pi-ui-extend 1.0.38 → 1.0.39

package/docs/desktop-task-manager.md

@@ -0,0 +1,124 @@
+# Spec: Desktop Project Task Manager
+
+## Type
+
+Change
+
+## Goal
+
+Add a project-scoped task list to Pix Desktop in a persistent left sidebar and
+allow a saved task to start work immediately in a new session tab.
+
+## Scope
+
+- A collapsible, resizable left sidebar with `Tasks` and `Project` tabs.
+- A compact task list with create, edit, delete, filtering, and manual status
+  changes.
+- Task type (`bug`, `feature`, `improvement`), status (`backlog`, `todo`,
+  `in-progress`, `done`), and priority (`low`, `medium`, `high`, `urgent`).
+- Project-local persistence in `.pi/tasks.json` with a versioned schema.
+- Starting a task in a new ACP session, automatically sending a generated
+  prompt, marking the task in progress, and storing the new session id.
+- Opening an already-linked session instead of creating a duplicate.
+
+## Non-goals
+
+- Kanban, drag and drop, subtasks, dependencies, assignees, due dates,
+  comments, or task history.
+- Automatic transition to `done` when an agent turn finishes.
+- Synchronizing the project task file with the agent's session-local todo list.
+- Concurrent multi-window file merging.
+
+## Behavior
+
+1. The sidebar is visible below the title/session chrome, starts on `Tasks`,
+   remembers its width and collapsed state locally, and preserves the main
+   session workspace at narrow window sizes.
+2. Task rows show a title, type, status, priority, and actions. Status and
+   priority are communicated by a Lucide icon, text, and color so meaning never
+   depends on color alone. `done` is green, `in-progress` is yellow, and idle
+   states are neutral; all colors support light and dark themes.
+3. Creating or editing validates a non-empty title. A description is optional.
+4. Filters affect only the visible list and do not modify persisted ordering.
+5. A missing `.pi/tasks.json` is treated as an empty version-1 document.
+6. Starting an unlinked task creates and selects a new ACP session, persists
+   its id and `in-progress` status, appends the generated user prompt to the
+   transcript, and sends it immediately.
+7. Starting a linked task loads that session. A missing/stale linked session is
+   reported as a recoverable error and does not silently create another one.
+8. Task completion remains manual.
+
+## Contracts
+
+- Project file: `.pi/tasks.json` containing `{ version: 1, tasks: [...] }`.
+- Every task has a unique id, title, type, status, priority, `createdAt`, and
+  `updatedAt`; description and `sessionId` may be omitted.
+- Tauri exposes narrow read/write commands for this one file. Workspace paths
+  are validated and `.pi` symlinks may not escape the workspace.
+- The webview sends a complete validated document on each mutation. Writes use
+  a same-directory temporary file and replacement to avoid partial JSON.
+
+## Invariants
+
+- The task document version must be supported before it is displayed or saved.
+- Duplicate ids, unknown enum values, empty titles, and malformed timestamps
+  are rejected.
+- A task is linked to at most one session in this MVP.
+- Starting a task is disabled while another prompt/session operation is active.
+- A failed save does not leave the UI claiming a task state that was not
+  persisted.
+
+## Edge cases
+
+- Switching workspaces discards the previous workspace's in-memory task view
+  and loads the new project file.
+- Empty and missing files have distinct behavior: missing means no tasks;
+  malformed or empty JSON is an error.
+- Save, session creation, and prompt failures remain visible and retryable.
+- Collapsing the sidebar keeps its tab rail available; choosing a tab expands
+  it.
+
+## Related files
+
+- `desktop/src/App.svelte`
+- `desktop/src/components/`
+- `desktop/src/lib/`
+- `desktop/src/styles.css`
+- `desktop/src-tauri/src/lib.rs`
+
+## Verification
+
+- Unit tests for task parsing, validation, filtering, prompt generation, and
+  sidebar preference parsing.
+- Rust tests for missing/read/write/malformed files and workspace escape.
+- `npm --prefix desktop test`
+- `npm --prefix desktop run check`
+- `npm --prefix desktop run build:web`
+- `cargo test --manifest-path desktop/src-tauri/Cargo.toml`
+
+## Risks / unknowns
+
+- Whole-document writes assume one active desktop writer per project.
+- A linked session may later be removed outside Pix Desktop; the first version
+  reports this rather than automatically unlinking it.
+- Windows replacement semantics may require a fallback around replacing an
+  existing task file while retaining best-effort crash safety.
+
+## Evidence
+
+- Confirmed by code: desktop session creation/loading and prompt submission are
+  orchestrated in `desktop/src/App.svelte` through `AcpClient`.
+- Confirmed by code: Tauri already validates workspace-contained project file
+  reads in `desktop/src-tauri/src/lib.rs`.
+- Confirmed by tests: session tab ordering and ACP request behavior have focused
+  Vitest coverage under `desktop/src/lib/`.
+- Confirmed by docs: `DESIGN.md` specifies a compact persistent sidebar,
+  semantic theme roles, Lucide icons, and non-color status cues.
+- Implemented: `.pi/tasks.json` now has an explicit version-1 contract and
+  rejects unsupported versions rather than guessing a migration.
+- Verified by tests: all 82 desktop Vitest tests and all 8 Rust tests pass;
+  Svelte/TypeScript checks report no errors or warnings.
+- Verified visually: browser QA covered clean preference state, resize/collapse,
+  tabs, filters, and semantic colors in light/dark themes.
+- Verified natively: actual Tauri UI/backend CRUD, Run, persisted linkage,
+  Open session without duplication, and delete cleanup all passed.

