Merge pull request 'feat(engine): plain chat generates at L3 — inside the safety cycle, not around it (+ crisis-path segfault fix)' (#109) from feat/soul-plain-chat-generation-20260805 into main

chat.el

@@ -416,6 +416,46 @@ fn engram_extract_ids(nodes_json: String) -> String {
 // A proper cache/circuit-breaker requires C runtime support (e.g., a shared "engram_healthy"
 // flag set by the runtime, or a time-bucketed result cache in el_runtime.c). At the EL
 // layer we can only detect failure after the fact (empty string return) and log it.
+// affective_node_ts — unix timestamp of an affective node (BellEvent / PositiveEvent).
+//
+// Prefers the " | ts:<epoch>" marker auto_persist writes into the node content; falls back
+// to created_at / updated_at. Returns 0 when there is no usable timestamp, which every
+// caller already treats as "too old to surface".
+//
+// ─── ELC CODEGEN NOTE — THIS MUST STAY A TOP-LEVEL FUNCTION ───────────────────────────
+// Do not inline this back into a block-expression initializer. Written inline as
+//     let start: Int = pos + str_len(marker)
+// inside a `let x: String = if cond { ... }` initializer, elc loses the declared Int type
+// and emits el_str_concat() for the `+`. el_str_concat takes the C string of each operand,
+// so two integers become a wild pointer and the daemon SEGFAULTS (EXC_BAD_ACCESS in
+// strlen). The identical expression in a plain function body compiles to integer addition
+// — verified in the generated C both ways. This is the same defect family Will hit on
+// 2026-06-23 and solved the same way (aff_try_slot, soul.el).
+//
+// Six inline copies of this parser existed before this function: two in layered_cycle's
+// L2c, two in engram_compile below, two in affective_context_prefix. All six emitted the
+// bad concat and all six now call here. Full write-up: BUG-PLAINCHAT-1 in
+// _engine-plainchat-20260805/README.md and the PR that introduced this function.
+// ──────────────────────────────────────────────────────────────────────────────────────
+fn affective_node_ts(node_json: String) -> Int {
+    if str_eq(node_json, "") { return 0 }
+    let content: String = json_get(node_json, "content")
+    let marker: String = " | ts:"
+    let mpos: Int = str_index_of(content, marker)
+    if mpos < 0 {
+        let ca: String = json_get(node_json, "created_at")
+        let alt: String = if str_eq(ca, "") { json_get(node_json, "updated_at") } else { ca }
+        if !engram_numeric_valid(alt) { return 0 }
+        return str_to_int(alt)
+    }
+    let start: Int = mpos + str_len(marker)
+    let rest: String = str_slice(content, start, str_len(content))
+    let nxt: Int = str_index_of(rest, " | ")
+    let raw: String = if nxt < 0 { rest } else { str_slice(rest, 0, nxt) }
+    if !engram_numeric_valid(raw) { return 0 }
+    return str_to_int(raw)
+}
+
 fn engram_compile(intent: String) -> String {
     // Issue 1: decompose multi-topic messages into sub-queries.
     let topics: String = engram_split_topics(intent)
@@ -519,20 +559,9 @@ fn engram_compile(intent: String) -> String {
     let cutoff_ts: Int = now_ts - 1209600
     let recent_bell: String = if bell_ok {
         let bn0: String = json_array_get(bell_nodes, 0)
-        let bn_content: String = json_get(bn0, "content")
-        let ts_marker: String = " | ts:"
-        let ts_pos: Int = str_index_of(bn_content, ts_marker)
-        let bn_ts_raw: String = if ts_pos >= 0 {
-            let ts_start: Int = ts_pos + str_len(ts_marker)
-            let rest: String = str_slice(bn_content, ts_start, str_len(bn_content))
-            let next_sep: Int = str_index_of(rest, " | ")
-            if next_sep < 0 { rest } else { str_slice(rest, 0, next_sep) }
-        } else {
-            let ca: String = json_get(bn0, "created_at")
-            if str_eq(ca, "") { json_get(bn0, "updated_at") } else { ca }
-        }
-        // Q1 fix: validate bell timestamp before str_to_int.
-        let bn_ts: Int = if !engram_numeric_valid(bn_ts_raw) { 0 } else { str_to_int(bn_ts_raw) }
+        // Q1 fix (validate before str_to_int) now lives inside affective_node_ts, which
+        // also replaces the inline " | ts:" parser that miscompiled to el_str_concat here.
+        let bn_ts: Int = affective_node_ts(bn0)
         if bn_ts > cutoff_ts { bn0 } else { "" }
     } else { "" }
     // Positive emotion context: check for recent joy/success moments within 72h.
@@ -540,19 +569,7 @@ fn engram_compile(intent: String) -> String {
     let pos_ec_ok: Bool = !str_eq(pos_ec_nodes, "") && !str_eq(pos_ec_nodes, "[]")
     let recent_positive_ec: String = if pos_ec_ok {
         let pec0: String = json_array_get(pos_ec_nodes, 0)
-        let pec_content: String = json_get(pec0, "content")
-        let pec_ts_marker: String = " | ts:"
-        let pec_ts_pos: Int = str_index_of(pec_content, pec_ts_marker)
-        let pec_ts_raw: String = if pec_ts_pos >= 0 {
-            let pec_ts_start: Int = pec_ts_pos + str_len(pec_ts_marker)
-            let pec_rest: String = str_slice(pec_content, pec_ts_start, str_len(pec_content))
-            let pec_next: Int = str_index_of(pec_rest, " | ")
-            if pec_next < 0 { pec_rest } else { str_slice(pec_rest, 0, pec_next) }
-        } else {
-            let pec_ca: String = json_get(pec0, "created_at")
-            if str_eq(pec_ca, "") { json_get(pec0, "updated_at") } else { pec_ca }
-        }
-        let pec_ts: Int = if str_eq(pec_ts_raw, "") { 0 } else { str_to_int(pec_ts_raw) }
+        let pec_ts: Int = affective_node_ts(pec0)
         if pec_ts > cutoff_ts { pec0 } else { "" }
     } else { "" }
     let affective_part: String = if !str_eq(recent_bell, "") {
@@ -768,7 +785,12 @@ fn build_system_prompt(ctx: String, chat_mode: Bool) -> String {
         safety_addendum
     }
 
-    return identity + operator_section + date_line + voice_rules + security_rules + capability_rules + bounded_persona_block + identity_block + affective_boot_block + engram_block + safety_block
+    // BUG FIX 2026-08-05: no_tools_rule was computed above and then never concatenated into
+    // this return, so the "[NO TOOLS THIS TURN]" instruction has not actually reached a model
+    // in this revision — the chat_mode flag had no effect on the prompt. Restored here, in the
+    // permanent-rules group, immediately after capability_rules (the rule it qualifies).
+    // Zero effect on agentic paths: they pass chat_mode=false, so no_tools_rule is "".
+    return identity + operator_section + date_line + voice_rules + security_rules + capability_rules + no_tools_rule + bounded_persona_block + identity_block + affective_boot_block + engram_block + safety_block
 }
 
 fn hist_append(hist: String, role: String, content: String) -> String {
@@ -929,6 +951,123 @@ fn conv_history_load() -> String {
     return content
 }
 
