@pencil-agent/nano-pencil 2.0.1 → 2.0.3

package/dist/extensions/builtin/discipline/skills/test-driven-development/SKILL.md

@@ -1,32 +1,32 @@
----
-name: test-driven-development
-description: Use when implementing features, bug fixes, refactors, or behavior changes before writing production code.
----
-
-# Test-Driven Development
-
-Use red, green, refactor for behavior changes.
-
-## Gate
-
-No production behavior change without first seeing a test fail for the intended reason, unless the user explicitly accepts an exception.
-
-## Cycle
-
-1. Write the smallest test that describes one desired behavior.
-2. Run it and verify it fails for the expected reason.
-3. Implement the minimum production code needed to pass.
-4. Run the focused test and verify it passes.
-5. Run nearby or broader tests appropriate to the blast radius.
-6. Refactor only after tests are green, then re-run verification.
-
-## Good Tests
-
-- Test observable behavior, not private implementation details.
-- Prefer real code paths over mocks unless isolation requires a mock.
-- Use names that describe the behavior being guaranteed.
-- Split tests when the name needs "and".
-
-## Exceptions
-
-Acceptable exceptions include pure documentation, generated artifacts, exploratory prototypes, and configuration-only changes. State the exception before proceeding.
+---
+name: test-driven-development
+description: Use when implementing features, bug fixes, refactors, or behavior changes before writing production code.
+---
+
+# Test-Driven Development
+
+Use red, green, refactor for behavior changes.
+
+## Gate
+
+No production behavior change without first seeing a test fail for the intended reason, unless the user explicitly accepts an exception.
+
+## Cycle
+
+1. Write the smallest test that describes one desired behavior.
+2. Run it and verify it fails for the expected reason.
+3. Implement the minimum production code needed to pass.
+4. Run the focused test and verify it passes.
+5. Run nearby or broader tests appropriate to the blast radius.
+6. Refactor only after tests are green, then re-run verification.
+
+## Good Tests
+
+- Test observable behavior, not private implementation details.
+- Prefer real code paths over mocks unless isolation requires a mock.
+- Use names that describe the behavior being guaranteed.
+- Split tests when the name needs "and".
+
+## Exceptions
+
+Acceptable exceptions include pure documentation, generated artifacts, exploratory prototypes, and configuration-only changes. State the exception before proceeding.

package/dist/extensions/builtin/discipline/skills/using-git-worktrees/SKILL.md

@@ -1,25 +1,25 @@
----
-name: using-git-worktrees
-description: Use before feature work that should be isolated from the current branch, or before executing a multi-step implementation plan.
----
-
-# Using Git Worktrees
-
-Prefer isolated workspaces for risky or multi-step changes.
-
-## Process
-
-1. Detect whether the current checkout is already a linked worktree:
-   - `git rev-parse --git-dir`
-   - `git rev-parse --git-common-dir`
-   - `git rev-parse --show-superproject-working-tree`
-2. If already isolated and not a submodule, continue there.
-3. If not isolated, ask before creating a worktree unless the user or plan already requested isolation.
-4. Prefer native workspace tools when available.
-5. If using `git worktree`, place project-local worktrees under `.worktrees/` when it is ignored.
-6. Run project setup only when dependencies are missing or stale.
-7. Establish a clean baseline with the relevant tests before changing behavior.
-
-## Safety
-
-Do not remove harness-owned worktrees. Do not proceed from a failing baseline without telling the user what failed and getting direction or investigating the failure.
+---
+name: using-git-worktrees
+description: Use before feature work that should be isolated from the current branch, or before executing a multi-step implementation plan.
+---
+
+# Using Git Worktrees
+
+Prefer isolated workspaces for risky or multi-step changes.
+
+## Process
+
+1. Detect whether the current checkout is already a linked worktree:
+   - `git rev-parse --git-dir`
+   - `git rev-parse --git-common-dir`
+   - `git rev-parse --show-superproject-working-tree`
+2. If already isolated and not a submodule, continue there.
+3. If not isolated, ask before creating a worktree unless the user or plan already requested isolation.
+4. Prefer native workspace tools when available.
+5. If using `git worktree`, place project-local worktrees under `.worktrees/` when it is ignored.
+6. Run project setup only when dependencies are missing or stale.
+7. Establish a clean baseline with the relevant tests before changing behavior.