package/external/pi-tools-suite/package.json

@@ -44,9 +44,9 @@
     "vscode-languageserver-protocol": "^3.17.5"
   },
   "peerDependencies": {
-    "@earendil-works/pi-ai": "0.84.4",
-    "@earendil-works/pi-coding-agent": "0.84.4",
-    "@earendil-works/pi-tui": "0.84.4",
+    "@earendil-works/pi-ai": "0.85.0",
+    "@earendil-works/pi-coding-agent": "0.85.0",
+    "@earendil-works/pi-tui": "0.85.0",
     "typebox": "*"
   },
   "devDependencies": {

package/external/pi-tools-suite/src/async-subagents/async-subagents.sample.jsonc

@@ -234,6 +234,14 @@
       "description": "Use when the sub-agent should make or plan code changes for a feature, bug fix, or refactor.",
       "model": "openai-codex/gpt-5.6-sol",
       "fallbackModels": ["zai/glm-5.3"],
+      // Avoid Sol recursively doing routine implementation work for a Sol parent.
+      // Luna also escalates substantial implementation to Terra. modelByParent
+      // wins over preset role models, so this prevents both Luna -> Sol and
+      // Sol -> Sol for implement tasks under the built-in `gpt` preset.
+      "modelByParent": {
+        "openai-codex/gpt-5.6-luna*": { "model": "openai-codex/gpt-5.6-terra", "fallbackModels": ["zai/glm-5.3"] },
+        "openai-codex/gpt-5.6-sol*": { "model": "openai-codex/gpt-5.6-terra", "fallbackModels": ["zai/glm-5.3"] }
+      },
       "thinking": "high"
     },
 

package/external/pi-tools-suite/src/async-subagents/core/agent-strategy.ts

@@ -1,6 +1,6 @@
 import { isGptLikeModel } from "./ultrawork-auto.js";
 
-export type AgentStrategyName = "parallel-first" | "deep-work";
+export type AgentStrategyName = "parallel-first" | "deep-work" | "escalation-aware" | "cost-aware-orchestrator";
 
 export interface AgentStrategyOptions {
 	modelRef?: string;
@@ -27,13 +27,40 @@ Default: autonomous deep worker. Build context directly, make progress, edit, an
 For broad work, keep delegation explicit and bounded: focused review/research/tests/frontend/deep tracks, plus one oracle only for high-stakes uncertainty or final plan checks. Read compact results, decide in the parent session, and report only what matters. If compressing unfinished work, preserve active objective + next step via todo/DCP rules.
 </agent_strategy>`;
 
