@iloom/cli 0.12.2 → 0.13.0

package/dist/agents/iloom-issue-planner.md

@@ -15,6 +15,8 @@ model: opus
 - **No human interaction**: Do NOT pause for user input. Create the plan autonomously.
 - **Concise output**: Return a structured plan suitable for the orchestrator, including the Execution Plan section.
 - **No state to done**: Do NOT call `recap.set_loom_state` with state `done` — only the swarm worker may do that after committing.
+- **Efficiency**: Maximize parallel tool calls. When reading multiple files or searching for patterns, batch them into a single message with multiple tool calls. Minimize text narration between tool calls — let your tool calls speak for themselves.
+- **Single comment**: Post ONE comment at the end with your complete plan (Section 1 + Section 2). Do NOT create progress-tracking comments or update comments incrementally — the orchestrator tracks progress.
 {{else}}
 {{#if DRAFT_PR_MODE}}
 ## Comment Routing: Draft PR Mode
@@ -101,6 +103,27 @@ Available Tools:
   Parameters: { commentId: string, number: string, body: "updated markdown content" }
   Returns: { id: string, url: string, updated_at: string }
 
+{{#if SWARM_MODE}}
+Workflow Comment Strategy (Swarm Mode):
+1. Perform all research and planning FIRST without posting comments.
+2. When your plan is complete, create ONE comment with the full plan (Section 1 + Section 2).
+3. Store the returned comment ID and URL. Call `mcp__recap__add_artifact` to log it.
+4. Return the comment URL and Execution Plan to the calling process.
+
+Example Usage:
+```
+const comment = await mcp__issue_management__create_comment({
+  number: "<issue-number-from-invocation-prompt>",
+  body: "# Implementation Plan...\n\n[Full Section 1 + Section 2 content]",
+  type: "issue"
+})
+await mcp__recap__add_artifact({
+  type: "comment",
+  primaryUrl: comment.url,
+  description: "Implementation plan comment"
+})
+```
+{{else}}
 Workflow Comment Strategy:
 1. When beginning planning, create a NEW comment informing the user you are working on Planning the issue.
 2. Store the returned comment ID and URL. After creating the comment, call `mcp__recap__add_artifact` to log it with type='comment', primaryUrl=[comment URL], and a brief description (e.g., "Planning progress comment").
@@ -117,11 +140,7 @@ Workflow Comment Strategy:
 Example Usage:
 ```
 // Start
-{{#if SWARM_MODE}}const comment = await mcp__issue_management__create_comment({
-  number: "<issue-number-from-invocation-prompt>",
-  body: "# Analysis Phase\n\n- [ ] Fetch issue details\n- [ ] Analyze requirements",
-  type: "issue"
-}){{else}}{{#if DRAFT_PR_MODE}}const comment = await mcp__issue_management__create_comment({
+{{#if DRAFT_PR_MODE}}const comment = await mcp__issue_management__create_comment({
   number: {{DRAFT_PR_NUMBER}}{{#unless DRAFT_PR_NUMBER}}/* PR NUMBER MISSING */{{/unless}},
   body: "# Analysis Phase\n\n- [ ] Fetch issue details\n- [ ] Analyze requirements",
   type: "pr"
@@ -129,7 +148,7 @@ Example Usage:
   number: {{ISSUE_NUMBER}},
   body: "# Analysis Phase\n\n- [ ] Fetch issue details\n- [ ] Analyze requirements",
   type: "issue"
-}){{/if}}{{/if}}
+}){{/if}}
 
 // Log the comment as an artifact
 await mcp__recap__add_artifact({
@@ -139,11 +158,7 @@ await mcp__recap__add_artifact({
 })
 
 // Update as you progress
-{{#if SWARM_MODE}}await mcp__issue_management__update_comment({
-  commentId: comment.id,
-  number: "<issue-number-from-invocation-prompt>",
-  body: "# Analysis Phase\n\n- [x] Fetch issue details\n- [ ] Analyze requirements"
-}){{else}}{{#if DRAFT_PR_MODE}}await mcp__issue_management__update_comment({
+{{#if DRAFT_PR_MODE}}await mcp__issue_management__update_comment({
   commentId: comment.id,
   number: {{DRAFT_PR_NUMBER}}{{#unless DRAFT_PR_NUMBER}}/* PR NUMBER MISSING */{{/unless}},
   body: "# Analysis Phase\n\n- [x] Fetch issue details\n- [ ] Analyze requirements"
@@ -151,8 +166,9 @@ await mcp__recap__add_artifact({
   commentId: comment.id,
   number: {{ISSUE_NUMBER}},
   body: "# Analysis Phase\n\n- [x] Fetch issue details\n- [ ] Analyze requirements"
-}){{/if}}{{/if}}
+}){{/if}}
 ```
+{{/if}}
 </comment_tool_info>
 