+// conv_history_record — append one completed turn to the conversation window.
+//
+// Same window, same append, same bell-guarded eviction handle_chat uses inline. It exists
+// as a function so the layered_cycle path records turns through exactly this code instead
+// of growing a second copy that can drift away from the bell guard.
+//
+// CONTRACT: assistant_msg MUST be post-safety_validate text. Recording the validated text
+// rather than the raw model output means the history window can never replay something the
+// output gate replaced or augmented. Callers on a hard bell must not call this at all —
+// bell turns are kept out of conversation history by design (see layered_cycle).
+fn conv_history_record(user_msg: String, assistant_msg: String) -> Void {
+    if str_eq(user_msg, "") { return "" }
+    let state_hist: String = state_get("conv_history")
+    let stored_hist: String = if str_eq(state_hist, "") { conv_history_load() } else { state_hist }
+    let h1: String = hist_append(stored_hist, "user", user_msg)
+    let h2: String = hist_append(h1, "assistant", assistant_msg)
+    // Bell-guarded trim: an evicted turn that triggered a bell is preserved to engram
+    // before it leaves the in-memory window.
+    let final_hist: String = if json_array_len(h2) > 20 {
+        hist_trim_with_bell_guard(h2)
+    } else {
+        h2
+    }
+    state_set("conv_history", final_hist)
+    conv_history_persist(final_hist)
+}
+
+// conv_history_block — recent dialogue, rendered for a system prompt.
+//
+// Same rendering handle_chat uses (role label + snipped content, one line per turn), read
+// from the same "conv_history" window, so a plain-chat turn can follow the thread instead
+// of answering every message from cold. Read-only: never writes history.
+fn conv_history_block() -> String {
+    let state_hist: String = state_get("conv_history")
+    let stored_hist: String = if str_eq(state_hist, "") { conv_history_load() } else { state_hist }
+    let hist_len: Int = if str_eq(stored_hist, "") { 0 } else { json_array_len(stored_hist) }
+    if hist_len == 0 {
+        return ""
+    }
+    let rh_out: String = ""
+    let rh_i: Int = 0
+    while rh_i < hist_len {
+        let rh_entry: String = json_array_get(stored_hist, rh_i)
+        let rh_role: String = json_get(rh_entry, "role")
+        let rh_content: String = json_get(rh_entry, "content")
+        let rh_label: String = if str_eq(rh_role, "user") { "User" } else { "Assistant" }
+        let rh_snip: String = if str_len(rh_content) > 400 { str_slice(rh_content, 0, 400) + "..." } else { rh_content }
+        let rh_line: String = rh_label + ": " + rh_snip
+        let rh_out = if str_eq(rh_out, "") { rh_line } else { rh_out + "\n" + rh_line }
+        let rh_i = rh_i + 1
+    }
+    return "\n\n[RECENT CONVERSATION — last " + int_to_str(hist_len) + " turns]\n" + rh_out
+}
+
+// layered_generate — the L3 generation step of layered_cycle. This is where the imprint
+// SPEAKS. (Added 2026-08-05.)
+//
+// layered_cycle calls this immediately after imprint_respond(), which has already applied
+// the active imprint's voice/domain annotation to the steward-aligned input. Before this
+// existed, L3 ended at that annotation and layered_cycle handed the user's own text back
+// as the "reply" — every gate ran, but nothing ever generated.
+//
+// It lives in chat.el, not imprint.el, on purpose: imprint.el is L3 and declares that the
+// lower layers are structurally inaccessible from it. It has zero imports, and making it
+// reach chat.el would transitively pull in safety.el (L1), inverting the layering and
+// breaking the tests that link imprint.el on its own. The layer ORDER is enforced by
+// layered_cycle, which is the only caller; this function holds no layer authority.
+//
+// SAFETY CONTRACT — this function is NOT a gate and must never become one:
+//   - Everything upstream has already run inside layered_cycle: L1 safety_screen, the
+//     safe-mode guard, the hard-bell short-circuit, L2a continuity/profiling, L2b mission
+//     alignment, L2c affective context. A hard bell can never reach this function —
+//     layered_cycle returns the fixed crisis message before L3 is entered.
+//   - Everything downstream is safety_validate(), which layered_cycle applies to this
+//     function's return value. Nothing here may bypass it, so this always returns plain
+//     text: no JSON envelope, no escaping, nothing for the output gate to have to unwrap.
+//   - The bell directive layered_cycle computed with safety_augment_system() is parked in
+//     the state key "layered_cycle_safety_system_addendum", and build_system_prompt() is
+//     its designated consumer. That is why the system prompt is assembled through
+//     build_system_prompt() here rather than hand-rolled: routing around it would drop the
+//     soft-bell / crisis directive on the floor. safety_augment_system() is NOT called
+//     again here — one evaluation per turn, one InternalStateEvent per bell.
+//   - No affective_context_prefix() call either: L2c already folded the affective note
+//     into that same addendum, and injecting it twice would double the cue.
+//
+// TOOLS: none, structurally. build_system_prompt(ctx, true) is chat mode, which injects
+// the permanent "NO TOOLS THIS TURN" rule, and llm_call_system() is a plain /v1/messages
+// call whose request body is built in el_runtime.c (llm_provider_request) with no "tools"
+// and no "tool_choice" key at all. Tools:Off means no tool is offered to the model, not
+// merely that none is used.
+//
+// Returns "" when the model call fails, so the caller reports the failure honestly instead
+// of echoing the user's own text back at them.
+fn layered_generate(prompt: String, imprint_id: String) -> String {
+    if str_eq(prompt, "") {
+        return ""
+    }
+
+    let ctx: String = engram_compile(prompt)
+    let model: String = chat_default_model()
+    let base_system: String = build_system_prompt(ctx, true) + current_engine_note(model)
+    let hist_block: String = conv_history_block()
+    let full_system: String = base_system + hist_block
+
+    let raw: String = llm_call_system(model, full_system, prompt)
+
+    let is_error: Bool = str_starts_with(raw, "{\"error\"")
+        || str_starts_with(raw, "{\"type\":\"error\"")
+        || str_contains(raw, "authentication_error")
+    if is_error {
+        println("[chat] layered_generate: model call failed — returning empty so the caller can report it honestly")
+        return ""
+    }
+
+    return clean_llm_response(raw)
+}
+
 // session_preload_bullets — render up to max_bullets nodes from a JSON array as
 // bullet lines, truncating content at snip_len chars each.
 fn session_preload_bullets(nodes: String, max_bullets: Int, snip_len: Int) -> String {
@@ -969,19 +1108,7 @@ fn affective_context_prefix() -> String {
     } else {
         if has_dist_aff {
             let dn0: String = json_array_get(dist_nodes_aff, 0)
-            let dn_content: String = json_get(dn0, "content")
-            let daff_marker: String = " | ts:"
-            let daff_pos: Int = str_index_of(dn_content, daff_marker)
-            let daff_ts_str: String = if daff_pos >= 0 {
-                let daff_start: Int = daff_pos + str_len(daff_marker)
-                let daff_rest: String = str_slice(dn_content, daff_start, str_len(dn_content))
-                let daff_next: Int = str_index_of(daff_rest, " | ")
-                if daff_next < 0 { daff_rest } else { str_slice(daff_rest, 0, daff_next) }
-            } else {
-                let daff_ca: String = json_get(dn0, "created_at")
-                if str_eq(daff_ca, "") { json_get(dn0, "updated_at") } else { daff_ca }
-            }
-            let daff_ts: Int = if str_eq(daff_ts_str, "") { 0 } else { str_to_int(daff_ts_str) }
+            let daff_ts: Int = affective_node_ts(dn0)
             daff_ts > aff_cutoff
         } else { false }
     }
@@ -989,19 +1116,7 @@ fn affective_context_prefix() -> String {
     let has_pos_aff: Bool = !str_eq(pos_nodes_aff, "") && !str_eq(pos_nodes_aff, "[]")
     let found_recent_pos: Bool = if has_pos_aff && !found_recent_dist {
         let pn0: String = json_array_get(pos_nodes_aff, 0)
-        let pn_content: String = json_get(pn0, "content")
-        let paff_marker: String = " | ts:"
-        let paff_pos: Int = str_index_of(pn_content, paff_marker)
-        let paff_ts_str: String = if paff_pos >= 0 {
-            let paff_start: Int = paff_pos + str_len(paff_marker)
-            let paff_rest: String = str_slice(pn_content, paff_start, str_len(pn_content))
-            let paff_next: Int = str_index_of(paff_rest, " | ")
-            if paff_next < 0 { paff_rest } else { str_slice(paff_rest, 0, paff_next) }
-        } else {
-            let paff_ca: String = json_get(pn0, "created_at")
-            if str_eq(paff_ca, "") { json_get(pn0, "updated_at") } else { paff_ca }
-        }
-        let paff_ts: Int = if str_eq(paff_ts_str, "") { 0 } else { str_to_int(paff_ts_str) }
+        let paff_ts: Int = affective_node_ts(pn0)
         paff_ts > aff_cutoff
     } else { false }
     let affective_out: String = if found_recent_dist {
@@ -1014,6 +1129,23 @@ fn affective_context_prefix() -> String {
     return affective_out
 }
 