+
+## Safety
+
+Do not remove harness-owned worktrees. Do not proceed from a failing baseline without telling the user what failed and getting direction or investigating the failure.

package/dist/extensions/builtin/discipline/skills/verification-before-completion/SKILL.md

@@ -1,27 +1,27 @@
----
-name: verification-before-completion
-description: Use before saying work is complete, fixed, passing, implemented, ready, or safe to merge.
----
-
-# Verification Before Completion
-
-Evidence must precede completion claims.
-
-## Gate
-
-Do not claim success from intent, plausibility, previous output, or another agent's report. Verify against the current state.
-
-## Process
-
-1. Identify each claim you are about to make.
-2. Identify the command, file inspection, diff, runtime check, or rendered artifact that would prove it.
-3. Run or inspect that evidence freshly.
-4. Read the output, exit code, or artifact carefully.
-5. Report the actual state:
-   - If verified, name the evidence.
-   - If not verified, say what remains unverified.
-   - If failed, report the failure and continue work.
-
-## Evidence Matching
-
-Use focused checks for narrow claims and broader checks for broad claims. A passing unit test does not prove a full build, and a successful build does not prove the requested behavior unless the behavior is covered.
+---
+name: verification-before-completion
+description: Use before saying work is complete, fixed, passing, implemented, ready, or safe to merge.
+---
+
+# Verification Before Completion
+
+Evidence must precede completion claims.
+
+## Gate
+
+Do not claim success from intent, plausibility, previous output, or another agent's report. Verify against the current state.
+
+## Process
+
+1. Identify each claim you are about to make.
+2. Identify the command, file inspection, diff, runtime check, or rendered artifact that would prove it.
+3. Run or inspect that evidence freshly.
+4. Read the output, exit code, or artifact carefully.
+5. Report the actual state:
+   - If verified, name the evidence.
+   - If not verified, say what remains unverified.
+   - If failed, report the failure and continue work.
+
+## Evidence Matching
+
+Use focused checks for narrow claims and broader checks for broad claims. A passing unit test does not prove a full build, and a successful build does not prove the requested behavior unless the behavior is covered.

package/dist/extensions/builtin/discipline/skills/writing-plans/SKILL.md

@@ -1,26 +1,26 @@
----
-name: writing-plans
-description: Use when a task needs multiple implementation steps, multiple files, handoff to another agent, or careful verification sequencing.
----
-
-# Writing Plans
-
-Create an implementation plan that can be executed without rediscovering context.
-
-## Required Sections
-
-- Goal: one sentence describing the user-visible outcome.
-- Context: what the codebase currently does and why the change is needed.
-- Architecture: the recommended approach and why it fits existing boundaries.
-- Files: exact files to create or modify and each file's responsibility.
-- Tasks: small ordered steps with verification after meaningful changes.
-- Test plan: exact commands and expected evidence.
-- Documentation impact: P1/P2/P3 or user docs that must change.
-
-## Task Quality
-
-Each task should be independently understandable. Include exact paths, APIs, data shapes, and expected behavior. Avoid placeholders such as "handle edge cases" or "add tests"; state the actual edge cases and tests.
-
-## Handoff
-
-When the plan is approved, execute it directly or use `executing-plans` for inline execution. Use subagents when tasks are independent and reviewable.
+---
+name: writing-plans
+description: Use when a task needs multiple implementation steps, multiple files, handoff to another agent, or careful verification sequencing.
+---
+
+# Writing Plans
+
+Create an implementation plan that can be executed without rediscovering context.
+
+## Required Sections
+
+- Goal: one sentence describing the user-visible outcome.
+- Context: what the codebase currently does and why the change is needed.
+- Architecture: the recommended approach and why it fits existing boundaries.
+- Files: exact files to create or modify and each file's responsibility.
+- Tasks: small ordered steps with verification after meaningful changes.
+- Test plan: exact commands and expected evidence.
+- Documentation impact: P1/P2/P3 or user docs that must change.
+
+## Task Quality
+
+Each task should be independently understandable. Include exact paths, APIs, data shapes, and expected behavior. Avoid placeholders such as "handle edge cases" or "add tests"; state the actual edge cases and tests.
+
+## Handoff
+
+When the plan is approved, execute it directly or use `executing-plans` for inline execution. Use subagents when tasks are independent and reviewable.