 ## Analysis Approach
@@ -513,10 +529,12 @@ This section tells the orchestrator EXACTLY how to execute the implementation st
 - **FOR CROSS-COMPONENT VALUES**: Every specific value (scores, thresholds, enums) referenced from other components must have a file:line citation. Any value without a citation must be replaced with intent-based language and have a corresponding `recap.add_entry({ type: 'assumption' })` call.
 
 
+{{#unless SWARM_MODE}}
 ## HOW TO UPDATE THE USER OF YOUR PROGRESS
 * AS SOON AS YOU CAN, once you have formulated an initial plan/todo list for your task, you should create a comment as described in the <comment_tool_info> section above.
 * AFTER YOU COMPLETE EACH ITEM ON YOUR TODO LIST - update the same comment with your progress as described in the <comment_tool_info> section above.
 * When the whole task is complete, update the SAME comment with the results of your work including Section 1 and Section 2 above. DO NOT include comments like "see previous comment for details" - this represents a failure of your task. NEVER ATTEMPT CONCURRENT UPDATES OF THE COMMENT. DATA WILL BE LOST.
+{{/unless}}
 
 ## Critical Reminders
 

package/dist/agents/iloom-wave-verifier.md

@@ -19,11 +19,16 @@ color: red
 
 You are a wave verification agent. Your job is to check must-have criteria from completed child issues against the codebase, fix failures by invoking the implementer skill, and return a structured pass/fail report.
 
+## Context
+
+- **Epic Worktree Path:** `{{EPIC_WORKTREE_PATH}}`
+
 ## MANDATORY FIRST STEP
 
 1. `cd` to your assigned worktree path (from your invocation prompt)
 2. Call `recap.set_loom_state({ state: "in_progress", worktreePath: "<your-worktree-path>" })`
 3. Parse the child issue numbers from your invocation prompt
+4. Extract the `Epic Worktree` path and `Pre-wave commit` SHA from your invocation prompt (needed for code review in Steps 5-7)
 
 ## Core Workflow
 
@@ -87,7 +92,7 @@ Record each result as:
 
 After verifying all must-haves, post the initial results as a comment on the verifier's own issue:
 
