+    // `stop processing: n_tokens` in the server's own log). The gap is the
+    // reply, so it grows with how much the model just said -- which is the
+    // worst direction for a figure somebody is watching to see how much room
+    // is left.
     let mut context = None;
     for line in std::io::BufRead::lines(reader) {
         if shared.cancel.load(Ordering::SeqCst) {
@@ -1510,13 +1634,13 @@ fn generate(
         let Ok(chunk) = serde_json::from_str::<Value>(payload) else {
             continue;
         };
-        if let Some(usage) = chunk.get("usage") {
-            if let Some(total) = usage.get("total_tokens").and_then(Value::as_u64) {
-                tokens = total;
-            }
-            if let Some(prompt) = usage.get("prompt_tokens").and_then(Value::as_u64) {
-                context = Some(prompt);
-            }
+        // One figure answering both questions, which for this dialect it
+        // genuinely does: what the call was charged for and what the model is
+        // left holding are the same tokens, because nothing here is billed and
+        // the whole conversation is resent every time.
+        if let Some(total) = chunk.pointer("/usage/total_tokens").and_then(Value::as_u64) {
+            tokens = total;
+            context = Some(total);
         }
         let Some(delta) = chunk.pointer("/choices/0/delta") else {
             continue;