ima-claude 2.14.1 → 2.15.0

package/README.md

@@ -94,7 +94,7 @@ See [platforms/shared/types.ts](platforms/shared/types.ts) for the interface con
 ## What's Included
 
 - **Multi-Platform Installer**: Interactive CLI with auto-detection, install preview, and per-item exclusion — supports Claude Code, Junie CLI, Gemini CLI, and GitHub Copilot
-- **48 Skills**: Foundational + FP implementation + domain expert + integration + meta-skills
+- **49 Skills**: Foundational + FP implementation + domain expert + integration + meta-skills
 - **6 Named Agents**: Explorer (haiku), Implementer (sonnet), Reviewer (sonnet), Tester (sonnet), WP Developer (sonnet), Memory (sonnet) — enforced constraints
 - **23 Hooks**: Automatic behavioral enforcement (security, memory, workflow, Serena, code quality) — translated to guidelines for platforms without hook support
 - **Default Persona**: "The Practitioner" - 25-year veteran mindset, collaborative, plan-first
@@ -302,6 +302,7 @@ Agents are auto-discovered from `plugins/ima-claude/agents/`. No manifest change
 | `ima-forms-expert` | WordPress form components (IMA Forms) |
 | `discourse-admin` | Discourse admin API (site settings, config export/import, groups) |
 | `ima-cancer-care-guides` | Cancer care guide document pipeline (DOCX → markdown → HTML → PDF, Canva mapping) |
+| `prompt-starter` | Zero-friction prompt templates (quick, brainstorm, plan-implement) with Jira pre-fill and editor spawn |
 
 ### Integration Skills
 

package/dist/cli.js

@@ -11,7 +11,7 @@ var HOOKS_DIR = join(CLAUDE_DIR, "hooks");
 var COMMANDS_DIR = join(CLAUDE_DIR, "commands");
 var RULES_DIR = join(CLAUDE_DIR, "rules");
 var SETTINGS_FILE = join(CLAUDE_DIR, "settings.json");
-var VERSION = "2.14.1";
+var VERSION = "2.15.0";
 var colors = {
   reset: "\x1B[0m",
   bright: "\x1B[1m",
@@ -117,6 +117,8 @@ var SKILLS_TO_INSTALL = [
   "mcp-memory",
   "mcp-vestige",
   "mcp-qdrant",
+  // Prompt workflow skills
+  "prompt-starter",
   // Quick reference
   "quickstart",
   "scorecard",

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "ima-claude",
-  "version": "2.14.1",
+  "version": "2.15.0",
   "description": "IMA's AI coding agent skills - FP patterns, architecture guidance, and team standards. Supports Claude Code, Junie CLI, Gemini CLI, and GitHub Copilot.",
   "type": "module",
   "scripts": {

package/plugins/ima-claude/.claude-plugin/plugin.json

@@ -1,7 +1,7 @@
 {
   "name": "ima-claude",
-  "version": "2.14.1",
-  "description": "IMA's Claude Code skills for functional programming, architecture, and team standards. 52 skills, 24 hooks, default persona, 3-tier memory system.",
+  "version": "2.15.0",
+  "description": "IMA's Claude Code skills for functional programming, architecture, and team standards. 53 skills, 24 hooks, default persona, 3-tier memory system.",
   "author": {
     "name": "IMA",
     "url": "https://github.com/Soabirw/ima-claude"