+const ESCALATION_AWARE_STRATEGY_PROMPT = `<agent_strategy name="escalation-aware">
+Execution hint for Pi, not a replacement for system/developer/user instructions.
+
+Default: self-sufficient with tiered escalation. Solve narrow, well-bounded work directly, but do not grind through a large or uncertain task at the current model tier when a stronger focused subagent is appropriate. Use todo for the plan and async subagents for escalation; read compact results first and keep the parent context lean.
+
+For a Luna parent, prefer Terra workers for substantial multi-file research, tests, or implementation, and escalate deep root-cause analysis, architecture/security review, high-risk decisions, or repeatedly failing complex work to Sol through the deep/review roles. For a Terra parent, handle routine research/tests/implementation directly; escalate deep root-cause analysis, architecture/security review, high-risk decisions, or stubborn complex failures to Sol through deep/review. Do not escalate merely because a plan has several steps, and do not delegate a tiny known-file edit or exact lookup.
+
+Keep user questions, plan/todo changes, integration decisions, and the final report in the parent. Independent read-only escalations may run in parallel; serialize overlapping edits unless scopes are clearly disjoint. When work is delegated, synchronize its todo lifecycle: mark it in progress, collect and verify the worker result, then complete/update it before moving on.
+</agent_strategy>`;
+
+const COST_AWARE_ORCHESTRATOR_STRATEGY_PROMPT = `<agent_strategy name="cost-aware-orchestrator">
+Execution hint for Pi, not a replacement for system/developer/user instructions.
+
+Default: cost-aware orchestration. You are an expensive parent model, so keep the parent session focused on planning, decisions, integration, verification, and the final user-facing answer. For non-trivial todo work, prefer focused async subagents for repo scanning, multi-file research, documentation, tests, frontend work, and implementation steps that would otherwise require several repository tool calls. Read compact subagent results first; inspect raw artifacts or redo work in the parent only when verification or uncertainty requires it.
+
+Keep user questions, todo/plan changes, architecture tradeoffs, cross-worker integration decisions, high-stakes review, and the final report in the parent. Do not delegate merely to avoid one cheap exact lookup or a tiny known-file edit. Independent read-only tracks may run in parallel; serialize overlapping edits unless scopes are clearly disjoint. When a todo item is delegated, keep its lifecycle synchronized: mark it in progress, collect and verify the worker result, then complete/update it before moving on.
+</agent_strategy>`;
+
 export function agentStrategyPrompt(options: AgentStrategyOptions = {}): string | undefined {
 	const env = options.env ?? process.env;
 	const override = strategyOverride(env);
 	if (override === "off") return undefined;
 	if (options.customPrompt && shouldSkipCustomPrompt(env)) return undefined;
 
-	const strategy = override ?? (isGptLikeModel(options.modelRef) ? "deep-work" : "parallel-first");
+	const strategy = override
+		?? (isExpensiveGptParent(options.modelRef)
+			? "cost-aware-orchestrator"
+			: isEscalationAwareGptParent(options.modelRef)
+				? "escalation-aware"
+			: isGptLikeModel(options.modelRef)
+				? "deep-work"
+				: "parallel-first");
+	if (strategy === "cost-aware-orchestrator") return COST_AWARE_ORCHESTRATOR_STRATEGY_PROMPT;
+	if (strategy === "escalation-aware") return ESCALATION_AWARE_STRATEGY_PROMPT;
 	return strategy === "deep-work" ? DEEP_WORK_STRATEGY_PROMPT : PARALLEL_FIRST_STRATEGY_PROMPT;
 }
 
@@ -49,10 +76,22 @@ function strategyOverride(env: NodeJS.ProcessEnv): AgentStrategyName | "off" | u
 	if (FALSE_ENV_PATTERN.test(value)) return "off";
 	if (value === "parallel-first") return "parallel-first";
 	if (value === "deep-work") return "deep-work";
+	if (value === "escalation" || value === "escalation-aware") return "escalation-aware";
+	if (value === "cost-aware" || value === "cost-aware-orchestrator" || value === "orchestrator") return "cost-aware-orchestrator";
 	if (TRUE_ENV_PATTERN.test(value)) return undefined;
 	return undefined;
 }
 
+function isExpensiveGptParent(modelRef: string | undefined): boolean {
+	if (!modelRef) return false;
+	return /(?:^|\/)gpt-5\.6-sol(?:$|[-.:])/i.test(modelRef.trim());
+}
+
+function isEscalationAwareGptParent(modelRef: string | undefined): boolean {
+	if (!modelRef) return false;
+	return /(?:^|\/)gpt-5\.6-(?:luna|terra)(?:$|[-.:])/i.test(modelRef.trim());
+}
+
 function shouldSkipCustomPrompt(env: NodeJS.ProcessEnv): boolean {
 	const raw = firstEnv(env, "PI_AGENT_STRATEGY_WITH_CUSTOM_PROMPT", "ASYNC_SUBAGENTS_AGENT_STRATEGY_WITH_CUSTOM_PROMPT");
 	return raw ? !TRUE_ENV_PATTERN.test(raw.trim()) : true;