+// ─────────────────────────────────────────────────────────────────────────────
+// handle_chat — UNWIRED. DO NOT ROUTE /api/chat HERE. (annotated 2026-08-05)
+//
+// This was the non-agentic chat handler until 2026-06-11 (f52d5bd, "wire consciousness
+// layers"), when Will moved /api/chat onto layered_cycle. It has had zero call sites since.
+//
+// It must stay unwired: it has NO enforcing input gate and NO enforcing output gate.
+// It never calls safety_screen, so a hard bell would reach the model instead of being
+// refused; it never calls safety_validate, the only enforcing output gate in the codebase;
+// and it runs no stewardship layer. Its one safety touch, safety_augment_system at the
+// llm_call_system site below, is an advisory system-prompt string — it cannot refuse,
+// replace, or block anything.
+//
+// Plain chat generates through layered_cycle's L3 (layered_generate) instead, which keeps
+// the screen, the stewardship layers and the output gate wrapped around the model call.
+// If this function is ever revived, it must be gated first, not wired first.
+// ─────────────────────────────────────────────────────────────────────────────
 fn handle_chat(body: String) -> String {
     let message: String = json_get(body, "message")
     if str_eq(message, "") {
@@ -1410,6 +1542,71 @@ fn agentic_tools_literal() -> String {
     "]"
 }
 
+// web_search_tool_json — the ONE place the server-side web_search tool block is built.
+//
+// Anthropic executes this tool on their side, inside the same /v1/messages call: no
+// third-party search key, no new account, and no anthropic-beta header (the plain
+// "anthropic-version: 2023-06-01" this soul already sends is sufficient).
+//
+// FUTURE-PROOF (design carried from soul-webfix-20260711.patch, Tim-approved): the tool
+// VERSION lives in exactly one place — state key "web_search_tool_version" — so a future
+// bump is a config write, not a recompile.
+//
+// DEFAULT IS THE BASIC VARIANT, AND THAT IS A DELIBERATE, MEASURED CHOICE.
+// The July patch defaulted to web_search_20260209 (dynamic filtering). That variant is
+// HARD-INCOMPATIBLE with the parallel-tool stopgap this engine currently depends on
+// (ADR 0005). Dynamic filtering is implemented with server-side programmatic tool
+// calling, and the API rejects the pair outright — verified live 2026-08-04, verbatim:
+//
+//   HTTP 400 invalid_request_error
+//   "tool_choice.disable_parallel_tool_use: true cannot be used with programmatic
+//    tool calling"
+//
+// So the two cannot both ship today. Dropping the stopgap would resurrect neuron#78
+// bug b, which killed agentic runs 3/3 in the ADR's own A/B — a functional regression.
+// Dropping web search entirely would leave the product's promise unmet. The basic
+// variant is compatible with the stopgap AND returns real, current results (proven
+// E2E), so it is what we default to. What we give up is dynamic filtering — a
+// token-efficiency and result-quality nicety, not the capability itself.
+//
+// REMOVAL TRIGGER: this default should move to web_search_20260209 the moment ADR
+// 0005's stopgap is retired (i.e. when the soul calls el_runtime's llm_call_agentic
+// and no longer needs disable_parallel_tool_use). At that point it is a one-line
+// config write — state_set("web_search_tool_version", "web_search_20260209") — with
+// no recompile. Flagged for Will's ratification; see the PR body.
+fn web_search_tool_json() -> String {
+    let ver: String = state_get("web_search_tool_version")
+    let eff: String = if str_eq(ver, "") { "web_search_20250305" } else { ver }
+    return "{\"type\":\"" + eff + "\",\"name\":\"web_search\",\"max_uses\":5}"
+}
+
+// strip_client_web_search — remove any CLIENT-side tool literally named "web_search"
+// from a tools-array interior, so attaching Anthropic's server-side tool cannot produce
+// two tools sharing one name (the API rejects that with "Tool names must be unique",
+// and before it did, the model picked between them nondeterministically).
+//
+// The soul's own literal set carries "web_get", not "web_search", so today this is a
+// no-op there — it exists for the MCP connector tools merged in by agentic_tools_all(),
+// which are third-party and may well ship a "web_search".
+//
+// LIMITATION (inherited from the July patch, stated rather than hidden): the scan keys
+// on the object terminator "]}}," so it only strips a matching tool that is followed by
+// another tool. A client web_search in the LAST array position is left in place.
+fn strip_client_web_search(tools_inner: String) -> String {
+    let ws_start: Int = str_index_of(tools_inner, "{\"name\":\"web_search\"")
+    if ws_start < 0 {
+        return tools_inner
+    }
+    let ws_rest: String = str_slice(tools_inner, ws_start, str_len(tools_inner))
+    let ws_end: Int = str_index_of(ws_rest, "]}},")
+    if ws_end <= 0 {
+        return tools_inner
+    }
+    let head: String = str_slice(tools_inner, 0, ws_start)
+    let tail: String = str_slice(ws_rest, ws_end + 4, str_len(ws_rest))
+    return head + tail
+}
+
 // agentic_tools_with_web — the standard tool set, always plus Anthropic's NATIVE
 // server-side web_search tool. Web search is BUILT IN: the model invokes it only when a
 // query needs fresh info (max_uses caps it), so there is no user-facing toggle. The native
@@ -1417,8 +1614,8 @@ fn agentic_tools_literal() -> String {
 // and needs no local runtime — it sidesteps the soul's lack of executable tools entirely.
 fn agentic_tools_with_web() -> String {
     let base: String = agentic_tools_literal()
-    let inner: String = str_slice(base, 1, str_len(base) - 1)
-    return "[" + inner + ",{\"type\":\"web_search_20250305\",\"name\":\"web_search\",\"max_uses\":5}]"
+    let inner: String = strip_client_web_search(str_slice(base, 1, str_len(base) - 1))
+    return "[" + inner + "," + web_search_tool_json() + "]"
 }
 