package/plugins/ima-claude/skills/prompt-starter/SKILL.md

@@ -0,0 +1,128 @@
+---
+name: prompt-starter
+description: Zero-friction prompt templates (quick, brainstorm, plan-implement). Selects template, pre-fills from Jira, opens in GUI editor, reads back on close.
+---
+
+# Prompt Starter
+
+Load a structured prompt template, pre-fill it with Jira context, open it in the user's GUI editor, and execute the result when they close the file.
+
+**Trigger words:**
+- `brainstorm`, `research`, `explore` → `references/brainstorm.md`
+- `plan`, `implement`, `build`, `execute` → `references/plan-implement.md`
+- `quick`, or a short one-liner task → `references/quick.md`
+- Just a Jira key with no clear workflow → ask the user which template
+
+## Relationship to prompt_coach.py
+
+- **prompt-starter** = "What should my prompt say?" (template + context + editor)
+- **prompt_coach.py** = "Is my prompt good enough?" (quality evaluation)
+
+Different lanes, complementary. The coach may fire on the final prompt — that's fine.
+
+---
+
+## Template Selection
+
+Match the user's trigger words to select a template. Examples:
+
+| User says | Template |
+|---|---|
+| "brainstorm FNR-1234" | brainstorm.md |
+| "research auth options" | brainstorm.md |
+| "plan and implement FNR-567" | plan-implement.md |
+| "build the PDF export" | plan-implement.md |
+| "quick task: add validation" | quick.md |
+| "FNR-1234" (ambiguous) | Ask which workflow |
+
+---
+
+## Flow
+
+### Step 1: Select template
+
+Read the appropriate template from `references/`.
+
+### Step 2: Fetch Jira context (if Jira key present)
+
+Use mcp-atlassian to fetch the issue. Extract:
+- **Summary** → fills the one-line goal / user story
+- **Description** → fills the Problem section
+- **Acceptance Criteria** (from description or subtasks) → fills Acceptance / Test sections
+
+Map Jira fields to template `[bracket]` placeholders naturally — don't create a rigid substitution engine. Use judgment: if the Jira description is rich, use it; if sparse, leave the bracket hint for the user to fill in.
+
+### Step 3: Check prior work (plan-implement only)
+
+For plan-implement templates, search Serena memory for `{feature-name}-brainstorm`. If found, pre-fill the **Prior Work** section with a reference to it and incorporate key decisions into the **Plan** section.
+
+### Step 4: Write the pre-filled template
+
+Create the prompt directory if needed:
+```bash
+mkdir -p ~/.claude/prompts
+```
+
+Write the pre-filled template to `~/.claude/prompts/{session-name}.md` where session-name is descriptive:
+- `brainstorm-pdf-export.md`
+- `quick-email-validation.md`
+- `plan-fnr-1234.md`
+
+### Step 5: Open in editor (or present inline)
+
+**Quick template exception:** Quick templates are short (12 lines). Present them inline in the conversation — no editor spawn. Skip to Step 7.
+
+**For brainstorm and plan-implement templates**, spawn the user's GUI editor.
+
+#### Editor Resolution
+
+Claude Code owns the terminal, so terminal editors (nano, vim) cannot be used — the user can't interact with them. Only GUI editors work.
+
+Resolution order:
+1. `$VISUAL` — the Unix convention for GUI editors
+2. `$EDITOR` — but **only** if it's a known GUI editor: `zed`, `code`, `subl`, `gedit`, `kate`
+3. Auto-detect: check `which zed`, then `which code`, then `which subl` — use the first found
+4. **Fallback:** no suitable editor found → present inline and modify via conversation
+
+#### Spawning the editor
+
+Use Bash with `run_in_background: true`:
+```bash
+zed --wait ~/.claude/prompts/{session-name}.md
+```
+
+All GUI editors use `--wait` so the process blocks until the user closes the file/tab.
+
+Tell the user:
+> Template is open in {editor}. Edit and close the tab when done — I'll pick up from there.
+
+### Step 6: Read the result
+
+When the background task completes (editor closed), read the file back:
+```
+Read ~/.claude/prompts/{session-name}.md
+```
+
+### Step 7: Confirm and execute
+
+Present the final prompt to the user. Ask for confirmation:
+> Here's your prompt. Ready to execute, or want to adjust anything?
+
+On confirmation, treat the prompt content as working instructions and execute accordingly.
+
+---
+
+## Editor Notes for Team Onboarding
+
+Team members should set `$VISUAL` in their shell profile for the best experience:
+
+```bash
+# Zed
+echo 'export VISUAL="zed --wait"' >> ~/.bashrc
+
+# VS Code
+echo 'export VISUAL="code --wait"' >> ~/.bashrc
+
+# Sublime Text
+echo 'export VISUAL="subl --wait"' >> ~/.bashrc
+```

package/plugins/ima-claude/skills/prompt-starter/references/brainstorm.md

@@ -0,0 +1,25 @@
+[One-line goal as a user story]
+
+## Problem
+[What's broken or missing. Be specific about the symptom.]
+
+## Context
+[Repo, branch, related systems, infrastructure details]
+
+## Research
+- [Specific things to investigate, with scoped locations]
+- [Reference patterns to follow: "like X" or "see Y"]
+
+## Q&A
+After completing research, interview me using the AskUserQuestion tool.
+Dig into edge cases, tradeoffs, and constraints — skip obvious questions.
+
+- [Seed questions/concerns you already have]
+
+## Document
+Save the outcome to Serena memory as `{feature-name}-brainstorm`.
+Include:
+- Original user story and problem statement
+- Key findings from research
+- Decisions and tradeoffs from Q&A
+- Recommended approach (not a full plan — just direction)

package/plugins/ima-claude/skills/prompt-starter/references/plan-implement.md

@@ -0,0 +1,35 @@
+[One-line goal as a user story]
+Save this prompt and plan to Serena memory as `{feature-name}-plan`.
+
+## Problem
+[What's broken or missing. Be specific about the symptom.]
+
+## Prior Work
+[Reference brainstorming: "See Serena memory `{feature-name}-brainstorm`"]
+
+## Context
+[Repo, branch, infrastructure — if not already in brainstorming]
+
+## Plan
+- [Approach, informed by brainstorming outcomes]
+- [Open implementation questions with `?`]
+- [Constraints: "prefer X unless evidence shows Y"]
+
+## Implement
+[Numbered deliverables in priority order]
+
+For each story: Implement → Test → Review, then repeat for the full picture.
+
+## Test
+- [How to verify EACH deliverable]
+
+## Review
+- [Which review tool/agent and what to check for]
+
+## Document
+- [Explicit list of files to update]
+
+## Agents (recommended for multi-story work)
+- ima-claude:implementer for [phase] with skills: [...]
+- ima-claude:tester with /unit-testing
+- ima-claude:reviewer after implementation

package/plugins/ima-claude/skills/prompt-starter/references/quick.md

@@ -0,0 +1,11 @@
+[One-line goal — what and where]
+
+## Context
+[Repo/branch/file if not obvious from working directory]
+
+## Constraints
+- [Pattern to follow: "like X" or "match Y"]
+- [What NOT to do, if relevant]
+
+## Acceptance
+- [How to verify it works]