@hyperdreamer/pi-webui 1.11.0-beta.5 → 1.11.0-beta.7

package/optional-skills/deterministic-subagent-driven-development/references/capability-contract.md

@@ -16,6 +16,12 @@ Frontier.
 A zero-parameter, read-only tool. It never mutates policy, never applies a tier,
 and never returns credentials or endpoints.
 
+It advertises no tier slash commands. An earlier draft of this file listed
+`/tier-economy`...`/tier-frontier` plus `/tier-up` and `/tier-down`. No such
+command is registered in the runtime: tier selection is the typed `tier` field on
+`spawn_subsession` and nothing else. The rendered `Model tier: <tier>` line is a
+human-readable consistency label, not a command.
+
 ```ts
 type ModelTier = "economy" | "fast" | "standard" | "advanced" | "capable" | "frontier";
 
@@ -35,13 +41,6 @@ interface GetModelPolicyV1 {
     blockedReason: string | null;
   };
   ladder: { valid: boolean; revision: string | null; blockedReason: string | null };
-  tierCommands: {
-    contractVersion: 1;
-    absolute: readonly ["/tier-economy", "/tier-fast", "/tier-standard", "/tier-advanced", "/tier-capable", "/tier-frontier"];
-    relative: readonly ["/tier-up", "/tier-down"];
-    leadingOnly: true;
-    exactOutcome: "ignored-exact";
-  };
   trackedDispatch: {
     contractVersion: 1;
     tierField: true;
@@ -59,10 +58,11 @@ thinkingLevel }` and carry model identity and supported thinking only.
 
 | Condition | Requirement |
 | --- | --- |
-| Exact mode | `currentTier` is `null`; an invalid ladder is permitted; `currentRuntime` and `nextRequestResolved` are both non-null and equal |
+| Exact mode | `currentTier` is `null`; `currentRuntime` and `nextRequestResolved` are both non-null and equal |
 | Valid tiered mode | `currentTier` is non-null; ladder is complete and valid; latest resolved tuple is non-null |
 | Invalid tiered mode | `nextRequestResolved` may be `null` only when `ladder.blockedReason` is a non-empty actionable reason; this state is capability-blocking |
 | Any policy blocked reason | Capability-blocking regardless of mode |
+| Invalid ladder, either mode | Capability-blocking. Children are dispatched *by tier* whatever the parent's mode, and reviewer/fixer tiers are derived by formula, so any unresolvable rung can fail a later round |
 
 A capability-blocking state yields `CAPABILITY_BLOCKED` before any worktree
 mutation, plan mutation, or dispatch.
@@ -102,14 +102,14 @@ without creating a child and without substituting a neighbouring tier. An omitte
 `tier` inherits the parent's model.
 
 Prompt text never selects a model. The runtime does not scan prompt bytes for
-slash directives, so a `/tier-*` line is a human-readable echo with zero control
-effect. A test fake that recovers a tier by splitting the prompt is exercising a
-channel the runtime does not implement, and cannot detect a child that ignored
-the directive.
+model-selection commands, so a `Model tier: <tier>` line is a human-readable echo
+with zero control effect. A test fake that recovers a tier from this label is
+exercising a channel the runtime does not implement, and cannot detect a child
+that ignored the typed field.
 
-The one exception is a guard, not a mechanism: a leading `/tier-*` line that
-*disagrees* with the typed `tier` is rejected before child creation, so a stale
-echoed directive cannot silently imply a tier that was not requested.
+The one exception is a guard, not a mechanism: a leading label that *disagrees*
+with the typed `tier` is rejected before child creation, so a stale rendered
+prompt cannot silently imply a tier that was not requested.
 
 ### No dispatch idempotency
 
@@ -129,7 +129,7 @@ rely on is therefore **detectable** non-idempotency, not prevented duplication:
 ### Fail-closed conditions
 
 Missing required fields, unknown tier values, a tier absent from the configured
-ladder, an unavailable model, and a leading directive disagreeing with the typed
+ladder, an unavailable model, and a leading tier label disagreeing with the typed
 tier all fail before a child is created.