package/dist/extensions/builtin/goal/README.md

@@ -1,67 +1,67 @@
-# Goal Extension
-
-Long-running task management for nanoPencil. Set a goal with `/goal <objective>` and the
-agent will auto-continue working on it across turns until the objective is achieved,
-the token budget runs out, or you pause/clear it.
-
-This extension mirrors the Codex `/goal` command semantics: a per-thread goal, persisted
-to disk, with idle-continuation prompts, token accounting, and budget enforcement.
-
-## Usage
-
-```
-/goal                  Show current goal summary menu
-/goal <objective>      Set or replace the goal
-/goal clear            Clear the goal
-/goal edit             Open the editor to change the objective
-/goal pause            Pause auto-continuation
-/goal resume           Resume auto-continuation
-/goal help             Show usage help
-```
-
-## LLM Tools
-
-The extension registers three LLM-facing tools:
-
-| Tool | Purpose | Who can call |
-|------|---------|--------------|
-| `get_goal` | Read the current goal | LLM |
-| `create_goal` | Create a new goal (only when the user explicitly asks) | LLM |
-| `update_goal` | Mark the goal `complete` or `blocked` | LLM |
-
-The LLM is only allowed to set the goal's status to `complete` or `blocked`. Pause /
-resume / budget limits are user-driven and happen exclusively through `/goal`.
-
-## Lifecycle
-
-The extension subscribes to `turn_start`, `turn_end`, `message_end`, `tool_execution_end`,
-and `agent_end` to track token usage and time per turn. When a turn ends with an
-`active` goal, the extension injects a follow-up user message containing the
-continuation prompt so the agent keeps working on the objective.
-
-When a turn causes the goal to cross its token budget, the extension injects a
-budget-limit steering prompt and marks the goal `budget_limited`. Once budget-limited,
-auto-continuation stops and the goal is terminal.
-
-## Persistence
-
-Goals are stored as JSON files under `<agentDir>/goals/<threadId>.json`. They survive
-session restarts and are keyed by the active session ID.
-
-## Status
-
-`Status: active` shows in the footer while a goal is running.
-
-## Architecture
-
-| File | Responsibility |
-|------|----------------|
-| `goal-types.ts` | `ThreadGoalStatus`, `ThreadGoal`, helper predicates |
-| `goal-store.ts` | Atomic JSON-file persistence (replace / insert / update / delete / account_usage) |
-| `goal-format.ts` | Time/token formatting, summary lines, status indicator, validators |
-| `goal-prompts.ts` | Continuation / budget-limit / objective-updated prompt templates |
-| `goal-controller.ts` | Per-thread runtime: mutex, turn accounting, idle continuation |
-| `goal-tools.ts` | `get_goal`, `create_goal`, `update_goal` LLM tool definitions |
-| `goal-parser.ts` | `/goal` slash-command argument parsing |
-| `goal-command.ts` | `/goal` slash-command handler (UI + controller dispatch) |
-| `index.ts` | Extension entry: tools, command, lifecycle hooks, status indicator |
+# Goal Extension
+
+Long-running task management for nanoPencil. Set a goal with `/goal <objective>` and the
+agent will auto-continue working on it across turns until the objective is achieved,
+the token budget runs out, or you pause/clear it.
+
+This extension mirrors the Codex `/goal` command semantics: a per-thread goal, persisted
+to disk, with idle-continuation prompts, token accounting, and budget enforcement.
+
+## Usage
+
+```
+/goal                  Show current goal summary menu
+/goal <objective>      Set or replace the goal
+/goal clear            Clear the goal
+/goal edit             Open the editor to change the objective
+/goal pause            Pause auto-continuation
+/goal resume           Resume auto-continuation
+/goal help             Show usage help
+```
+
+## LLM Tools
+
+The extension registers three LLM-facing tools:
+
+| Tool | Purpose | Who can call |
+|------|---------|--------------|
+| `get_goal` | Read the current goal | LLM |
+| `create_goal` | Create a new goal (only when the user explicitly asks) | LLM |
+| `update_goal` | Mark the goal `complete` or `blocked` | LLM |
+
+The LLM is only allowed to set the goal's status to `complete` or `blocked`. Pause /
+resume / budget limits are user-driven and happen exclusively through `/goal`.