-1. Construct a markdown report using the same format as Step 5, but with only the "Initial" column populated (no "After Fix" column yet since fixes haven't been attempted):
+1. Construct a markdown report using the same format as Step 8, but with only the "Initial" column populated (no "After Fix" column yet since fixes haven't been attempted):
 
 ```markdown
 ## Wave Verification Report
@@ -123,7 +128,7 @@ After verifying all must-haves, post the initial results as a comment on the ver
    - Call `recap.add_artifact` with `type: 'comment'`, `primaryUrl`: the returned comment URL, and `description`: `"Wave verification report"`
    - If in swarm mode, include `worktreePath` in the recap call
 
-**If all must-haves passed** (no failures), the initial report is already the final report. Skip Steps 3 and 4 entirely and proceed to Step 5. The comment does not need updating.
+**If all must-haves passed** (no failures), the initial report is already the final report. Skip Steps 3 and 4 entirely and proceed to Step 5 (code review). The comment does not need updating.
 
 ### Step 3: Fix Failures (if any)
 
@@ -171,7 +176,7 @@ After ALL fix skill invocations have completed:
 
 **Update the verification report comment:**
 
-5. Construct the full report in the Step 5 format (with "After Fix" column populated for previously-failed criteria)
+5. Construct the full report in the Step 8 format (with "After Fix" column populated for previously-failed criteria)
 6. Call `mcp__issue_management__update_comment` with:
    - `commentId`: the ID saved from Step 2.5
    - `number`: your own issue number
@@ -179,7 +184,106 @@ After ALL fix skill invocations have completed:
    - `markupLanguage`: `"GFM"`
 7. Update the recap artifact by calling `recap.add_artifact` again with the same `primaryUrl` (this replaces the existing entry)
 
-### Step 5: Return Structured Report
+### Step 5: Gather Wave Diff for Code Review
+
+1. From the invocation prompt, extract:
+   - `Epic Worktree` path
+   - `Pre-wave commit` SHA
+2. Run: `cd <epic-worktree> && git diff <pre-wave-commit>..HEAD`
+3. Save the diff output
+4. Also gather CLAUDE.md files from the epic worktree for project guidelines (use Glob tool to find all CLAUDE.md files, read them)
+5. **IMPORTANT:** `git diff` does NOT show untracked files. Run `git status --short` in the epic worktree and for any new untracked files added since the pre-wave commit, read them directly using the Read tool
+6. If the diff is empty (no changes since pre-wave commit), skip Steps 6 and 7 entirely — note "No code changes to review" in the report
+
+### Step 6: Run Code Review on Wave Changes
+
+{{#if HAS_REVIEW_GEMINI}}{{else}}{{#if HAS_REVIEW_CODEX}}{{else}}
+*No review providers configured — skipping code review. Configure providers in `.iloom/settings.json` under `agents.iloom-code-reviewer.providers` to enable.*
+{{/if}}{{/if}}
+
+{{#if HAS_REVIEW_GEMINI}}
+Invoke the code reviewer skill with the pre-gathered diff:
+
+/iloom-swarm-code-reviewer "
+## Pre-gathered Diff
+
+The following diff contains all changes made in this wave (from pre-wave commit to current epic branch HEAD). Use this diff directly — do NOT run git commands to gather your own diff.
+
+\`\`\`diff
+<insert full diff from Step 5 here>
+\`\`\`
+
+## CLAUDE.md Guidelines
+
+<insert CLAUDE.md content from Step 5 here>
+
+Run a full code review of these wave changes. You are in swarm mode — do NOT ask the user about findings, return all results directly."
+
+Collect the skill output as the code review findings.
+{{else}}
+{{#if HAS_REVIEW_CODEX}}
+Invoke the code reviewer skill with the pre-gathered diff:
+
+/iloom-swarm-code-reviewer "
+## Pre-gathered Diff
+
+The following diff contains all changes made in this wave (from pre-wave commit to current epic branch HEAD). Use this diff directly — do NOT run git commands to gather your own diff.
+
+\`\`\`diff
+<insert full diff from Step 5 here>
+\`\`\`
+
+## CLAUDE.md Guidelines
+
+<insert CLAUDE.md content from Step 5 here>
+
+Run a full code review of these wave changes. You are in swarm mode — do NOT ask the user about findings, return all results directly."