 
 ## Recovery-input properties

package/optional-skills/deterministic-subagent-driven-development/references/plan-contract.md

@@ -100,10 +100,10 @@ re-reviewer formula takes no round. That is pinned in tests as a deliberate
 decision; changing it requires changing this contract.
 
 `tier` is the binding channel. The rendered prompt also opens with
-`/tier-<lowercase>` as a human-readable echo. The echo carries no control effect —
-the runtime never parses prompt text to select a model — but PI WEBUI does reject
-a leading directive that *disagrees* with the typed tier, so the echo works as a
-cross-check. An absent echo is not an error; a disagreeing one is.
+`Model tier: <lowercase>` as a human-readable echo. The echo carries no control
+effect — the runtime never parses prompt text to select a model — but PI WEBUI
+rejects a leading label that *disagrees* with the typed tier, so the echo works as
+a cross-check. An absent echo is not an error; a disagreeing one is.
 
 ## Identity pinning
 

package/optional-skills/deterministic-subagent-driven-development/references/state-machine.md

@@ -157,9 +157,9 @@ An intent stores the exact rendered prompt bytes, bounded at 384 KiB, before any
 session exists. Recovery reissues those bytes verbatim and never re-renders,
 because re-rendering couples recovery to renderer output.
 
-The typed `tier` selects the model. A leading `/tier-*` line is a human-readable
-echo with no control effect: absent is fine, and disagreeing with the typed tier
-is reported as renderer/formula divergence.
+The typed `tier` selects the model. A leading `Model tier: <tier>` line is a
+human-readable echo with no control effect: absent is fine, and disagreement with
+the typed tier is reported as renderer/formula divergence.
 
 ## Recovery authority
 

package/optional-skills/deterministic-subagent-driven-development/scripts/lib/plan-policy.mjs

@@ -100,15 +100,21 @@ export function finalReviewerTier() {
 }
 
 /**
- * The human-readable echo for a tier.
+ * The human-readable tier echo.
  *
  * This is display text, not a control channel. `spawn_subsession` selects a
- * model from its typed `tier` parameter; a `/tier-*` line in a rendered prompt
- * has no effect on which model runs. It exists so a human reading a transcript
- * can see the intended tier, and so renderer/formula divergence is detectable.
+ * model from its typed `tier` parameter; a `Model tier: <tier>` line in a
+ * rendered prompt has no effect on which model runs. It exists so a human
+ * reading a transcript can see the intended tier, and so renderer/formula
+ * divergence is detectable.
  */
+export function tierEcho(tier) {
+  return `Model tier: ${TIERS[tierIndex(tier)]}`;
+}
+
+/** @deprecated Use `tierEcho`; retained for older controller instructions. */
 export function tierDirective(tier) {
-  return `/tier-${TIERS[tierIndex(tier)]}`;
+  return tierEcho(tier);
 }
 
 const ROLES = Object.freeze({
@@ -120,10 +126,12 @@ const ROLES = Object.freeze({
 });
 
 /**
- * Resolve one role's tier and its directive.
+ * Resolve one role's tier and its display echo.
  *
- * The round argument is required for the fixer role and rejected for every other
- * role, so a caller cannot silently pass a round that has no effect.
+ * The round argument is required for the fixer role and rejected for every
+ * other role, so a caller cannot silently pass a round that has no effect.
+ * `directive` is a compatibility alias for controllers written before the
+ * display line stopped looking like a slash command.
  */
 export function roleTier({ implementer, role, round }) {
   const definition = ROLES[role];
@@ -139,7 +147,8 @@ export function roleTier({ implementer, role, round }) {
   const tier = definition.needsRound
     ? definition.resolve(implementer, round)
     : definition.resolve(implementer);
-  return { tier, directive: tierDirective(tier) };
+  const echo = tierEcho(tier);
+  return { tier, echo, directive: echo };
 }
 
 class PlanError extends Error {

package/optional-skills/deterministic-subagent-driven-development/scripts/lib/prompt-renderer.mjs

@@ -7,7 +7,7 @@
  * template evaluation.
  *
  * There is deliberately no general-purpose templating here. A template engine
- * would let a context value introduce a directive, a heading, or another
+ * would let a context value introduce a tier label, a heading, or another
  * placeholder, and the prompt is the one artifact a child treats as instructions.