package/external/pi-tools-suite/src/async-subagents/core/config.ts

@@ -224,6 +224,10 @@ const BUILTIN_CONFIG: SubagentConfig = {
 		},
 		implement: {
 			description: "Use when the sub-agent should make or plan code changes for a feature, bug fix, or refactor.",
+			modelByParent: {
+				"openai-codex/gpt-5.6-luna*": { model: "openai-codex/gpt-5.6-terra", fallbackModels: ["zai/glm-5.3"] },
+				"openai-codex/gpt-5.6-sol*": { model: "openai-codex/gpt-5.6-terra", fallbackModels: ["zai/glm-5.3"] },
+			},
 			thinking: "high",
 		},
 		tests: {
@@ -234,10 +238,18 @@ const BUILTIN_CONFIG: SubagentConfig = {
 		},
 		review: {
 			description: "Use for review/audit of existing code or changes: correctness, security, performance, maintainability, API risks, quality. Do not implement new code.",
+			modelByParent: {
+				"openai-codex/gpt-5.6-luna*": { model: "openai-codex/gpt-5.6-sol", fallbackModels: ["zai/glm-5.3"] },
+				"openai-codex/gpt-5.6-terra*": { model: "openai-codex/gpt-5.6-sol", fallbackModels: ["zai/glm-5.3"] },
+			},
 			thinking: "high",
 		},
 		deep: {
 			description: "Use for broad hard reasoning: architecture, root-cause analysis, cross-module impact, complex debugging or tradeoffs.",
+			modelByParent: {
+				"openai-codex/gpt-5.6-luna*": { model: "openai-codex/gpt-5.6-sol", fallbackModels: ["zai/glm-5.3"] },
+				"openai-codex/gpt-5.6-terra*": { model: "openai-codex/gpt-5.6-sol", fallbackModels: ["zai/glm-5.3"] },
+			},
 			thinking: "high",
 		},
 		oracle: {

package/external/pi-tools-suite/src/async-subagents/index.ts

@@ -32,6 +32,7 @@ import { registerSubagentsTool } from "./tools/subagents.js";
 import type { LiveAgent, SubagentsLiveStateEvent } from "./types.js";
 import type { AgentState } from "./core/types.js";
 import { publishStartupSection } from "../startup-section.js";
+import { publishRpcSessionState } from "../lib/rpc-session-state.js";
 
 function isTerminalAgentStatus(status: AgentState["status"]): boolean {
 	return status === "done" || status === "failed" || status === "stopped";
@@ -83,8 +84,8 @@ function createLiveStatePayload(
 }
 
 function agentMatchesSession(agent: LiveAgent, sessionFile: string | undefined): boolean {
-	if (!sessionFile || !agent.parentSession) return true;
-	return pathsEqual(sessionFile, agent.parentSession);
+	if (!sessionFile) return true;
+	return agent.parentSession !== undefined && pathsEqual(sessionFile, agent.parentSession);
 }
 
 function isStaleExtensionContextError(error: unknown): boolean {
@@ -100,6 +101,7 @@ export default function (pi: ExtensionAPI) {
 	const subagentOverlay = new SubagentOverlay(liveAgents);
 	let sawAutoUltraworkCandidate = false;
 	let currentSessionFile: string | undefined;
+	let currentSessionStateContext: Parameters<typeof publishRpcSessionState>[0];
 	let completionWatchTimer: ReturnType<typeof setInterval> | undefined;
 	publishSubagentPresetsStartupSection();
 
@@ -109,6 +111,7 @@ export default function (pi: ExtensionAPI) {
 			const liveState = createLiveStatePayload(liveAgents, currentSessionFile);
 			pi.events?.emit?.(SUBAGENTS_LIVE_COUNT_EVENT, { count: liveState.count });
 			pi.events?.emit?.(SUBAGENTS_LIVE_STATE_EVENT, liveState);
+			publishRpcSessionState(currentSessionStateContext, SUBAGENTS_LIVE_STATE_EVENT, liveState);
 			updateCompletionWatcher();
 		} catch (error) {
 			ignoreStaleExtensionContextError(error);
@@ -168,6 +171,7 @@ export default function (pi: ExtensionAPI) {
 		try {
 			sawAutoUltraworkCandidate = false;
 			currentSessionFile = sessionFileFromContext(ctx);
+			currentSessionStateContext = ctx;
 			subagentOverlay.restoreRunningAgents(ctx.cwd, currentSessionFile);
 			refreshSubagentOverlay();
 		} catch (error) {

package/external/pi-tools-suite/src/async-subagents/subagent-overlay.ts

@@ -23,7 +23,7 @@ export class SubagentOverlay {
 				if (liveRun.has(agent.id)) continue;
 				const agentDir = path.join(runDir, agent.id);
 				const agentParentSession = readParentSessionLink(agentDir);
-				if (parentSession && agentParentSession && !pathsEqual(parentSession, agentParentSession)) continue;
+				if (parentSession && (!agentParentSession || !pathsEqual(parentSession, agentParentSession))) continue;
 				liveRun.set(agent.id, { runDir, agentId: agent.id, parentSession: agentParentSession, completed: Promise.resolve() });
 			}
 			if (liveRun.size === 0) this.liveAgents.delete(runDir);

package/external/pi-tools-suite/src/dcp/auto-compress.ts

@@ -116,7 +116,7 @@ export function buildProgrammaticSummary(
 	return lines.join("\n")
 }
 