+
+## Lifecycle
+
+The extension subscribes to `turn_start`, `turn_end`, `message_end`, `tool_execution_end`,
+and `agent_end` to track token usage and time per turn. When a turn ends with an
+`active` goal, the extension injects a follow-up user message containing the
+continuation prompt so the agent keeps working on the objective.
+
+When a turn causes the goal to cross its token budget, the extension injects a
+budget-limit steering prompt and marks the goal `budget_limited`. Once budget-limited,
+auto-continuation stops and the goal is terminal.
+
+## Persistence
+
+Goals are stored as JSON files under `<agentDir>/goals/<threadId>.json`. They survive
+session restarts and are keyed by the active session ID.
+
+## Status
+
+`Status: active` shows in the footer while a goal is running.
+
+## Architecture
+
+| File | Responsibility |
+|------|----------------|
+| `goal-types.ts` | `ThreadGoalStatus`, `ThreadGoal`, helper predicates |
+| `goal-store.ts` | Atomic JSON-file persistence (replace / insert / update / delete / account_usage) |
+| `goal-format.ts` | Time/token formatting, summary lines, status indicator, validators |
+| `goal-prompts.ts` | Continuation / budget-limit / objective-updated prompt templates |
+| `goal-controller.ts` | Per-thread runtime: mutex, turn accounting, idle continuation |
+| `goal-tools.ts` | `get_goal`, `create_goal`, `update_goal` LLM tool definitions |
+| `goal-parser.ts` | `/goal` slash-command argument parsing |
+| `goal-command.ts` | `/goal` slash-command handler (UI + controller dispatch) |
+| `index.ts` | Extension entry: tools, command, lifecycle hooks, status indicator |

package/dist/extensions/builtin/goal/index.js

@@ -1,10 +1,10 @@
-var U=Object.defineProperty;var i=(t,r)=>U(t,"name",{value:r,configurable:!0});import*as B from"node:fs";import*as F from"node:os";import*as O from"node:path";import{Box as N,Container as j,Spacer as D,Text as K}from"@pencil-agent/tui";import{runGoalCommand as H}from"./goal-command.js";import{GoalController as J}from"./goal-controller.js";import{getGoalArgumentCompletions as Q}from"./goal-parser.js";import{buildAllGoalTools as X,setGoalToolHost as q}from"./goal-tools.js";import{goalStatusIndicator as G,goalSummaryLines as z,goalUsageSummary as W}from"./goal-format.js";import{isActiveStatus as C}from"./goal-types.js";const y="goal",Y="weak-model-compatible",V=O.join(F.homedir(),".nanopencil","agent","nanopencil-debug.log");function g(t){B.appendFileSync(V,`[${new Date().toISOString()}] [goal] ${t}
-`)}i(g,"dbg");function w(t,r){return t.length>r?t.slice(0,r)+"\u2026":t}i(w,"truncate");function Z(t){const r=t.content;if(typeof r=="string")return w(r,150);if(Array.isArray(r)){const c=r.filter(u=>u.type==="text"&&typeof u.text=="string").map(u=>u.text);if(c.length>0)return w(c.join(" "),150);const m=r.filter(u=>u.type==="tool_use").map(u=>`[tool_use:${u.name??"?"}]`);if(m.length>0)return m.join(" ")}return"(empty)"}i(Z,"extractContentPreview");const h=new Map;let k=null,S=null,_=null,$=null;function tt(t,r){const c=r.sessionManager.getSessionId();if(!c)return null;const m=`${r.agentDir}/${c}`;k=m;let u=h.get(m);return u||(u=new J(t,c),h.set(m,u)),S=u,u}i(tt,"resolveController");const et={getController(t,r){const c=`${t}/${r}`;return h.get(c)??null}};function nt(t){return"normal"}i(nt,"detectRunKind");function ot(t){return!!t.getSettings().plan}i(ot,"isPlanMode");function st(t){let r=0;for(const c of t){if(c.role!