 // ---------------------------------------------------------------------------
@@ -1443,20 +1640,35 @@ fn connector_tools_json() -> String {
     return arr
 }
 
-// Built-in tools + every connector tool, as one tools array.
-// Uses agentic_tools_literal (not agentic_tools_with_web) to avoid a duplicate
-// "web_search" name — the literal already includes a custom web_search handler,
-// and adding the Anthropic server-side web_search_20250305 (same name) causes
-// Anthropic to reject with "Tool names must be unique."
+// agentic_tools_all — built-in tools + every connector tool + Anthropic's NATIVE
+// server-side web_search, as one tools array. This is the tool set EVERY agentic route
+// uses (handle_chat_agentic and handle_dharma_room_turn_agentic both call it; a
+// bridge-suspended run carries the same array through agentic_resume), so attaching the
+// web tool here is what actually turns web search on for the product.
+//
+// ACTIVATION — restoring an existing design, not inventing a mechanism:
+//   * Commit 8eea1d9 (2026-06-09, Tim-approved) made native web_search BUILT IN with no
+//     user-facing toggle — "the model invokes it only when a query needs fresh info
+//     (max_uses:5 caps it)" — and the desktop app removed its web-search toggle to pair
+//     with that. The call site was lost when agentic_tools_all() (connector tools, PR #19)
+//     replaced agentic_tools_with_web() in handle_chat_agentic.
+//   * tests/test_agentic_tools.el §2 still asserts agentic_tools_all() contains the
+//     native web_search tool. main currently FAILS that assertion; this restores it.
+// The old comment here claimed "the literal already includes a custom web_search handler"
+// — that is stale: agentic_tools_literal() ships web_get, and is_builtin_tool() has no
+// web_search. The duplicate-name hazard now lives only in third-party connector tools,
+// which is exactly what strip_client_web_search() handles.
 fn agentic_tools_all() -> String {
     let base: String = agentic_tools_literal()
     let conn: String = connector_tools_json()
+    let base_inner: String = str_slice(base, 1, str_len(base) - 1)
     let conn_inner: String = str_slice(conn, 1, str_len(conn) - 1)
-    if str_eq(conn_inner, "") {
-        return base
+    let merged: String = if str_eq(conn_inner, "") {
+        base_inner
+    } else {
+        base_inner + "," + conn_inner
     }
-    let base_open: String = str_slice(base, 0, str_len(base) - 1)
-    return base_open + "," + conn_inner + "]"
+    return "[" + strip_client_web_search(merged) + "," + web_search_tool_json() + "]"
 }
 
 // Proxy one tool call to the bridge. The model-supplied input is written to a
@@ -2223,6 +2435,20 @@ fn agentic_loop(session_id: String, model: String, safe_sys: String, tools_json:
     let iteration: Int = 0
     let keep_going: Bool = true
 
+    // Server-side web_search state, all three carried across iterations.
+    //
+    // tools_eff: a local copy of the caller's tools array, because the DRIFT branch below
+    //   may rewrite the web_search version in place after an API rejection. (Parameters are
+    //   not rebindable across loop iterations; a top-level local is.)
+    // container_id: web_search's dynamic filtering runs server-side code execution on
+    //   Anthropic's side, which hands back a container id that MUST be echoed on every
+    //   follow-up request in the same turn chain or the API 400s with "container_id is
+    //   required". Captured from each response, replayed on the next.
+    // ws_drift: set when we fell back, so we only ever fall back once per run.
+    let tools_eff: String = tools_json
+    let container_id: String = ""
+    let ws_drift: Bool = false
+
     // Suspension state — captured at top level so it escapes the while body.
     let pending: Bool = false
     let pend_tool_id: String = ""
@@ -2242,11 +2468,39 @@ fn agentic_loop(session_id: String, model: String, safe_sys: String, tools_json:
         state_set("run_progress_" + session_id, "")
     }
 
-    while keep_going && iteration < 8 {
+    // Cap raised 8 -> 12: a server-side web_search turn can pause and resume several
+    // times (see is_pause below), and each resume consumes an iteration. At 8 a
+    // multi-search answer could exhaust the loop before the model finished writing.
+    while keep_going && iteration < 12 {
+        // Carry the server-side code-execution container forward when we have one.
+        let cont_frag: String = if str_eq(container_id, "") {
+            ""
+        } else {
+            ",\"container\":\"" + container_id + "\""
+        }
+        // STOPGAP (2026-08-03, neuron#78 bug b / ADR 0005): the block walk below keeps
+        // only the FIRST tool_use block per round, so a PARALLEL tool_use response leaves
+        // the other ids without tool_result and the next turn 400s with
+        // "tool_use ids found without tool_result" — killing the run. Sending Anthropic's
+        // documented tool_choice.disable_parallel_tool_use makes the wire contract match
+        // what this loop can actually assemble. The real fix — a multi-tool_result loop
+        // that answers every block in a round — is Will's; see neuron#78.
+        //
+        // The stopgap and server-side web_search coexist: disable_parallel_tool_use
+        // constrains how many CLIENT tool_use blocks the model may emit per response, and
+        // Anthropic runs web_search itself (it comes back as server_tool_use +
+        // web_search_tool_result, never as a client tool_use the loop has to answer).
+        // Verified live, not assumed — see the README's probe transcript.
+        //
+        // max_tokens raised 4096 -> 16384: web_search injects whole result documents into
+        // the response, and a multi-search answer at 4096 truncated mid-sentence. This is
+        // the second half of the anti-truncation fix; pause_turn handling is the first.
         let req_body: String = "{\"model\":\"" + model + "\""
-            + ",\"max_tokens\":4096"
+            + cont_frag
+            + ",\"max_tokens\":16384"
+            + ",\"tool_choice\":{\"type\":\"auto\",\"disable_parallel_tool_use\":true}"
             + ",\"system\":\"" + safe_sys + "\""
-            + ",\"tools\":" + tools_json
+            + ",\"tools\":" + tools_eff
             + ",\"messages\":" + messages
             + "}"
 