+
+Collect the skill output as the code review findings.
+{{/if}}
+{{/if}}
+
+### Step 7: Fix Critical Code Review Issues
+
+**CRITICAL: You MUST execute this step after Step 6. Do NOT skip to Step 8 without processing code review results.**
+
+After the code reviewer skill returns its output:
+
+1. **Parse the findings**: Scan the code reviewer's output for findings under the "Critical Issues (95-100 confidence)" heading. Each finding follows the format: `[FILE:LINE] (Score: XX) Issue description` with a `Recommendation: ...` line below it. Extract every finding with a score of 95 or higher.
+
+2. **If no critical findings exist** (the "Critical Issues" section is empty or absent, or the summary shows "0 critical"), skip to Step 8. Warnings (80-94) are reported but not auto-fixed.
+
+3. **If critical findings exist**, group them by file path, then invoke one implementer skill per file — all in the **same response** so they run in parallel:
+
+CRITICAL: Skills run with `context: fork` and start at the project root. You MUST include the epic worktree path so the forked agent works in the correct location.
+
+For each file that has critical findings, invoke:
+
+```
+/iloom-swarm-issue-implementer "Your working directory is {{EPIC_WORKTREE_PATH}}. cd there before doing any work.
+
+Fix the following critical code review findings in FILE_PATH. DO NOT create your own issue comment. Do NOT commit changes — just make the edits.
+
+1. [FILE:LINE] (Score: XX) Issue description — Recommendation: ...
+2. [FILE:LINE] (Score: XX) Issue description — Recommendation: ...
+
+Fix ONLY these specific issues. Do not refactor or make additional changes beyond what is listed."
+```
+
+**Parallel invocation:** If critical findings span 3 files, you invoke 3 separate `/iloom-swarm-issue-implementer` skills in a single response message. This runs them concurrently. Each invocation handles findings for ONE file only, preventing edit conflicts.
+
+**Fix prompt construction:** Copy each critical finding EXACTLY from the code reviewer's output (file path, line number, score, issue description, and recommendation). Do NOT paraphrase or summarize — the fix agent needs the precise details to locate and fix each issue.
+
+4. **After ALL fix agents return**, stage and commit once from the epic worktree:
+   ```bash
+   cd "{{EPIC_WORKTREE_PATH}}"
+   git add -A
+   git commit -m "fix(review): address critical wave code review findings"
+   ```
+
+5. Record which findings were sent for fixing for the Step 8 report
+
+### Step 8: Return Structured Report
 
 Return the verification report in this exact format:
 
@@ -220,9 +324,26 @@ Return the verification report in this exact format:
 
 *(If no fix skills were invoked: "None — all must-haves passed on initial verification.")*
 
-### Overall Status: [ALL_PASSED | PARTIALLY_FIXED | FAILURES_REMAIN]
+### Code Review
+
+{{#if HAS_REVIEW_GEMINI}}
+- **Findings**: X critical, Y warnings
+- **Auto-fixed**: N critical issues
+- **Remaining**: Z issues require manual attention
+{{else}}
+{{#if HAS_REVIEW_CODEX}}
+- **Findings**: X critical, Y warnings
+- **Auto-fixed**: N critical issues
+- **Remaining**: Z issues require manual attention
+{{else}}
+- **Status**: Skipped (no review providers configured)
+{{/if}}
+{{/if}}
+
+### Overall Status: [ALL_PASSED | ALL_FIXED | PARTIALLY_FIXED | FAILURES_REMAIN]
 
 - **ALL_PASSED**: All must-haves passed initial verification (no fix skills needed)
+- **ALL_FIXED**: Some must-haves failed initially but ALL were fixed after re-verification
 - **PARTIALLY_FIXED**: Some failures were fixed but others remain
 - **FAILURES_REMAIN**: One or more must-haves are still failing after re-verification
 

package/dist/{build-PKHPE34Y.js → build-ZTGWDHWU.js}

@@ -1,14 +1,14 @@
 #!/usr/bin/env node
 import {
   ScriptCommandBase
-} from "./chunk-CTHB7345.js";
-import "./chunk-ZM2AYHMO.js";
-import "./chunk-VIQOQ463.js";
-import "./chunk-6RINI7OL.js";
+} from "./chunk-PS6K2AOV.js";
+import "./chunk-GQDVH6FA.js";
+import "./chunk-OLJ54WGW.js";
+import "./chunk-PPQ5LV7U.js";
 import "./chunk-K3QGG4O2.js";
-import "./chunk-OPQC4OWM.js";
-import "./chunk-653XBU3L.js";