=="assistant")continue;const u=c.usage;u&&typeof u.totalTokens=="number"&&(r+=u.totalTokens)}return r}i(st,"getRunningTotalTokens");async function rt(t){q(et);const r=i(o=>{const s=tt(t,o);return s&&(S=s),s},"ensureController"),[c,m,u]=X();t.registerTool(c),t.registerTool(m),t.registerTool(u),t.registerMessageRenderer(y,(o,s,e)=>{const l=typeof o.content=="string"?o.content:Array.isArray(o.content)?o.content.filter(a=>a.type==="text").map(a=>a.text).join(`
-`):JSON.stringify(o.content??""),d=new N(1,1,a=>e.bg("customMessageBg",a));d.addChild(new K(e.fg("dim",l),0,0));const n=new j;return n.addChild(new D(1)),n.addChild(d),n}),t.registerCommand("goal",{description:"Set, show, edit, pause, resume, or clear the thread goal.",getArgumentCompletions:Q,handler:i(async(o,s)=>{g(`/goal command invoked: args="${o}"`);const e=r(s);await H(o,s,e),_=null,$=null,g("/goal command done, reportedTerminalGoalId reset")},"handler")}),t.on("session_start",(o,s)=>{g("session_start"),r(s)?.resetIdleContinuationFlag(),_=null,$=null}),t.on("session_shutdown",()=>{k&&h.delete(k),k=null,S=null});const E=i((o,s)=>{g(`turn_start index=${o.turnIndex} timestamp=${o.timestamp}`);const e=r(s);if(!e)return;const l=ot(s)?"plan":"normal";e.on_turn_start(`turn-${o.turnIndex}-${o.timestamp}`,l,0)},"onTurnStart");t.on("turn_start",E);const I=i((o,s)=>{r(s);const e=o.message;g(`message_start role=${e.role??"?"}`)},"onMessageStart");t.on("message_start",I);const A=i((o,s)=>{const e=r(s);if(!e)return;const l=o.message,d=Z(l);if(g(`message_end role=${l.role??"?"} tokens=${l.usage?.totalTokens??"?"} content=${d}`),l.role!=="assistant")return;const n=l.usage?.totalTokens;if(typeof n!="number")return;const a=e.on_token_usage(n);if(a.crossed&&a.goal){const f=e.maybe_build_budget_limit_steering();f&&t.sendMessage({customType:y,content:f,display:!0,details:{kind:"budget_limit",goal:a.goal}})}},"onMessageEnd");t.on("message_end",A);const M=i((o,s)=>{g(`tool_execution_end name=${o.toolName} isError=${o.isError} result=${w(JSON.stringify(o.result??""),200)}`);const e=r(s);e&&(o.isError||e.on_tool_finish(o.toolName))},"onToolExecutionEnd");t.on("tool_execution_end",M);const P=i(async(o,s)=>{try{const e=r(s);if(!e)return;const l=e.currentTurnSnapshot(),d=l?((Date.now()-l.turnStartedAt)/1e3).toFixed(1):"?";g(`turn_end BEGIN index=${o.turnIndex} duration=${d}s tokens=${l?.tokensNow??"?"}`);const n=await e.on_turn_end();if(g(`turn_end RESULT: reason=${n.reason} goalId=${n.goal?.goal_id} goalStatus=${n.goal?.status} reportedTerminalGoalId=${_}`),n.reason!=="not_active_status"||!n.goal)return;const a=n.goal.status;if(a==="complete"||a==="blocked"||a==="budget_limited"||a==="paused"){g(`turn_end \u2192 clearing followUp queue (goal is ${a})`);try{t.clearFollowUpQueue()}catch(R){g(`turn_end \u2192 clearFollowUpQueue FAILED: ${R}`)}}const f=n.goal.goal_id,T=n.goal.status;if(_===f&&$===T)return;const p=G(n.goal,null);(p.type==="Complete"||p.type==="BudgetLimited")&&(_=f,$=T,g(`turn_end \u2192 sending terminal message: ${p.type} for goal ${f}`),t.sendMessage({customType:y,content:`Goal ${p.type==="Complete"?"complete":"budget_limited"}.