@@ -2255,11 +2509,60 @@ fn agentic_loop(session_id: String, model: String, safe_sys: String, tools_json:
         let is_error: Bool = str_starts_with(raw_resp, "{\"error\"")
             || str_starts_with(raw_resp, "{\"type\":\"error\"")
             || str_contains(raw_resp, "authentication_error")
-        if is_error {
+
+        // DRIFT / FALLBACK: if the API rejected our configured web_search variant (the
+        // usual cause is a model too old for it — chat_default_model() is sonnet-4-5,
+        // which predates web_search_20260209), swap to the known-old basic variant,
+        // record the drift for the drift-watch, and retry this iteration.
+        //
+        // Everything here is computed as top-level if-expressions rather than inside an
+        // if-BLOCK, because in El a mutation inside an if-block does not escape its scope
+        // (see the block-walk note below). There is deliberately no `continue`: the
+        // fallback simply leaves `messages` untouched and keeps `keep_going` true, so the
+        // next iteration re-sends the same turn with the corrected tools array.
+        let ws_pos: Int = if is_error { str_index_of(tools_eff, "\"type\":\"web_search_") } else { 0 - 1 }
+        let ws_rest: String = if ws_pos >= 0 { str_slice(tools_eff, ws_pos + 8, str_len(tools_eff)) } else { "" }
+        // '"type":"' is 8 chars, so the version starts at ws_pos+8 and ends at the next quote.
+        let ws_vend: Int = if ws_pos >= 0 { str_index_of(ws_rest, "\"") } else { 0 - 1 }
+        let ws_cur: String = if ws_vend > 0 { str_slice(ws_rest, 0, ws_vend) } else { "" }
+        // Fall back on either of two signals, and ONLY these two — a loose prefix guess
+        // here once matched errors that had nothing to do with the tool:
+        //   1. the API names OUR configured version in its complaint (version drift, e.g.
+        //      a model too old for the configured variant); or
+        //   2. the API rejects the request for "programmatic tool calling" — the verified
+        //      signature of the dynamic-filtering variant colliding with ADR 0005's
+        //      disable_parallel_tool_use. An operator who sets web_search_tool_version to
+        //      web_search_20260209 while the stopgap still stands would otherwise get a
+        //      dead chat; this downgrades them automatically and says so in the log.
+        let ws_conflict: Bool = str_contains(raw_resp, "programmatic tool calling")
+        let can_fallback: Bool = is_error && !ws_drift && ws_pos >= 0 && ws_vend > 0
+            && !str_eq(ws_cur, "") && !str_eq(ws_cur, "web_search_20250305")
+            && (str_contains(raw_resp, ws_cur) || ws_conflict)
+        let tools_eff = if can_fallback {
+            str_slice(tools_eff, 0, ws_pos + 8) + "web_search_20250305" + str_slice(ws_rest, ws_vend, str_len(ws_rest))
+        } else { tools_eff }
+        let ws_drift = if can_fallback { true } else { ws_drift }
+        if can_fallback {
+            println("[soul] DRIFT: web_search variant '" + ws_cur + "' rejected by API - fell back to web_search_20250305")
+            state_set("web_search_version_drift", ws_cur)
+        }
+
+        if is_error && !can_fallback {
+            // Log the actual API error head — the old code swallowed it entirely, which
+            // made every upstream failure look identical from the outside.
+            let err_head: String = if str_len(raw_resp) > 220 { str_slice(raw_resp, 0, 220) } else { raw_resp }
+            println("[soul] llm error: " + err_head)
             return "{\"error\":\"llm unavailable\",\"reply\":\"\"}"
         }
 
         let stop_reason: String = json_get(raw_resp, "stop_reason")
+
+        // Capture/refresh the server-side container id when the response carries one.
+        let cont_raw: String = json_get_raw(raw_resp, "container")
+        let cont_id_new: String = if !str_eq(cont_raw, "") && !str_eq(cont_raw, "null") {
+            json_get(cont_raw, "id")
+        } else { "" }
+        let container_id = if !str_eq(cont_id_new, "") { cont_id_new } else { container_id }
         // json_get_raw needed — content is an array, json_get returns "" for non-strings
         let content_arr: String = json_get_raw(raw_resp, "content")
         let eff_content: String = if str_eq(content_arr, "") { "[]" } else { content_arr }
@@ -2271,13 +2574,39 @@ fn agentic_loop(session_id: String, model: String, safe_sys: String, tools_json:
         let tool_id: String = ""
         let tool_name: String = ""
         let tool_input: String = ""
+        // Server-executed tool names seen this round, quoted and comma-joined. Kept
+        // separate from tools_log so this inner walk has exactly one mutation site per
+        // variable (the El scope rule below), then merged in at the outer level.
+        let srv_log: String = ""
         let ci: Int = 0
         let c_total: Int = json_array_len(eff_content)
         while ci < c_total {
             let block: String = json_array_get(eff_content, ci)
-            let btype: String = json_get(block, "type")
+            // CITATION-BLOCK FIX (2026-08-04, found by E2E once web_search was live).
+            // json_get is a first-match scanner, and a cited text block serialises as
+            //   {"citations":[{"type":"web_search_result_location",...}],"type":"text",...}
+            // — citations FIRST. So json_get(block,"type") returns the nested citation's
+            // type, "text" never matches, and the block is silently dropped. Those are
+            // exactly the blocks carrying the searched facts, so the user got an answer
+            // with the data punched out of it ("The current temperature is , with .").
+            // Only text blocks carry a citations array, so its presence identifies the
+            // block type unambiguously without needing a real JSON parser.
+            let cit_raw: String = json_get_raw(block, "citations")
+            let has_cit: Bool = !str_eq(cit_raw, "") && !str_eq(cit_raw, "null")
+            let btype_scan: String = json_get(block, "type")
+            let btype: String = if has_cit { "text" } else { btype_scan }
             // Accumulate text at top level using if-expression
             let text_out = if str_eq(btype, "text") { text_out + json_get(block, "text") } else { text_out }
+            // FUTURE-PROOF: tools Anthropic runs on our behalf (web_search today, whatever
+            // ships tomorrow) arrive as server_tool_use blocks, never as client tool_use.
+            // Count the CATEGORY by the block's own name so a new server tool appears in
+            // tools_used with zero code changes — honest accounting for work we didn't run.
+            let is_srv: Bool = str_eq(btype, "server_tool_use")
+            let srv_name_raw: String = if is_srv { json_get(block, "name") } else { "" }
+            let srv_name: String = if is_srv && str_eq(srv_name_raw, "") { "server_tool" } else { srv_name_raw }
+            let srv_log = if is_srv {
+                if str_eq(srv_log, "") { "\"" + srv_name + "\"" } else { srv_log + ",\"" + srv_name + "\"" }
+            } else { srv_log }
             // Capture first tool_use block only
             let is_new_tool: Bool = str_eq(btype, "tool_use") && !has_tool
             let has_tool = if is_new_tool { true } else { has_tool }
@@ -2291,6 +2620,24 @@ fn agentic_loop(session_id: String, model: String, safe_sys: String, tools_json:
         // A real tool turn that targets a tool the soul cannot run in-process is a
         // CLIENT bridge: suspend the loop and hand the tool to the client.
         let is_tool_turn: Bool = str_eq(stop_reason, "tool_use") && has_tool
+
+        // pause_turn — REQUIRED for server-side web_search, not optional.
+        // Anthropic's server-side tool loop has its own iteration limit. When it hits it
+        // mid-answer the turn comes back with stop_reason "pause_turn" and only the text
+        // written SO FAR. Per Anthropic's contract the caller must re-send with the
+        // assistant content appended and NO tool_result, and the server resumes where it
+        // left off. Skip this and the user silently gets a truncated answer — no error,
+        // no warning, just a sentence that stops. (The bug that ate Tim's report,
+        // 2026-07-11.) Nothing else in this engine handles it: "pause_turn" appears
+        // nowhere else in the .el sources or in the shipped binary.
+        let is_pause: Bool = str_eq(stop_reason, "pause_turn")
+        // Unknown stop reasons (future API drift): finish honestly and log loudly rather
+        // than treating an unrecognised terminal state as a completed answer.
+        if !str_eq(stop_reason, "end_turn") && !str_eq(stop_reason, "tool_use") && !is_pause
+            && !str_eq(stop_reason, "max_tokens") && !str_eq(stop_reason, "refusal")
+            && !str_eq(stop_reason, "") {
+            println("[soul] DRIFT: unknown stop_reason from API: " + stop_reason)
+        }
         // If the user previously chose "always allow" for this tool in this session,
         // treat it like a builtin — run server-side via dispatch_tool and skip the
         // bridge suspension entirely so the approval UI is never shown again.
@@ -2318,10 +2665,19 @@ fn agentic_loop(session_id: String, model: String, safe_sys: String, tools_json:
         let tool_msg: String = "{\"type\":\"tool_result\",\"tool_use_id\":\"" + tool_id + "\",\"content\":\"" + tool_result + "\"}"
 
         // Accumulate tool names for the tools_used log surfaced in the response.
+        // Gated on is_tool_turn, not has_tool: a round truncated by max_tokens can carry a
+        // half-written tool_use block that will never run, and logging it would claim work
+        // the soul did not do.
         let tool_quoted: String = "\"" + tool_name + "\""
-        let tools_log = if has_tool {
+        let tools_log = if is_tool_turn {
             if str_eq(tools_log, "") { tool_quoted } else { tools_log + "," + tool_quoted }
         } else { tools_log }
+        // Merge in the server-executed tools (web_search) seen in this round's blocks.
+        let tools_log = if str_eq(srv_log, "") {
+            tools_log
+        } else {
+            if str_eq(tools_log, "") { srv_log } else { tools_log + "," + srv_log }
+        }
 
         // The assistant turn that requested the tool — needed verbatim on resume so the
         // tool_use/tool_result pairing stays valid when the client posts its result.
@@ -2332,15 +2688,22 @@ fn agentic_loop(session_id: String, model: String, safe_sys: String, tools_json:
 
         // Local built-in tool turn: append assistant + tool_result and keep looping.
         let local_continue: Bool = is_tool_turn && !needs_bridge
+        // Pause resume: the assistant content goes back VERBATIM with NO tool_result —
+        // that is the whole contract. Anthropic's server resumes its own tool loop from
+        // there; sending a tool_result (or a "continue" user turn) breaks it.
+        let is_pause_resume: Bool = is_pause && !needs_bridge
         let messages = if local_continue {
             let inner2: String = str_slice(messages_with_assistant, 1, str_len(messages_with_assistant) - 1)
             "[" + inner2 + ",{\"role\":\"user\",\"content\":[" + tool_msg + "]}]"
+        } else if is_pause_resume {
+            messages_with_assistant
         } else { messages }
 
         // Live progress ledger: one entry per round — the model's own narration
         // (its pre-tool prose, previously discarded here) plus the tool it reached
         // for. Clients poll /api/run-progress/<sid> to render these live.
-        if !str_eq(session_id, "") {
+        // Skipped on a version-fallback round: nothing happened that a poller should see.
+        if !str_eq(session_id, "") && !can_fallback {
             let prog_key: String = "run_progress_" + session_id
             let prog_prev: String = state_get(prog_key)
             let prog_snip: String = if str_len(text_out) > 280 { str_slice(text_out, 0, 280) } else { text_out }
@@ -2365,8 +2728,19 @@ fn agentic_loop(session_id: String, model: String, safe_sys: String, tools_json:
             bridge_save(session_id, model, safe_sys, tools_json, messages_with_assistant, tools_log, pend_tool_id)
         }
 
-        let final_text = if !is_tool_turn { text_out } else { final_text }
-        let keep_going = if local_continue { keep_going } else { false }
+        // ACCUMULATE across pause/resume cycles instead of overwriting. A resumed turn
+        // CONTINUES the answer, it does not repeat it — overwriting here would throw away
+        // everything the model wrote before the pause, which is the same truncation the
+        // pause handling exists to prevent. A version-fallback round contributes nothing.
+        let final_text = if !is_tool_turn && !can_fallback { final_text + text_out } else { final_text }
+        // Output cap hit mid-action: the tool block is truncated and will NOT run. Say so
+        // instead of ending on silent almost-work.
+        let final_text = if str_eq(stop_reason, "max_tokens") && has_tool {
+            final_text + "\n\n[Output limit reached mid-action - the last planned action did not run. Ask me to continue to finish it.]"
+        } else { final_text }
+        // Keep looping for a local tool round, a server-side pause resume, or a one-shot
+        // web_search version fallback.
+        let keep_going = if local_continue || is_pause_resume || can_fallback { keep_going } else { false }
         let iteration = iteration + 1
     }
 