-const SUMMARIZER_SYSTEM_PROMPT = `You summarize a slice of a coding agent's conversation so it can replace the raw messages in context. Produce a dense, continuation-focused summary: preserve user intent, decisions made, files/symbols changed or inspected, exact errors still actionable, verification status, and next steps. Do not infer, invent, or add facts absent from the source; preserve uncertainty instead of filling gaps. Drop full logs, repeated output, and incidental detail. Be concise (roughly 4-10 bullets). Output ONLY the summary text, no preamble.`
+const SUMMARIZER_SYSTEM_PROMPT = `You summarize a slice of a coding agent's conversation so it can replace the raw messages in context. Produce a dense, continuation-focused summary: preserve user intent, decisions made, files/symbols changed or inspected, exact errors still actionable, verification status, and next steps. Preserve exact identifiers and explicit continuity markers verbatim, including uppercase labels before colons; never paraphrase or omit those labels. Do not infer, invent, or add facts absent from the source; preserve uncertainty instead of filling gaps. Drop full logs, repeated output, and incidental detail without quoting or naming the discarded log lines or their markers. Be concise (roughly 4-10 bullets). Output ONLY the summary text, no preamble.`
 
 /** Outcome of one summarizer-model attempt, surfaced in DCP debug logs. */
 export interface ModelSummaryAttempt {

package/external/pi-tools-suite/src/dcp/compression-blocks.ts

@@ -388,55 +388,6 @@ export function resolveIdToBoundary(
   const ts = state.messageIdSnapshot.get(id)
   if (ts !== undefined) return { timestamp: ts }
 
-  // ── Stale mNNN fallback: direction-only clamp ────────────────────────
-  // When compression or pruning removes messages between context passes,
-  // positional mNNN IDs shift (e.g. an end boundary m145 becomes m123
-  // after 22 messages are removed).  Clamp to the closest valid ID, but
-  // ONLY in a direction that preserves the range's semantics:
-  //   - start boundary: clamp upward to the first available ID at or after
-  //     the requested number.  The start must never move backwards into
-  //     older content; if no such ID exists, the requested start is gone
-  //     and we cannot safely compress — throw.
-  //   - end boundary: clamp downward to the last available ID at or before
-  //     the requested number.  The end must never move forwards into newer
-  //     content; if no such ID exists, throw.
-  // The previous implementation fell back to the highest available ID in
-  // both "no match" cases, which could clamp e.g. m010..m010 over a
-  // snapshot of m001..m003 to a single-message block over m003 — silently
-  // compressing the wrong content.
-  const mMatch = id.match(/^m(\d+)$/i)
-  if (mMatch && state.messageIdSnapshot.size > 0) {
-    const requestedNum = parseInt(mMatch[1]!, 10)
-    const allIds = sortIds([...state.messageIdSnapshot.keys()])
-    const allNums = allIds
-      .map((mid) => {
-        const n = mid.match(/^m(\d+)$/i)
-        return n ? { id: mid, num: parseInt(n[1]!, 10) } : null
-      })
-      .filter((entry): entry is { id: string; num: number } => entry !== null)
-      .sort((a, b) => a.num - b.num)
-
-    if (allNums.length > 0) {
-      let clamped: { id: string; num: number } | undefined
-      if (field === "startTimestamp") {
-        clamped = allNums.find((entry) => entry.num >= requestedNum)
-      } else {
-        for (let i = allNums.length - 1; i >= 0; i--) {
-          if (allNums[i]!.num <= requestedNum) {
-            clamped = allNums[i]
-            break
-          }
-        }
-      }
-      if (clamped) {
-        const clampedMeta = state.messageMetaSnapshot.get(clamped.id)
-        if (clampedMeta) return resolveMetaBoundary(clampedMeta, field, state)
-        const clampedTs = state.messageIdSnapshot.get(clamped.id)
-        if (clampedTs !== undefined) return { timestamp: clampedTs }
-      }
-    }
-  }
-
   throw unknownCompressionIdError(id, state)
 }
 