+var R=Object.defineProperty;var i=(t,r)=>R(t,"name",{value:r,configurable:!0});import*as E from"node:fs";import*as F from"node:os";import*as G from"node:path";import{Box as j,Container as D,Spacer as K,Text as H}from"@pencil-agent/tui";import{runGoalCommand as J}from"./goal-command.js";import{GoalController as Q}from"./goal-controller.js";import{getGoalArgumentCompletions as X}from"./goal-parser.js";import{buildAllGoalTools as q,setGoalToolHost as z}from"./goal-tools.js";import{goalStatusIndicator as C,goalSummaryLines as W,goalUsageSummary as Y}from"./goal-format.js";import{isActiveStatus as I}from"./goal-types.js";const y="goal",V="weak-model-compatible",Z=process.env.NANOPENCIL_DEBUG==="1",A=G.join(F.homedir(),".nanopencil","agent","nanopencil-debug.log");function g(t){if(Z)try{E.mkdirSync(G.dirname(A),{recursive:!0}),E.appendFileSync(A,`[${new Date().toISOString()}] [goal] ${t}
+`)}catch{}}i(g,"dbg");function w(t,r){return t.length>r?t.slice(0,r)+"\u2026":t}i(w,"truncate");function tt(t){const r=t.content;if(typeof r=="string")return w(r,150);if(Array.isArray(r)){const c=r.filter(u=>u.type==="text"&&typeof u.text=="string").map(u=>u.text);if(c.length>0)return w(c.join(" "),150);const m=r.filter(u=>u.type==="tool_use").map(u=>`[tool_use:${u.name??"?"}]`);if(m.length>0)return m.join(" ")}return"(empty)"}i(tt,"extractContentPreview");const h=new Map;let k=null,S=null,_=null,$=null;function et(t,r){const c=r.sessionManager.getSessionId();if(!c)return null;const m=`${r.agentDir}/${c}`;k=m;let u=h.get(m);return u||(u=new Q(t,c),h.set(m,u)),S=u,u}i(et,"resolveController");const nt={getController(t,r){const c=`${t}/${r}`;return h.get(c)??null}};function ot(t){return"normal"}i(ot,"detectRunKind");function st(t){return!!t.getSettings().plan}i(st,"isPlanMode");function rt(t){let r=0;for(const c of t){if(c.role!=="assistant")continue;const u=c.usage;u&&typeof u.totalTokens=="number"&&(r+=u.totalTokens)}return r}i(rt,"getRunningTotalTokens");async function at(t){z(nt);const r=i(o=>{const s=et(t,o);return s&&(S=s),s},"ensureController"),[c,m,u]=q();t.registerTool(c),t.registerTool(m),t.registerTool(u),t.registerMessageRenderer(y,(o,s,e)=>{const l=typeof o.content=="string"?o.content:Array.isArray(o.content)?o.content.filter(a=>a.type==="text").map(a=>a.text).join(`
+`):JSON.stringify(o.content??""),d=new j(1,1,a=>e.bg("customMessageBg",a));d.addChild(new H(e.fg("dim",l),0,0));const n=new D;return n.addChild(new K(1)),n.addChild(d),n}),t.registerCommand("goal",{description:"Set, show, edit, pause, resume, or clear the thread goal.",getArgumentCompletions:X,handler:i(async(o,s)=>{g(`/goal command invoked: args="${o}"`);const e=r(s);await J(o,s,e),_=null,$=null,g("/goal command done, reportedTerminalGoalId reset")},"handler")}),t.on("session_start",(o,s)=>{g("session_start"),r(s)?.resetIdleContinuationFlag(),_=null,$=null}),t.on("session_shutdown",()=>{k&&h.delete(k),k=null,S=null});const P=i((o,s)=>{g(`turn_start index=${o.turnIndex} timestamp=${o.timestamp}`);const e=r(s);if(!e)return;const l=st(s)?"plan":"normal";e.on_turn_start(`turn-${o.turnIndex}-${o.timestamp}`,l,0)},"onTurnStart");t.on("turn_start",P);const M=i((o,s)=>{r(s);const e=o.message;g(`message_start role=${e.role??"?"}`)},"onMessageStart");t.on("message_start",M);const v=i((o,s)=>{const e=r(s);if(!e)return;const l=o.message,d=tt(l);if(g(`message_end role=${l.role??"?"} tokens=${l.usage?.totalTokens??"?"} content=${d}`),l.role!=="assistant")return;const n=l.usage?.totalTokens;if(typeof n!