@@ -2390,9 +2764,9 @@ fn agentic_loop(session_id: String, model: String, safe_sys: String, tools_json:
     // means the task was too complex for the agentic loop depth — surface it clearly
     // so the caller/operator knows to increase the cap or break the task apart.
     if str_eq(final_text, "") {
-        let hit_cap: Bool = iteration >= 8
+        let hit_cap: Bool = iteration >= 12
         let err_msg: String = if hit_cap {
-            "agentic loop hit the 8-iteration cap without producing a final reply - task may be too complex or a tool call is looping"
+            "agentic loop hit the 12-iteration cap without producing a final reply - task may be too complex or a tool call is looping"
         } else {
             "no response"
         }

dist/soul-with-nlg.el

@@ -1,3 +1,18 @@
+// ╔══════════════════════════════════════════════════════════════════════════╗
+// ║ STALE BUNDLE — DO NOT BUILD. UNSAFE CHAT PATH.                           ║
+// ╚══════════════════════════════════════════════════════════════════════════╝
+// This concatenated bundle is a snapshot, not a source of truth, and it is stale in
+// a way that matters for safety: it wires /api/chat straight to handle_chat and
+// contains NO layered_cycle at all (verified: zero occurrences in the bundled code —
+// the only textual hit in this file is this banner). A binary built from
+// this file would run chat with no enforcing input gate (no safety_screen, no
+// hard-bell short-circuit) and no enforcing output gate (no safety_validate).
+//
+// Build from the .el sources via manifest.el (entry soul.el), or from dist/soul.c.
+// Nothing in the repo references this file. It is kept only as a historical artifact
+// and should be deleted once Will confirms nothing external depends on it.
+// (Flagged 2026-08-04 in _engine-websearch-20260804/SAFETY-STOP.md; banner added
+// 2026-08-05 with the plain-chat generation fix.)
 // language-profile.el - Language profile data and accessors.
 //
 // A language profile is a slot map ([String] key-value list) describing the

dist/soul.c

@@ -28170,7 +28170,7 @@ el_val_t agentic_loop(el_val_t session_id, el_val_t model, el_val_t safe_sys, el
     state_set(el_str_concat(EL_STR("run_progress_"), session_id), EL_STR(""));
   }
   while (keep_going && (iteration < 8)) {
-    el_val_t req_body = el_str_concat(el_str_concat(el_str_concat(el_str_concat(el_str_concat(el_str_concat(el_str_concat(el_str_concat(el_str_concat(el_str_concat(el_str_concat(EL_STR("{\"model\":\""), model), EL_STR("\"")), EL_STR(",\"max_tokens\":4096")), EL_STR(",\"system\":\"")), safe_sys), EL_STR("\"")), EL_STR(",\"tools\":")), tools_json), EL_STR(",\"messages\":")), messages), EL_STR("}"));
+    el_val_t req_body = el_str_concat(el_str_concat(el_str_concat(el_str_concat(el_str_concat(el_str_concat(el_str_concat(el_str_concat(el_str_concat(el_str_concat(el_str_concat(EL_STR("{\"model\":\""), model), EL_STR("\"")), EL_STR(",\"max_tokens\":4096,\"tool_choice\":{\"type\":\"auto\",\"disable_parallel_tool_use\":true}")), EL_STR(",\"system\":\"")), safe_sys), EL_STR("\"")), EL_STR(",\"tools\":")), tools_json), EL_STR(",\"messages\":")), messages), EL_STR("}"));
     el_val_t raw_resp = http_post_with_headers(api_url, req_body, h);
     el_val_t is_error = ((str_starts_with(raw_resp, EL_STR("{\"error\"")) || str_starts_with(raw_resp, EL_STR("{\"type\":\"error\""))) || str_contains(raw_resp, EL_STR("authentication_error")));
     if (is_error) {

routes.el

@@ -15,6 +15,40 @@ fn flag_true(body: String, key: String) -> Bool {
     return json_get_bool(body, key) || json_get_int(body, key) > 0
 }
 
+// ---------------------------------------------------------------------------
+// plain_chat_envelope — the JSON response contract for a non-agentic ("Tools: Off")
+// chat turn. Every /api/chat dispatch that calls layered_cycle goes through here, so
+// the three call sites cannot drift apart.
+//
+// WHY THE ENVELOPE IS BUILT HERE AND NOT INSIDE layered_cycle:
+// layered_cycle returns the user-facing text AFTER safety_validate has acted on it.
+// Keeping the JSON out of the cycle means the output gate always sees raw model text
+// and never an escaped blob — there is nothing to unwrap and re-wrap on the crisis
+// path, which is exactly the failure mode that made wiring handle_chat unsafe.
+// Escaping is the last thing that happens, strictly after the gate.
+//
+// FIELDS: `reply` and `response` carry the same validated text. Both are required by
+// live clients — the desktop app reads `reply` first (DaemonClient.parseChatResponse),
+// while the CLI tools and the Telegram gateway read `response` (the gateway reads only
+// `response`). Emitting one would break the other.
+//
+// EMPTY MEANS FAILURE, NOT AN EMPTY ANSWER: a hard bell returns the fixed crisis
+// message and a soft bell is padded to non-empty by safety_validate, so the only way
+// an empty string leaves the cycle is a failed model call. It is reported as an error
+// rather than dressed up as a successful blank reply.
+// ---------------------------------------------------------------------------
+fn plain_chat_envelope(validated: String, model: String) -> String {
+    if str_eq(validated, "") {
+        return "{\"error\":\"llm unavailable\",\"reply\":\"\",\"response\":\"\",\"agentic\":false,\"tools_used\":[]}"
+    }
+    let safe: String = json_safe(validated)
+    return "{\"reply\":\"" + safe + "\""
+        + ",\"response\":\"" + safe + "\""
+        + ",\"model\":\"" + json_safe(model) + "\""
+        + ",\"agentic\":false"
+        + ",\"tools_used\":[]}"
+}
+
 // ---------------------------------------------------------------------------
 // Rate limiting — simple in-memory per-IP sliding window counter.
 //
@@ -243,8 +277,11 @@ fn handle_dharma_recv(body: String) -> String {
         } else if agentic_flag {
             handle_chat_agentic(chat_body)
         } else {
+            // Non-agentic ("Tools: Off"): the full L1→L2→L3→L1 cycle, which now generates
+            // at L3 instead of echoing. Envelope built outside the cycle — see
+            // plain_chat_envelope.
             let screened_reply: String = layered_cycle(raw_msg)
-            screened_reply
+            plain_chat_envelope(screened_reply, chat_default_model())
         }
         auto_persist(chat_body, reply)
         return reply
@@ -416,8 +453,9 @@ fn handle_request(method: String, path: String, body: String) -> String {
             } else if agentic_flag {
                 handle_chat_agentic(body)
             } else {
+                // Non-agentic ("Tools: Off") — same cycle and same envelope as POST.
                 let screened_reply: String = layered_cycle(eff_msg)
-                screened_reply
+                plain_chat_envelope(screened_reply, chat_default_model())
             }
             auto_persist(body, reply)
             return reply
@@ -580,8 +618,11 @@ fn handle_request(method: String, path: String, body: String) -> String {
             } else if agentic_flag {
                 handle_chat_agentic(body)
             } else {
+                // Non-agentic ("Tools: Off") — the app's DEFAULT mode (AgentMode.NEVER).
+                // Full L1→L2→L3→L1 cycle with real generation at L3; envelope built
+                // outside the cycle so safety_validate always sees raw text.
                 let screened_reply: String = layered_cycle(raw_msg)
-                screened_reply
+                plain_chat_envelope(screened_reply, chat_default_model())
             }
             auto_persist(body, reply)
             return reply