@@ -445,8 +396,7 @@ export function resolveIdToBoundary(
  * synthetic placeholder for an active compression block (`meta.blockId` is
  * set), resolve to that block's stored boundary so the caller rolls the
  * block up instead of nesting a new block on top of the placeholder. This
- * matters for both exact-match and clamped mNNN IDs: a model-visible mNNN
- * may itself represent a previously compressed section.
+ * A model-visible mNNN may itself represent a previously compressed section.
  */
 function resolveMetaBoundary(
   meta: MessageIdMeta,

package/external/pi-tools-suite/src/dcp/debug-log.ts

@@ -127,7 +127,13 @@ export function summarizeDcpState(state: DcpState): Record<string, unknown> {
       nextBlockId: state.nextBlockId,
     },
     inactiveBlocksTail: inactiveBlocks,
+    persistentMessageIds: state.messageIdsByStableId.size,
+    nextMessageId: state.nextMessageId,
     prunedTools: state.prunedToolIds.size,
+    automaticPruneCheckpoint: {
+      turn: state.lastAutomaticPruneTurn,
+      blockId: state.lastAutomaticPruneBlockId,
+    },
     providerSeenTools: state.providerSeenToolIds.size,
     consecutiveEmergencyPasses: state.consecutiveIgnoredStrongNudges,
     nudgeAnchors: state.nudgeAnchors.map((anchor) => ({

package/external/pi-tools-suite/src/dcp/index.ts

@@ -45,13 +45,7 @@ import {
 	resolveContextThresholds,
 	estimateTokens,
 } from "./pruner.js"
-import {
-	stripStaleDcpMetadataFromAssistantMessage,
-	stripStaleDcpMetadataFromMessage,
-} from "./pruner-metadata.js"
-import {
-	buildMessageIdControlText,
-} from "./pruner-message-ids.js"
+import { stripStaleDcpMetadataFromMessage } from "./pruner-metadata.js"
 import { summarizeDcpState, writeDcpDebugLog } from "./debug-log.js"
 import type { DcpNudgeType } from "./pruner-types.js"
 import { registerCompressTool } from "./compress-tool.js"
@@ -119,104 +113,6 @@ function isDcpControlPlaneMessage(message: any): boolean {
 	return message?.role === "custom" && DCP_CONTROL_PLANE_CUSTOM_TYPES.has(message.customType)
 }
 