="number")return;const a=e.on_token_usage(n);if(a.crossed&&a.goal){const f=e.maybe_build_budget_limit_steering();f&&t.sendMessage({customType:y,content:f,display:!0,details:{kind:"budget_limit",goal:a.goal}})}},"onMessageEnd");t.on("message_end",v);const L=i((o,s)=>{g(`tool_execution_end name=${o.toolName} isError=${o.isError} result=${w(JSON.stringify(o.result??""),200)}`);const e=r(s);e&&(o.isError||e.on_tool_finish(o.toolName))},"onToolExecutionEnd");t.on("tool_execution_end",L);const N=i(async(o,s)=>{try{const e=r(s);if(!e)return;const l=e.currentTurnSnapshot(),d=l?((Date.now()-l.turnStartedAt)/1e3).toFixed(1):"?";g(`turn_end BEGIN index=${o.turnIndex} duration=${d}s tokens=${l?.tokensNow??"?"}`);const n=await e.on_turn_end();if(g(`turn_end RESULT: reason=${n.reason} goalId=${n.goal?.goal_id} goalStatus=${n.goal?.status} reportedTerminalGoalId=${_}`),n.reason!=="not_active_status"||!n.goal)return;const a=n.goal.status;if(a==="complete"||a==="blocked"||a==="budget_limited"||a==="paused"){g(`turn_end \u2192 clearing followUp queue (goal is ${a})`);try{t.clearFollowUpQueue()}catch(O){g(`turn_end \u2192 clearFollowUpQueue FAILED: ${O}`)}}const f=n.goal.goal_id,T=n.goal.status;if(_===f&&$===T)return;const p=C(n.goal,null);(p.type==="Complete"||p.type==="BudgetLimited")&&(_=f,$=T,g(`turn_end \u2192 sending terminal message: ${p.type} for goal ${f}`),t.sendMessage({customType:y,content:`Goal ${p.type==="Complete"?"complete":"budget_limited"}.
 ${b(n.goal)}`,display:!0,details:{kind:p.type.toLowerCase(),goal:n.goal}},{triggerTurn:!1}))}catch(e){g(`turn_end EXCEPTION: ${e instanceof Error?e.message:String(e)}
-${e instanceof Error?e.stack:""}`)}},"onTurnEnd");t.on("turn_end",P);const L=i(async(o,s)=>{try{const e=r(s);if(!e)return;const l=st(o.messages);e.on_token_usage(l);const d=e.runStopReason;if(g(`agent_end stopReason=${d??"unknown"}`),d==="aborted")return;if(d==="error"){const a=e.on_turn_error();a&&a.status==="blocked"&&(g(`agent_end \u2192 goal ${a.goal_id} blocked after run error`),t.sendMessage({customType:y,content:`Goal blocked: the run ended with an error. Use /goal resume to continue.
+${e instanceof Error?e.stack:""}`)}},"onTurnEnd");t.on("turn_end",N);const U=i(async(o,s)=>{try{const e=r(s);if(!e)return;const l=rt(o.messages);e.on_token_usage(l);const d=e.runStopReason;if(g(`agent_end stopReason=${d??"unknown"}`),d==="aborted")return;if(d==="error"){const a=e.on_turn_error();a&&a.status==="blocked"&&(g(`agent_end \u2192 goal ${a.goal_id} blocked after run error`),t.sendMessage({customType:y,content:`Goal blocked: the run ended with an error. Use /goal resume to continue.
 ${b(a)}`,display:!0,details:{kind:"blocked_on_error",goal:a}},{triggerTurn:!1}));return}const n=e.maybe_dispatch_continuation({hasPendingMessages:s.hasPendingMessages()});if(g(`agent_end DISPATCH: dispatched=${n.dispatched} reason=${n.reason} goalStatus=${n.goal?.status}`),n.dispatched)t.sendMessage({customType:y,content:`Goal continuation dispatched.
 ${b(n.goal)}`,display:!0,details:{kind:"continuation",goal:n.goal}},{triggerTurn:!1});else if(n.goal&&n.reason==="continuation_limit_reached")s.hasUI&&s.ui.setStatus("goal","Continuation limit \u2014 /goal resume to continue");else if(n.goal&&n.reason==="total_continuation_limit_reached"){if(_===n.goal.goal_id&&$==="paused")return;_=n.goal.goal_id,$="paused",t.sendMessage({customType:y,content:`Goal auto-paused after ${n.consecutiveContinuations??"?"} continuation turns. Use /goal resume to continue.