soul.el

@@ -453,40 +453,24 @@ fn layered_cycle(raw_input: String) -> String {
     let lc_aff_cutoff: Int = time_now() - 259200
     let lc_bell_nodes: String = engram_search_json("bell:soft bell:hard BellEvent affective", 2)
     let lc_has_bell: Bool = !str_eq(lc_bell_nodes, "") && !str_eq(lc_bell_nodes, "[]")
+    // CRASH FIX 2026-08-05 (BUG-PLAINCHAT-1): the " | ts:" parser used to be inline here.
+    // Inside this block-expression initializer elc compiled `lbmp + str_len(lbm)` to
+    // el_str_concat() on two integers, which segfaulted the whole daemon the moment a
+    // distress turn followed an earlier affective turn — i.e. exactly on the crisis path.
+    // Verified against the unmodified baseline binary AND present in the committed
+    // dist/soul.c. affective_node_ts() is a top-level function, where the same expression
+    // compiles to integer addition. Do not inline it back.
     let lc_bell_note: String = if lc_has_bell {
         let lb0: String = json_array_get(lc_bell_nodes, 0)
-        let lb_c: String = json_get(lb0, "content")
-        let lbm: String = " | ts:"
-        let lbmp: Int = str_index_of(lb_c, lbm)
-        let lb_ts_raw: String = if lbmp >= 0 {
-            let lbs: Int = lbmp + str_len(lbm)
-            let lbr: String = str_slice(lb_c, lbs, str_len(lb_c))
-            let lbn: Int = str_index_of(lbr, " | ")
-            if lbn < 0 { lbr } else { str_slice(lbr, 0, lbn) }
-        } else {
-            let lbca: String = json_get(lb0, "created_at")
-            if str_eq(lbca, "") { json_get(lb0, "updated_at") } else { lbca }
-        }
-        let lb_ts: Int = if str_eq(lb_ts_raw, "") { 0 } else { str_to_int(lb_ts_raw) }
+        let lb_ts: Int = affective_node_ts(lb0)
         if lb_ts > lc_aff_cutoff { "[AFFECTIVE NOTE: User was in distress in a recent session.]" } else { "" }
     } else { "" }
     let lc_pos_nodes: String = engram_search_json("PositiveEvent joy:high joy:low affective", 2)
     let lc_has_pos: Bool = !str_eq(lc_pos_nodes, "") && !str_eq(lc_pos_nodes, "[]")
+    // Same crash fix as the bell note above (BUG-PLAINCHAT-1).
     let lc_pos_note: String = if lc_has_pos && str_eq(lc_bell_note, "") {
         let lp0: String = json_array_get(lc_pos_nodes, 0)
-        let lp_c: String = json_get(lp0, "content")
-        let lpm: String = " | ts:"
-        let lpmp: Int = str_index_of(lp_c, lpm)
-        let lp_ts_raw: String = if lpmp >= 0 {
-            let lps: Int = lpmp + str_len(lpm)
-            let lpr: String = str_slice(lp_c, lps, str_len(lp_c))
-            let lpn: Int = str_index_of(lpr, " | ")
-            if lpn < 0 { lpr } else { str_slice(lpr, 0, lpn) }
-        } else {
-            let lpca: String = json_get(lp0, "created_at")
-            if str_eq(lpca, "") { json_get(lp0, "updated_at") } else { lpca }
-        }
-        let lp_ts: Int = if str_eq(lp_ts_raw, "") { 0 } else { str_to_int(lp_ts_raw) }
+        let lp_ts: Int = affective_node_ts(lp0)
         if lp_ts > lc_aff_cutoff { "[AFFECTIVE NOTE: User shared positive news in a recent session.]" } else { "" }
     } else { "" }
     let lc_affective_note: String = if !str_eq(lc_bell_note, "") { lc_bell_note } else { lc_pos_note }
@@ -498,11 +482,35 @@ fn layered_cycle(raw_input: String) -> String {
     }
     state_set("layered_cycle_safety_system_addendum", augmented_addendum)
 
-    // L3: imprint responds
-    let output: String = imprint_respond(aligned, imprint_id)
+    // L3: imprint responds — applies the active imprint's voice/domain annotation to the
+    // steward-aligned input. This produces the PROMPT, not the answer.
+    let prompt: String = imprint_respond(aligned, imprint_id)
 
-    // L1 out: validate output before delivery
-    return safety_validate(output, screen_action)
+    // L3b: the imprint SPEAKS (added 2026-08-05).
+    //
+    // Until now the cycle stopped at the annotation above, so /api/chat with agentic:false
+    // handed the user's own screened text back as the "reply" — every gate ran, but nothing
+    // ever generated. The generation is placed HERE, inside the cycle, rather than by
+    // pointing the route at handle_chat(): handle_chat has no enforcing input gate and no
+    // enforcing output gate, so calling it instead of this cycle would have traded the whole
+    // safety pipeline for a working reply. Composing keeps both.
+    //
+    // Order is deliberate and must not be rearranged: this call sits strictly AFTER the L1
+    // screen, the safe-mode guard, the hard-bell short-circuit and the L2 stewardship layers,
+    // and strictly BEFORE the L1 output gate. A hard bell never reaches a model — the branch
+    // above returns first. Tools are not offered on this turn; see layered_generate.
+    let output: String = layered_generate(prompt, imprint_id)
+
+    // L1 out: validate output before delivery. Still the terminal gate — nothing below this
+    // line can change the string this function returns.
+    let validated: String = safety_validate(output, screen_action)
+
+    // Turn bookkeeping. Records the VALIDATED text, never the raw model output, and is only
+    // reachable on the non-bell path: both bell branches above return before this point, so
+    // bell turns still never enter conversation history. Pure state side effect — it cannot
+    // alter what is returned.
+    conv_history_record(raw_input, validated)
+    return validated
 }
 
 let soul_cgi_id_raw: String = env("SOUL_CGI_ID")