-const DCP_PROVIDER_CONTROL_HEADER = "DCP message ID control data (do not quote or output):"
-
-function appendTextToContent(content: unknown, text: string): unknown {
-	if (typeof content === "string") return `${content}\n\n${text}`
-	if (Array.isArray(content)) {
-		const textType = content.some((part: any) => part?.type === "input_text") ? "input_text" : "text"
-		return [...content, { type: textType, text }]
-	}
-	return text
-}
-
-function appendDcpControlToMessages(messages: unknown, text: string): unknown {
-	if (!Array.isArray(messages)) return messages
-	const block = `${DCP_PROVIDER_CONTROL_HEADER}\n${text}`
-
-	// Keep DCP's volatile ID map at the tail of the provider context instead of
-	// mutating the stable system/developer prefix. Prefix cache reuse is much
-	// better when only the newest transcript item changes on each request.
-	let targetIndex = -1
-	for (let index = messages.length - 1; index >= 0; index--) {
-		const message = messages[index] as any
-		if (!message || typeof message !== "object") continue
-		// Responses items such as reasoning/function_call_output do not accept a
-		// `content` field. Keep a function output at the tail by appending to its
-		// valid `output` string; otherwise scan back to an actual message item.
-		if (typeof message.type === "string" && message.type !== "message" && message.role === undefined) {
-			if (message.type === "function_call_output" && typeof message.output === "string") {
-				return messages.map((candidate: any, candidateIndex) => candidateIndex === index
-					? { ...candidate, output: `${candidate.output}\n\n${block}` }
-					: candidate)
-			}
-			continue
-		}
-		if (message.role === "system" || message.role === "developer") continue
-		targetIndex = index
-		break
-	}
-
-	if (targetIndex < 0) {
-		targetIndex = messages.findIndex((message: any) =>
-			message?.role === "system" || message?.role === "developer"
-		)
-	}
-
-	if (targetIndex >= 0) {
-		return messages.map((message: any, index) => index === targetIndex
-			? { ...message, content: appendTextToContent(message.content, block) }
-			: message)
-	}
-	return [{ role: "system", content: block }, ...messages]
-}
-
-function appendDcpControlToAnthropicSystem(system: unknown, text: string): unknown {
-	const block = `${DCP_PROVIDER_CONTROL_HEADER}\n${text}`
-	if (typeof system === "string") return `${system}\n\n${block}`
-	if (Array.isArray(system)) return [...system, { type: "text", text: block }]
-	if (system === undefined || system === null) return [{ type: "text", text: block }]
-	return system
-}
-
-function appendDcpControlToGoogleSystemInstruction(systemInstruction: unknown, text: string): unknown {
-	const block = `${DCP_PROVIDER_CONTROL_HEADER}\n${text}`
-	if (typeof systemInstruction === "string") return `${systemInstruction}\n\n${block}`
-	if (systemInstruction === undefined || systemInstruction === null) return block
-	return systemInstruction
-}
-
-function appendDcpControlToProviderPayload(payload: unknown, text: string): unknown {
-	if (Array.isArray(payload)) return appendDcpControlToMessages(payload, text)
-	if (!payload || typeof payload !== "object") return payload
-	const record = payload as Record<string, unknown>
-
-	if (Array.isArray(record.input)) {
-		return { ...record, input: appendDcpControlToMessages(record.input, text) }
-	}
-
-	if (Array.isArray(record.messages)) {
-		return { ...record, messages: appendDcpControlToMessages(record.messages, text) }
-	}
-
-	if ("system" in record) {
-		return { ...record, system: appendDcpControlToAnthropicSystem(record.system, text) }
-	}
-
-	if (record.config && typeof record.config === "object") {
-		const config = record.config as Record<string, unknown>
-		return {
-			...record,
-			config: {
-				...config,
-				systemInstruction: appendDcpControlToGoogleSystemInstruction(config.systemInstruction, text),
-			},
-		}
-	}
-
-	return payload
-}
-
 // ---------------------------------------------------------------------------
 // Module export
 // ---------------------------------------------------------------------------
@@ -344,15 +240,6 @@ export default async function dcpModule(pi: ExtensionAPI): Promise<void> {
 		}
 	})
 
-	// ── 7b. message_end: never persist provider-echoed DCP control markers ─────
-	pi.on("message_end", async (event, ctx) => {
-		const effectiveConfig = configForContext(ctx)
-		if (!effectiveConfig.enabled || event.message?.role !== "assistant") return undefined
-
-		const sanitized = stripStaleDcpMetadataFromAssistantMessage(event.message)
-		return { message: sanitized }
-	})
-
 	// ── 8. tool_call: record input args for dedup / purge fingerprinting ───────
 	pi.on("tool_call", async (event, _ctx) => {
 		if (!state.toolCalls.has(event.toolCallId)) {
@@ -418,7 +305,7 @@ export default async function dcpModule(pi: ExtensionAPI): Promise<void> {
 				inputMessages: event.messages.length,
 				filteredMessages: contextMessages.length,
 				outputMessages: messages.length,
-				messageIdControl: "provider-payload",
+				messageIdControl: "distributed-carriers",
 				state: summarizeDcpState(state),
 				...details,
 			}, ctx)
@@ -454,7 +341,20 @@ export default async function dcpModule(pi: ExtensionAPI): Promise<void> {
 				await saveDcpState(ctx, state)
 			}
 		}
+		const prunedToolCountBeforeCheckpoint = state.prunedToolIds.size
 		let prunedMessages = applyPruning(contextMessages, state, effectiveConfig)
+		const automaticPrunesCommitted = state.prunedToolIds.size - prunedToolCountBeforeCheckpoint
+		if (automaticPrunesCommitted > 0) {
+			const clearedAnchors = clearDcpNudgeAnchors(state)
+			await saveDcpState(ctx, state)
+			writeDcpDebugLog(effectiveConfig, "prune.tool_checkpoint", {
+				committed: automaticPrunesCommitted,
+				clearedAnchors,
+				turn: state.currentTurn,
+				blockId: state.lastAutomaticPruneBlockId,
+				state: summarizeDcpState(state),
+			}, ctx)
+		}
 		let candidate = null as ReturnType<typeof detectCompressionCandidate>
 		let messageCandidates = [] as ReturnType<typeof detectMessageCompressionCandidates>
 		let emergencySelection = null as ReturnType<typeof analyzeEmergencyCurrentTurn> | null
@@ -746,6 +646,7 @@ export default async function dcpModule(pi: ExtensionAPI): Promise<void> {
 				)
 				if (emergencyPruneResult.prunedToolCallIds.length > 0) {
 					prunedMessages = applyPruning(contextMessages, state, effectiveConfig)
+					const clearedAnchors = clearDcpNudgeAnchors(state)
 					state.consecutiveIgnoredStrongNudges = 0
 					emergencySelection = analyzeEmergencyCurrentTurn(prunedMessages, state, effectiveConfig)
 					messageCandidates = emergencyCurrentTurnMessageCandidates(emergencySelection, effectiveConfig)
@@ -758,6 +659,7 @@ export default async function dcpModule(pi: ExtensionAPI): Promise<void> {
 						targetContextPercent,
 						targetRecoveryTokens,
 						prunedOutputs: emergencyPruneResult.prunedToolCallIds.length,
+						clearedAnchors,
 						estimatedTokensRecovered: emergencyPruneResult.estimatedTokensRecovered,
 						estimatedContextPercentAfter: Math.max(
 							0,
@@ -786,7 +688,7 @@ export default async function dcpModule(pi: ExtensionAPI): Promise<void> {
 					prunedMessages,
 					state,
 					nudgeType,
-					{ contextPercent },
+					{ contextPercent, renderedReminder: nudgeText },
 				)
 				if (anchorResult.anchor) {
 					if (anchorResult.updated) {
@@ -836,10 +738,10 @@ export default async function dcpModule(pi: ExtensionAPI): Promise<void> {
 				anchor.type === "context-strong" || anchor.type === "context-soft",
 			)
 		}
-		applyAnchoredNudges(prunedMessages, state, (anchor) =>
+		const nudgeApplication = applyAnchoredNudges(prunedMessages, state, (anchor) =>
 			appendConcreteNudgeGuidance(baseNudgeText(anchor.type), candidate, messageCandidates, state),
 		)
-		if (state.nudgeAnchors.length !== anchorsBeforeFinalization) {
+		if (state.nudgeAnchors.length !== anchorsBeforeFinalization || nudgeApplication.stateChanged) {
 			await saveDcpState(ctx, state)
 		}
 