 ${b(n.goal)}`,display:!0,details:{kind:"total_continuation_limit",goal:n.goal}},{triggerTurn:!1})}}catch(e){g(`agent_end EXCEPTION: ${e instanceof Error?e.message:String(e)}
-${e instanceof Error?e.stack:""}`)}},"onAgentEnd");t.on("agent_end",L);const v=i((o,s)=>{const e=r(s);if(!e)return;e.note_run_stop_reason(o.stopReason);const l=o.usage;l&&typeof l.totalTokens=="number"&&e.on_token_usage(l.totalTokens);const d=o.loopFramework;nt(d)},"onAgentResult");t.on("agent_result",v);const x=i(async(o,s)=>{const e=r(s);if(!e)return;const l=await e.get_goal();!l||!C(l.status)||(g(`agent_abort \u2192 pausing goal ${l.goal_id}`),await e.set_status("paused"),_=null,$=null,s.hasUI&&s.ui.notify("Goal paused (agent aborted).","info"))},"onAgentAbort");t.on("agent_abort",x),t.on("session_start",(o,s)=>{if(!s.hasUI)return;const l=setInterval(i(()=>{const n=S;n&&n.get_goal().then(a=>{if(!a||!C(a.status)&&a.status!=="budget_limited"&&a.status!=="complete"){s.ui.setStatus("goal",void 0);return}const f=n.currentTurnSnapshot()?.turnStartedAt??null,T=G(a,f),p=at(T);s.ui.setStatus("goal",p)}).catch(()=>{})},"tick"),200),d=i(()=>clearInterval(l),"cleanup");t.on("session_shutdown",()=>d())})}i(rt,"goalExtension");function at(t){switch(t.type){case"Active":return`Pursuing goal (${t.usage})`;case"Paused":return"Goal paused (/goal resume)";case"Blocked":return"Goal blocked (/goal resume)";case"UsageLimited":return"Goal hit usage limits (/goal resume)";case"BudgetLimited":return t.usage?`Goal unmet (${t.usage})`:"Goal abandoned";case"Complete":return`Goal achieved (${t.usage})`}}i(at,"indicatorLabel");function b(t){if(!t)return"";const r=z(t),c=W(t);return[`  Status: ${t.status}`,`  Objective: ${t.objective}`,`  Time used: ${c.elapsed}`,`  Tokens used: ${c.tokensLabel}${c.hasBudget?" tokens":""}`].join(`
-`)}i(b,"summarizeGoalStatus");export{rt as default};
+${e instanceof Error?e.stack:""}`)}},"onAgentEnd");t.on("agent_end",U);const x=i((o,s)=>{const e=r(s);if(!e)return;e.note_run_stop_reason(o.stopReason);const l=o.usage;l&&typeof l.totalTokens=="number"&&e.on_token_usage(l.totalTokens);const d=o.loopFramework;ot(d)},"onAgentResult");t.on("agent_result",x);const B=i(async(o,s)=>{const e=r(s);if(!e)return;const l=await e.get_goal();!l||!I(l.status)||(g(`agent_abort \u2192 pausing goal ${l.goal_id}`),await e.set_status("paused"),_=null,$=null,s.hasUI&&s.ui.notify("Goal paused (agent aborted).","info"))},"onAgentAbort");t.on("agent_abort",B),t.on("session_start",(o,s)=>{if(!s.hasUI)return;const l=setInterval(i(()=>{const n=S;n&&n.get_goal().then(a=>{if(!a||!I(a.status)&&a.status!=="budget_limited"&&a.status!=="complete"){s.ui.setStatus("goal",void 0);return}const f=n.currentTurnSnapshot()?.turnStartedAt??null,T=C(a,f),p=lt(T);s.ui.setStatus("goal",p)}).catch(()=>{})},"tick"),200),d=i(()=>clearInterval(l),"cleanup");t.on("session_shutdown",()=>d())})}i(at,"goalExtension");function lt(t){switch(t.type){case"Active":return`Pursuing goal (${t.usage})`;case"Paused":return"Goal paused (/goal resume)";case"Blocked":return"Goal blocked (/goal resume)";case"UsageLimited":return"Goal hit usage limits (/goal resume)";case"BudgetLimited":return t.usage?`Goal unmet (${t.usage})`:"Goal abandoned";case"Complete":return`Goal achieved (${t.usage})`}}i(lt,"indicatorLabel");function b(t){if(!t)return"";const r=W(t),c=Y(t);return[`  Status: ${t.status}`,`  Objective: ${t.objective}`,`  Time used: ${c.elapsed}`,`  Tokens used: ${c.tokensLabel}${c.hasBudget?" tokens":""}`].join(`
+`)}i(b,"summarizeGoalStatus");export{at as default};