@benzotti/jedi 0.1.12 → 0.1.14

package/action/action.yml

@@ -17,7 +17,7 @@ inputs:
   trigger_phrase:
     description: 'The trigger phrase to look for in comments'
     required: false
-    default: '@jedi'
+    default: 'Hey Jedi'
   jedi_version:
     description: 'Version of @benzotti/jedi to install for init'
     required: false

package/action/workflow-template.yml

@@ -9,12 +9,12 @@
 #   - Set CLICKUP_API_TOKEN secret for ClickUp ticket integration
 #   - Run `npx @benzotti/jedi init` locally to customise framework files
 #
-# Usage: Comment on any issue or PR with @jedi followed by a command:
-#   @jedi plan <description>        — Create an implementation plan
-#   @jedi quick <description>       — Make a small, focused change
-#   @jedi review                    — Review the current PR
-#   @jedi feedback                  — Address PR review comments
-#   @jedi do <clickup-ticket-url>   — Full flow: plan + implement from ticket
+# Usage: Comment on any issue or PR with "Hey Jedi" followed by a command:
+#   Hey Jedi plan <description>        — Create an implementation plan
+#   Hey Jedi quick <description>       — Make a small, focused change
+#   Hey Jedi review                    — Review the current PR
+#   Hey Jedi feedback                  — Address PR review comments
+#   Hey Jedi do <clickup-ticket-url>   — Full flow: plan + implement from ticket
 #
 # Conversation: Jedi supports back-and-forth iteration. After Jedi responds,
 # reply with feedback to refine, or say "approved" to finalise.
@@ -39,11 +39,11 @@ permissions:
   id-token: write
 
 jobs:
-  # ── Triggered by @jedi comments on issues/PRs ──
+  # ── Triggered by "Hey Jedi" comments on issues/PRs ──
   jedi:
     if: >-
       (github.event_name == 'issue_comment' || github.event_name == 'pull_request_review_comment')
-      && contains(github.event.comment.body, '@jedi')
+      && contains(github.event.comment.body, 'Hey Jedi')
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v4
@@ -94,7 +94,7 @@ jobs:
         uses: anthropics/claude-code-action@v1
         with:
           anthropic_api_key: ${{ secrets.ANTHROPIC_API_KEY }}
-          trigger_phrase: '@jedi'
+          trigger_phrase: 'Hey Jedi'
           bot_name: 'jedi[bot]'
         env:
           CLICKUP_API_TOKEN: ${{ secrets.CLICKUP_API_TOKEN }}

package/dist/index.js

@@ -9250,7 +9250,7 @@ async function copyFrameworkFiles(cwd, projectType, force, ci = false) {
   const frameworkDest = join2(cwd, ".jdi", "framework");
   const glob = new Bun.Glob("**/*");
   for await (const file of glob.scan({ cwd: frameworkDir })) {
-    if (file.startsWith("adapters/") || file.startsWith("commands/"))
+    if (file.startsWith("adapters/"))
       continue;
     const src2 = join2(frameworkDir, file);
     const dest = join2(frameworkDest, file);
@@ -9366,6 +9366,8 @@ Recognise natural language JDI intents and invoke the matching skill via the Ski
 
 Extract flags from context: "in a worktree" \u2192 \`--worktree\`, "lightweight" \u2192 \`--worktree-lightweight\`, "single agent" \u2192 \`--single\`, "use teams" \u2192 \`--team\`. If the intent is unclear, ask. Never guess.
 
+Planning and implementation are separate gates \u2014 NEVER auto-proceed to implementation after plan approval.
+
 ## Iterative Refinement
 
 After \`/jdi:create-plan\` or \`/jdi:implement-plan\` completes, the conversation continues naturally \u2014 no new command invocation needed. When the user provides feedback (e.g. "change task 2", "move this to a helper", "add error handling"), apply the changes directly, update state, and present the updated summary. When the user approves (e.g. "approved", "looks good", "lgtm"), finalise the review state. The conversation IS the feedback loop.
@@ -9388,6 +9390,8 @@ Recognise natural language JDI intents and invoke the matching skill via the Ski
 
 Extract flags from context: "in a worktree" \u2192 \`--worktree\`, "lightweight" \u2192 \`--worktree-lightweight\`, "single agent" \u2192 \`--single\`, "use teams" \u2192 \`--team\`. If the intent is unclear, ask. Never guess.
 
+Planning and implementation are separate gates \u2014 NEVER auto-proceed to implementation after plan approval.
+
 ## Iterative Refinement
 
 After \`/jdi:create-plan\` or \`/jdi:implement-plan\` completes, the conversation continues naturally \u2014 no new command invocation needed. When the user provides feedback (e.g. "change task 2", "move this to a helper", "add error handling"), apply the changes directly, update state, and present the updated summary. When the user approves (e.g. "approved", "looks good", "lgtm"), finalise the review state. The conversation IS the feedback loop.
@@ -11093,7 +11097,7 @@ function buildConversationContext(thread, currentCommentId) {
   for (const comment of thread) {
     if (comment.id === currentCommentId)
       break;
-    if (comment.body.includes("@jedi")) {
+    if (/hey\s+jedi/i.test(comment.body)) {
       inJediConversation = true;
       jediSegments.push(comment);
     } else if (inJediConversation) {
@@ -11143,7 +11147,7 @@ function formatResultComment(command, success, summary) {
 
 // src/commands/action.ts
 function parseComment(comment, isFollowUp) {
-  const match = comment.match(/@jedi\s+(.+)/is);
+  const match = comment.match(/hey\s+jedi\s+(.+)/is);
   if (!match) {
     if (isFollowUp) {
       return {
@@ -11205,12 +11209,12 @@ function parseComment(comment, isFollowUp) {
 var actionCommand = defineCommand({
   meta: {
     name: "action",
-    description: "GitHub Action entry point \u2014 parse @jedi comment and run workflow"
+    description: "GitHub Action entry point \u2014 parse 'Hey Jedi' comment and run workflow"
   },
   args: {
     comment: {
       type: "positional",
-      description: "The raw comment body containing @jedi mention",
+      description: "The raw comment body containing 'Hey Jedi' mention",
       required: true
     },
     "comment-id": {
@@ -11248,7 +11252,7 @@ var actionCommand = defineCommand({
     }
     const intent = parseComment(args.comment, isFollowUp);
     if (!intent) {
-      consola.error("Could not parse @jedi intent from comment");
+      consola.error("Could not parse 'Hey Jedi' intent from comment");
       process.exit(1);
     }
     consola.info(`Parsed intent: ${intent.isFeedback ? "feedback on previous" : intent.command}${intent.fullFlow ? " (full flow)" : ""}`);
@@ -11487,7 +11491,7 @@ var setupActionCommand = defineCommand({
       "Jedi GitHub Action Setup",
       "",
       "Uses: anthropics/claude-code-action@v1",
-      "Trigger: @jedi in issue/PR comments",
+      "Trigger: 'Hey Jedi' in issue/PR comments",
       "",
       "Required secrets (set via GitHub UI or CLI):",
       "",
@@ -11499,11 +11503,11 @@ var setupActionCommand = defineCommand({
       "",
       "Usage: Comment on any issue or PR with:",
       "",
-      "  @jedi plan <description>",
-      "  @jedi quick <small fix>",
-      "  @jedi do <clickup-ticket-url>",
-      "  @jedi review",
-      "  @jedi feedback",
+      "  Hey Jedi plan <description>",
+      "  Hey Jedi quick <small fix>",
+      "  Hey Jedi do <clickup-ticket-url>",
+      "  Hey Jedi review",
+      "  Hey Jedi feedback",
       "",
       "Conversation: Reply to Jedi with feedback to iterate,",
       "or say 'approved' to finalise."
@@ -11514,7 +11518,7 @@ var setupActionCommand = defineCommand({
 // package.json
 var package_default = {
   name: "@benzotti/jedi",
-  version: "0.1.12",
+  version: "0.1.14",
   description: "JDI - Context-efficient AI development framework for Claude Code",
   type: "module",
   bin: {

package/framework/commands/create-plan.md

@@ -27,6 +27,6 @@ Create an implementation plan using a single planner agent (includes research).
 9. **Present summary** (name, objective, task table, files) then ask: _"Provide feedback to refine, or say **approved** to finalise."_
 10. **Review loop**: approval → update state to `"approved"`, confirm. Feedback → revise plan in-place, increment revision, re-present summary. Repeat until approved. This is natural conversation — no separate command needed.
 
-Agent base (read FIRST for cache): ./components/meta/AgentBase.md | Agent spec: ./agents/jdi-planner.md
+Agent base (read FIRST for cache): .jdi/framework/components/meta/AgentBase.md | Agent spec: .jdi/framework/agents/jdi-planner.md
 
 Feature to plan: $ARGUMENTS

package/framework/commands/implement-plan.md

@@ -26,8 +26,8 @@ Execute a PLAN.md with complexity-based routing.
 11. **Present summary** (tasks completed, files changed, verification results, deviations) then ask: _"Provide feedback to adjust, or say **approved** to finalise."_
 12. **Review loop**: approval → update state to `"complete"`, suggest commit/PR. Feedback → apply code changes, run tests, increment revision, re-present. Repeat until approved. Natural conversation — no separate command needed.
 
-Agent base (read FIRST for cache): ./components/meta/AgentBase.md | Agent specs: ./agents/jdi-backend.md, ./agents/jdi-frontend.md
-Orchestration: ./components/meta/AgentTeamsOrchestration.md | Routing: ./components/meta/ComplexityRouter.md
+Agent base (read FIRST for cache): .jdi/framework/components/meta/AgentBase.md | Agent specs: .jdi/framework/agents/jdi-backend.md, .jdi/framework/agents/jdi-frontend.md
+Orchestration: .jdi/framework/components/meta/AgentTeamsOrchestration.md | Routing: .jdi/framework/components/meta/ComplexityRouter.md
 
 When spawning agents, detect project type and include a `## Project Context` block (type, tech stack, quality gates, working directory) in the spawn prompt. This saves agents 2-3 discovery tool calls.
 

package/framework/commands/worktree-remove.md

@@ -27,6 +27,6 @@ Remove a git worktree and clean up all associated resources.
 7. **Update state**: set `worktree.active: false`, clear `worktree.path`, `worktree.branch` in `.jdi/config/state.yaml`
 8. **Report**: what was removed
 
-Reference: ./hooks/jdi-worktree-cleanup.md
+Reference: .jdi/framework/hooks/jdi-worktree-cleanup.md
 
 Worktree to remove: $ARGUMENTS

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@benzotti/jedi",
-  "version": "0.1.12",
+  "version": "0.1.14",
   "description": "JDI - Context-efficient AI development framework for Claude Code",
   "type": "module",
   "bin": {