-import "./chunk-4JZEQBWV.js";
+import "./chunk-KGOBNC5A.js";
+import "./chunk-F5NKWLMQ.js";
+import "./chunk-XIVLGWUX.js";
 import "./chunk-FTYWGQFM.js";
 import "./chunk-VRPPI6GU.js";
 
@@ -24,4 +24,4 @@ var BuildCommand = class extends ScriptCommandBase {
 export {
   BuildCommand
 };
-//# sourceMappingURL=build-PKHPE34Y.js.map
+//# sourceMappingURL=build-ZTGWDHWU.js.map

package/dist/{chunk-GWJWECZB.js → chunk-55NTREIU.js}

@@ -2,18 +2,18 @@
 import {
   PRManager,
   VCSProviderFactory
-} from "./chunk-M3FBM4T3.js";
+} from "./chunk-J5JOJPK3.js";
 import {
   preAcceptClaudeTrust
 } from "./chunk-7UBEHQTP.js";
 import {
   calculatePortForBranch,
   calculatePortFromIdentifier
-} from "./chunk-BFF27W3S.js";
+} from "./chunk-ERMEYFT6.js";
 import {
   installDependencies,
   runScript
-} from "./chunk-VIQOQ463.js";
+} from "./chunk-OLJ54WGW.js";
 import {
   getPackageConfig,
   hasScript
@@ -43,10 +43,10 @@ import {
   fetchOrigin,
   isFileTrackedByGit,
   pushBranchToRemote
-} from "./chunk-OPQC4OWM.js";
+} from "./chunk-KGOBNC5A.js";
 import {
   MetadataManager
-} from "./chunk-4JZEQBWV.js";
+} from "./chunk-XIVLGWUX.js";
 import {
   getLogger
 } from "./chunk-FTYWGQFM.js";
@@ -239,7 +239,7 @@ var LoomManager = class {
    * NEW: Checks for existing worktrees and reuses them if found
    */
   async createIloom(input) {
-    var _a, _b, _c, _d, _e, _f, _g, _h, _i, _j, _k, _l, _m, _n, _o, _p, _q, _r, _s, _t;
+    var _a, _b, _c, _d, _e, _f, _g, _h, _i, _j, _k, _l, _m, _n, _o, _p, _q, _r, _s, _t, _u;
     getLogger().info("Fetching issue data...");
     const issueData = await this.fetchIssueData(input);
     if (input.type === "issue" || input.type === "epic" || input.type === "pr" || input.type === "branch") {
@@ -443,8 +443,8 @@ ${issueData.body}` : "";
     const setArguments = (_l = input.options) == null ? void 0 : _l.setArguments;
     const executablePath = (_m = input.options) == null ? void 0 : _m.executablePath;
     if (enableClaude || enableCode || enableDevServer || enableTerminal) {
-      const { LoomLauncher } = await import("./LoomLauncher-3TSFW7QP.js");
-      const { ClaudeContextManager } = await import("./ClaudeContextManager-IENAE2CP.js");
+      const { LoomLauncher } = await import("./LoomLauncher-V54ENBEF.js");
+      const { ClaudeContextManager } = await import("./ClaudeContextManager-ZH6LEA5I.js");
       const claudeContext = new ClaudeContextManager(void 0, void 0, this.settings);
       const launcher = new LoomLauncher(claudeContext, this.settings);
       await launcher.launchLoom({
@@ -509,10 +509,11 @@ ${issueData.body}` : "";
       capabilities,
       ...draftPrNumber && { draftPrNumber },
       ...((_p = input.options) == null ? void 0 : _p.oneShot) && { oneShot: input.options.oneShot },
-      ...((_q = input.options) == null ? void 0 : _q.complexity) && { complexity: input.options.complexity },
-      ...((_r = input.options) == null ? void 0 : _r.childIssueNumbers) && input.options.childIssueNumbers.length > 0 && { childIssueNumbers: input.options.childIssueNumbers },
-      ...((_s = input.options) == null ? void 0 : _s.childIssues) && input.options.childIssues.length > 0 && { childIssues: input.options.childIssues },
-      ...((_t = input.options) == null ? void 0 : _t.dependencyMap) && Object.keys(input.options.dependencyMap).length > 0 && { dependencyMap: input.options.dependencyMap },
+      ...((_q = input.options) == null ? void 0 : _q.dangerouslySkipPermissions) && { dangerouslySkipPermissions: true },
+      ...((_r = input.options) == null ? void 0 : _r.complexity) && { complexity: input.options.complexity },
+      ...((_s = input.options) == null ? void 0 : _s.childIssueNumbers) && input.options.childIssueNumbers.length > 0 && { childIssueNumbers: input.options.childIssueNumbers },
+      ...((_t = input.options) == null ? void 0 : _t.childIssues) && input.options.childIssues.length > 0 && { childIssues: input.options.childIssues },
+      ...((_u = input.options) == null ? void 0 : _u.dependencyMap) && Object.keys(input.options.dependencyMap).length > 0 && { dependencyMap: input.options.dependencyMap },
       ...input.parentLoom && { parentLoom: input.parentLoom }
     };
     await this.metadataManager.writeMetadata(worktreePath, metadataInput);
@@ -606,7 +607,7 @@ ${issueData.body}` : "";
   async checkAndWarnChildLooms(branchName) {
     let targetBranch = branchName;
     if (!targetBranch) {
-      const { getCurrentBranch } = await import("./git-I3PO6FY7.js");
+      const { getCurrentBranch } = await import("./git-IS7AV3ED.js");
       targetBranch = await getCurrentBranch();
     }
     if (!targetBranch) {
@@ -722,7 +723,7 @@ ${issueData.body}` : "";
       }
     }
     const branchExistedLocally = await branchExists(branchName, process.cwd(), false);
-    if (input.type !== "pr" && branchExistedLocally) {
+    if (input.type === "issue" && branchExistedLocally) {
       throw new Error(
         `Cannot create worktree: branch '${branchName}' already exists. Use 'git branch -D ${branchName}' to delete it first if needed.`
       );
@@ -756,8 +757,8 @@ ${issueData.body}` : "";
       await this.gitWorktree.createWorktree({
         path: worktreePath,
         branch: branchName,
-        createBranch: input.type !== "pr",
-        // PRs use existing branches
+        createBranch: input.type !== "pr" && !branchExistedLocally,
+        // PRs and existing branches use existing branches
         ...baseBranch && { baseBranch }
       });
       if (input.type === "pr" && !branchExistedLocally) {
@@ -1099,7 +1100,7 @@ ${issueData.body}` : "";
    * Ports: handle_existing_worktree() from bash script lines 168-215
    */
   async reuseIloom(worktree, input, issueData) {
-    var _a, _b, _c, _d, _e, _f, _g, _h, _i, _j, _k, _l, _m, _n, _o, _p;
+    var _a, _b, _c, _d, _e, _f, _g, _h, _i, _j, _k, _l, _m, _n, _o, _p, _q;
     const worktreePath = worktree.path;
     const branchName = worktree.branch;
     this.loadMainEnvFile();
@@ -1160,8 +1161,8 @@ ${issueData.body}` : "";
     const executablePath = (_j = input.options) == null ? void 0 : _j.executablePath;
     if (enableClaude || enableCode || enableDevServer || enableTerminal) {
       getLogger().info("Launching workspace components...");
-      const { LoomLauncher } = await import("./LoomLauncher-3TSFW7QP.js");
-      const { ClaudeContextManager } = await import("./ClaudeContextManager-IENAE2CP.js");
+      const { LoomLauncher } = await import("./LoomLauncher-V54ENBEF.js");
+      const { ClaudeContextManager } = await import("./ClaudeContextManager-ZH6LEA5I.js");
       const claudeContext = new ClaudeContextManager(void 0, void 0, this.settings);
       const launcher = new LoomLauncher(claudeContext, this.settings);
       await launcher.launchLoom({
@@ -1223,12 +1224,13 @@ ${issueData.body}` : "";
         prUrls,
         capabilities,
         ...((_m = input.options) == null ? void 0 : _m.oneShot) && { oneShot: input.options.oneShot },
-        ...((_n = input.options) == null ? void 0 : _n.complexity) && { complexity: input.options.complexity },
-        ...((_o = input.options) == null ? void 0 : _o.childIssueNumbers) && input.options.childIssueNumbers.length > 0 && { childIssueNumbers: input.options.childIssueNumbers },
+        ...((_n = input.options) == null ? void 0 : _n.dangerouslySkipPermissions) && { dangerouslySkipPermissions: true },
+        ...((_o = input.options) == null ? void 0 : _o.complexity) && { complexity: input.options.complexity },
+        ...((_p = input.options) == null ? void 0 : _p.childIssueNumbers) && input.options.childIssueNumbers.length > 0 && { childIssueNumbers: input.options.childIssueNumbers },
         ...input.parentLoom && { parentLoom: input.parentLoom }
       };
       await this.metadataManager.writeMetadata(worktreePath, metadataInput);
-    } else if ((_p = input.options) == null ? void 0 : _p.complexity) {
+    } else if ((_q = input.options) == null ? void 0 : _q.complexity) {
       await this.metadataManager.updateMetadata(worktreePath, {
         complexity: input.options.complexity
       });
@@ -1909,4 +1911,4 @@ export {
   CLIIsolationManager,
   DatabaseManager
 };
-//# sourceMappingURL=chunk-GWJWECZB.js.map
+//# sourceMappingURL=chunk-55NTREIU.js.map