  * Role contracts are static files copied verbatim; context is emitted as a
  * validated, escaped key/value list.
@@ -242,7 +242,7 @@ const validateContext = (role, context) => {
  *
  * Byte layout, fixed by contract because dispatch stores and reissues these bytes:
  *
- *   /tier-<tier>\n
+ *   Model tier: <tier>\n
  *   \n
  *   <role contract, trailing whitespace trimmed>\n
  *   \n
@@ -263,7 +263,7 @@ export function renderPrompt({ tier, role, context, skillRoot }) {
   const template = readFileSync(join(skillRoot, "prompts", definition.template), "utf8");
 
   const rendered = [
-    `/tier-${tier}`,
+    `Model tier: ${tier}`,
     "",
     template.trimEnd(),
     "",

package/optional-skills/deterministic-subagent-driven-development/scripts/lib/state-machine.mjs

@@ -494,13 +494,14 @@ const buildIntent = (state, event, intentPhase) => {
     fail(`renderedPrompt exceeds ${String(MAX_PROMPT_BYTES)} bytes (384 KiB)`);
   }
 
-  // A leading `/tier-*` line is a human-readable echo with no control effect.
-  // Its absence is fine; a disagreement with the typed tier means the renderer
-  // and the contract have diverged and must not be papered over.
-  const echo = /^\/tier-([a-z]+)[ \t]*(?:\r?\n|$)/u.exec(event.renderedPrompt);
+  // A leading `Model tier: <tier>` line is a human-readable echo with no
+  // control effect. Its absence is fine; a disagreement with the typed tier
+  // means the renderer and the contract have diverged and must not be papered
+  // over.
+  const echo = /^Model tier: ([a-z]+)[ \t]*(?:\r?\n|$)/u.exec(event.renderedPrompt);
   if (echo !== null && echo[1] !== event.tier) {
     fail(
-      `renderer/formula divergence: prompt echoes /tier-${echo[1]} while the typed tier is ${event.tier}`,
+      `renderer/formula divergence: prompt says Model tier: ${echo[1]} while the typed tier is ${event.tier}`,
     );
   }
 

package/optional-skills/deterministic-subagent-driven-development/scripts/sdd-state.mjs

@@ -75,6 +75,7 @@ export {
   reviewerTier,
   roleTier,
   tierDirective,
+  tierEcho,
   tierLabel,
   TIERS,
 } from "./lib/plan-policy.mjs";

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@hyperdreamer/pi-webui",
-  "version": "1.11.0-beta.5",
+  "version": "1.11.0-beta.7",
   "description": "Web UI for persistent Pi Coding Agent sessions in real workspaces.",
   "license": "MIT",
   "author": "Federico Jaramillo Martinez and HyperDreamer",
@@ -84,9 +84,9 @@
   },
   "devDependencies": {
     "@changesets/cli": "^2.31.0",
-    "@earendil-works/pi-agent-core": "^0.82.1",
-    "@earendil-works/pi-ai": "^0.82.1",
-    "@earendil-works/pi-coding-agent": "^0.82.1",
+    "@earendil-works/pi-agent-core": "^0.83.0",
+    "@earendil-works/pi-ai": "^0.83.0",
+    "@earendil-works/pi-coding-agent": "^0.83.0",
     "@eslint/js": "^10.0.1",
     "@types/node": "^24.13.3",
     "@types/ws": "^8.18.1",
@@ -116,9 +116,9 @@
   "homepage": "https://pi-webui.dev/",
   "packageManager": "npm@11.11.0",
   "peerDependencies": {
-    "@earendil-works/pi-agent-core": ">=0.82.1 <0.83",
-    "@earendil-works/pi-ai": ">=0.82.1 <0.83",
-    "@earendil-works/pi-coding-agent": ">=0.82.1 <0.83"
+    "@earendil-works/pi-agent-core": ">=0.83.0 <0.84",
+    "@earendil-works/pi-ai": ">=0.83.0 <0.84",
+    "@earendil-works/pi-coding-agent": ">=0.83.0 <0.84"
   },
   "keywords": [
     "pi-package",