@@ -851,7 +753,7 @@ export default async function dcpModule(pi: ExtensionAPI): Promise<void> {
 		})
 	})
 
-	// ── 10b. before_provider_request: inject DCP IDs outside transcript ────────
+	// ── 10b. before_provider_request: observe accepted tool-result payloads ────
 	pi.on("before_provider_request", async (event, ctx) => {
 		const effectiveConfig = configForContext(ctx)
 		pendingProviderToolIds.clear()
@@ -868,16 +770,16 @@ export default async function dcpModule(pi: ExtensionAPI): Promise<void> {
 			}
 		}
 
-		const controlText = buildMessageIdControlText(state)
-		if (!controlText) return undefined
-
-		const payload = appendDcpControlToProviderPayload(event.payload, controlText)
 		writeDcpDebugLog(effectiveConfig, "provider_payload.message_ids", {
-			injected: payload !== event.payload,
+			injected: false,
+			delivery: "distributed-carriers",
 			pendingToolResults: pendingProviderToolIds.size,
 			state: summarizeDcpState(state),
 		}, ctx)
-		return payload === event.payload ? undefined : payload
+		// IDs are already attached to deterministic user/tool-result context
+		// carriers. Replacing the payload here would move metadata from the old
+		// tail to the new one and break strict append-only Responses continuation.
+		return undefined
 	})
 
 	// Promote pending IDs only after the provider accepted the exact payload.

package/external/pi-tools-suite/src/dcp/prompts.ts

@@ -118,9 +118,9 @@ You specify boundaries by ID using the injected metadata IDs present in the conv
 - \`mNNN\` IDs identify raw messages (3 digits, zero-padded, e.g. \`m001\`, \`m042\`)
 - \`bN\` IDs identify previously compressed blocks
 
-Current raw message IDs are provided in hidden DCP control metadata at the end of the model context.
+Raw message IDs are provided in hidden DCP control metadata attached to stable user/tool-result carriers. IDs may be sparse; determine range order from where their carriers appear in context, not from the numeric suffix.
 Some message-compression candidate hints include a low/medium/high priority; prefer high-priority stale message IDs for message-mode compression when a full range would be too broad.
-The ID reference line appears at the end of the message it belongs to — it identifies the message above it, not the one below it.
+Each carrier's metadata labels that carrier and any immediately preceding assistant message(s).
 Treat these reference lines as boundary metadata only, not as tool result content.
 
 Rules: