@sixsevenai/ai-dlc-installer 1.4.4 → 1.5.0

package/library/agents/ai-dlc-navigator.md

@@ -194,7 +194,7 @@ When users express these intents, recommend activating these skills:
 
 | User Says | Skill to Recommend |
 |-----------|-------------------|
-| "How do I manage sessions?" | `session-lifecycle` |
+| "How do I manage sessions?" | `ai-dlc-knowledge` |
 | "How do mob elaboration?" | `mob-elaboration` |
 | "How to model domains?" | `ddd-modeling` |
 | "How to write tests?" | `test-suite-builder` |

package/library/agents/ai-dlc-trainer.md

@@ -18,7 +18,7 @@ color: cyan
 **NEVER output internal CLI commands to users.**
 
 When presenting next steps to the user, ALWAYS use slash commands:
-- `/talk:aidlc`
+- `/ai-dlc:talk`
 - `/ai-dlc:status`
 - `/ai-dlc:start-intent`
 
@@ -67,7 +67,7 @@ When spawned by orchestrating commands, you receive enforced context:
 ## Activation Triggers
 
 Use this agent proactively when:
-1. User invokes `/talk:aidlc` command
+1. User invokes `/ai-dlc:talk` command
 2. Questions about AI-DLC methodology
 3. Need explanation of phases, artifacts, or workflows
 4. Training new team members on AI-DLC
@@ -153,7 +153,7 @@ Help teams implement AI-DLC:
 After creating training artifacts, the orchestrator verifies:
 
 ```bash
-ai-dlc verify talk:aidlc "artifacts/training/training-session-20251219-001.md"
+ai-dlc verify ai-dlc:talk "artifacts/training/training-session-20251219-001.md"
 ```
 
 **Success Response:**
@@ -240,8 +240,8 @@ As Dr. Chen, respond to:
 
 | Purpose | CLI Command |
 |---------|-------------|
-| Prepare Training | `ai-dlc prepare talk:aidlc --module <module>` |
-| Verify Session | `ai-dlc verify talk:aidlc <path>` |
+| Prepare Training | `ai-dlc prepare ai-dlc:talk --module <module>` |
+| Verify Session | `ai-dlc verify ai-dlc:talk <path>` |
 | List Modules | `ai-dlc training list` |
 | Progress | `ai-dlc training progress` |
 

package/library/commands/ai-dlc/cheatsheet.md

@@ -0,0 +1,63 @@
+---
+command: ai-dlc:cheatsheet
+description: Display a quick-reference card of all AI-DLC slash commands organized by phase
+phase: cross-cutting
+allowed-tools: Read
+---
+
+# AI-DLC Cheatsheet
+
+Output the following quick-reference card directly to the user. Do not add commentary — just present the card as-is in a single, readable block.
+
+---
+
+```
+AI-DLC Quick Reference
+======================
+
+INCEPTION FLOW
+  /ai-dlc:start-intent       Define business intent (WHAT and WHY)
+  /ai-dlc:create-units        Decompose into atomic Units
+  /ai-dlc:generate-stories    Create INVEST user stories
+  /ai-dlc:elicit-nfrs         Define non-functional requirements
+  /ai-dlc:assess-risks        Identify risks and mitigations
+  /ai-dlc:plan-bolt           Estimate work in 1-3 day Bolts
+  /ai-dlc:validate-phase inception   Verify phase completeness
+
+CONSTRUCTION FLOW
+  /ai-dlc:domain-model        DDD domain modeling
+  /ai-dlc:logical-design      Architecture and ADRs
+  /ai-dlc:design-api          OpenAPI contract design
+  /ai-dlc:generate-code       Production code generation
+  /ai-dlc:build-tests         Test suite generation
+  /ai-dlc:review-code         7-dimension code review
+  /ai-dlc:validate-phase construction   Verify phase completeness
+
+OPERATIONS FLOW
+  /ai-dlc:package-unit        Create deployment artifacts
+  /ai-dlc:deploy              Deploy to environment
+  /ai-dlc:setup-monitoring    Configure observability
+  /ai-dlc:create-runbook      Incident response procedures
+  /ai-dlc:validate-phase operations   Verify phase completeness
+
+UTILITY COMMANDS
+  /ai-dlc:status              Show current phase and progress
+  /ai-dlc:next-phase          Transition to next phase
+  /ai-dlc:archive-session     Archive completed work
+  /ai-dlc:talk                Interactive training with AI-DLC expert
+  /ai-dlc:quickstart          Guided onboarding (~5 min)
+
+COMPANION: SPECKIT (spec-driven development)
+  /spec:specify               Create feature specification
+  /spec:clarify               Fill spec gaps with questions
+  /spec:plan                  Generate implementation plan
+  /spec:tasks                 Generate dependency-ordered tasks
+  /spec:implement             Execute tasks from tasks.md
+  /spec:to-intent             Bridge spec to AI-DLC intent
+
+COMPANION: TASK (lightweight task planning)
+  /task:plan                  Decompose prompt into task plan
+  /task:execute               Execute plan step-by-step
+  /task:clarify               Refine plan with questions
+  /task:complete              Finalize with Complete.md + commit
+```

package/library/commands/ai-dlc/quickstart.md

@@ -0,0 +1,150 @@
+---
+command: ai-dlc:quickstart
+description: Interactive 5-minute onboarding walkthrough for new AI-DLC users — covers core concepts, runs your first Intent, and shows next steps
+argument-hint: ""
+phase: cross-cutting
+allowed-tools: Task, Read, AskUserQuestion
+args: []
+---
+
+# AI-DLC Quickstart
+
+A guided, conversational onboarding flow for new team members. Covers the three phases, core concepts, and walks you through creating your first Intent.
+
+## Command: /ai-dlc:quickstart
+
+## Context Isolation: Required
+
+**MANDATORY**: Use the Task tool to spawn the ai-dlc-trainer subagent. Do NOT execute the walkthrough in this context.
+
+Use Task tool with these parameters:
+- **subagent_type**: `ai-dlc-trainer`
+- **prompt**: See below
+
+## Subagent Prompt
+
+```
+You are running the AI-DLC Quickstart walkthrough for a new user. Keep everything concise and actionable — aim for ~5 minutes total.
+
+Follow these steps IN ORDER. After each step, pause and let the user respond before continuing.
+
+---
+
+STEP 1: WELCOME AND OVERVIEW
+
+Print exactly:
+
+## Welcome to AI-DLC
+
+**AI-DLC** (AI-Driven Lifecycle) replaces the traditional SDLC with three phases:
+
+| Phase | Purpose | Key Output |
+|-------|---------|------------|
+| **Inception** | Define WHAT and WHY | Intent, Units, Stories, NFRs, Risks, Bolt Plan |
+| **Construction** | Build production software | Domain Models, Code, Tests, Reviews |
+| **Operations** | Deploy, monitor, maintain | Deployments, Monitoring, Runbooks |
+
+**Two concepts to know:**
+- **Unit** — An atomic work item estimated at 1-3 Bolts
+- **Bolt** — A 1-3 day timebox that delivers a tangible artifact
+
+Every project starts with an **Intent** — a plain-language statement of your business goal.
+
+Then ask: "Ready to create your first Intent? (yes / I have questions first)"
+
+If the user has questions, answer them briefly, then proceed.
+
+---
+
+STEP 2: CREATE YOUR FIRST INTENT
+
+Say:
+
+Let's start Inception. Run this command now:
+
+```
+/ai-dlc:start-intent "Build a task management app for small teams"
+```
+
+Or replace the text in quotes with your own idea. I'll wait while you run it.
+
+Wait for the user to confirm they ran it.
+
+---
+
+STEP 3: CHECK YOUR PROGRESS
+
+Say:
+
+Now let's see what AI-DLC recorded. Run:
+
+```
+/ai-dlc:status
+```
+
+This shows your current phase, active session, and completed artifacts. You should see your new Intent listed under Inception.
+
+Wait for the user to confirm.
+
+---
+
+STEP 4: EXPLAIN NEXT STEPS
+
+Print exactly:
+
+## What Comes Next
+
+With your Intent created, here is the Inception sequence:
+
+1. `/ai-dlc:create-units` — Break your Intent into Units
+2. `/ai-dlc:generate-stories` — Create user stories for each Unit
+3. `/ai-dlc:elicit-nfrs` — Capture non-functional requirements
+4. `/ai-dlc:assess-risks project` — Identify risks and mitigations
+5. `/ai-dlc:plan-bolt` — Estimate each Unit in Bolts
+6. `/ai-dlc:validate-phase inception` — Confirm Inception is complete
+
+Once Inception passes validation, run `/ai-dlc:next-phase` to move into Construction.
+
+**Alternative:** Use `/ai-dlc:mob-elaborate` for a single collaborative session that produces Units, NFRs, and Risks together.
+
+---
+
+STEP 5: POINT TO DEEPER RESOURCES
+
+Say:
+
+## Resources
+
+- **Getting Started Guide**: Read `.claude/docs/GETTING-STARTED-GUIDE.md` for a full walkthrough
+- **Ask the Expert**: Run `/ai-dlc:talk` to chat with Dr. Alex Chen about methodology
+- **Agent Directory**: Run `/agent-directory` to see all 58 specialized agents
+- **Status Anytime**: Run `/ai-dlc:status` to see where you are
+
+You're all set. Run `/ai-dlc:create-units` to continue, or ask me anything.
+
+---
+
+End the session after Step 5 unless the user asks follow-up questions.
+```
+
+## Usage Examples
+
+```bash
+/ai-dlc:quickstart
+```
+
+## What You'll Learn
+
+- **Core Concepts**: Phases, Units, and Bolts
+- **First Command**: Create an Intent hands-on
+- **Status Check**: See your progress in AI-DLC
+- **Next Steps**: The full Inception command sequence
+- **Resources**: Where to learn more
+
+## Prerequisites
+
+None — designed for first-time users.
+
+## Duration
+
+Approximately 5 minutes.

package/library/commands/ai-dlc/setup.md

@@ -434,7 +434,7 @@ This command provides one-command project initialization by:
    info("  AI-DLC Commands: See README.md");
    info("  Full Documentation: .ai-dlc/README.md");
    info("  Quick Reference: .ai-dlc/QUICK_REFERENCE.md");
-   info("  Interactive Training: /talk:aidlc");
+   info("  Interactive Training: /ai-dlc:talk");
    ```
 
 ## Usage Examples

package/library/commands/im-lost.md

@@ -0,0 +1,245 @@
+---
+command: im-lost
+description: Search previous conversation sessions and memory files across Claude project folders to recover lost context across context windows
+argument-hint: "<search-terms> [--deep] [--memory-only] [--sessions-only]"
+allowed-tools: Read, Glob, Grep, Bash
+args:
+  - name: search-terms
+    description: "Keywords, phrases, or topics to search for across sessions and memory"
+    required: true
+  - name: --deep
+    description: "Also search inside .jsonl conversation transcripts (slower, more thorough)"
+    required: false
+  - name: --memory-only
+    description: "Only search memory files, skip session index"
+    required: false
+  - name: --sessions-only
+    description: "Only search session index, skip memory files"
+    required: false
+---
+
+# /im-lost
+
+Search previous conversation sessions and memory files to recover context lost across context windows. Searches the local project folder first, then discovers and searches sibling project folders that share the same project name suffix.
+
+## User Input
+
+```text
+$ARGUMENTS
+```
+
+<rules enforcement="strict">
+<rule id="1" severity="blocking">Parse $ARGUMENTS before any search. If empty, ask the user what they are looking for -- do not search with no terms.</rule>
+<rule id="2" severity="required">Always search the primary project folder first, then related folders. Present results grouped by source.</rule>
+<rule id="3" severity="required">Never modify any file. This command is read-only.</rule>
+<rule id="4" severity="warning">Limit .jsonl deep search to the 20 most recent sessions to avoid excessive context consumption.</rule>
+</rules>
+
+<step id="1" name="Parse Input and Detect Flags">
+<action>
+- Extract search terms from `$ARGUMENTS` (everything that is not a flag).
+- Detect flags: `--deep`, `--memory-only`, `--sessions-only`.
+- Store the search terms as `SEARCH_TERMS`.
+</action>
+<verify>
+- `SEARCH_TERMS` is non-empty after flag extraction.
+</verify>
+<on-fail>
+Ask the user: "What are you looking for? Provide keywords, a topic, or a phrase from a previous conversation."
+Stop execution.
+</on-fail>
+</step>
+
+<step id="2" name="Identify Project Folders">
+<action>
+- Determine the **primary project folder** path. This is the `.claude/projects/` subfolder for the current working directory. On Windows, the folder name is the path with `\` and `:` replaced by `-` (e.g., `C--dev-GitHub-SixSevenAI`).
+- Extract the **project name suffix** from the primary folder name. The suffix is the last segment after the final `-` that is NOT a single letter or drive prefix. For this project, the suffix is `SixSevenAI`.
+  - Implementation: split the folder name by `--` to get path segments, then take the last segment. If the last segment contains `-`, treat the full last `--`-delimited segment as the suffix (e.g., `SixSevenAI-src-installer` from `C--dev-GitHub-SixSevenAI-src-installer`).
+- List all directories in `~/.claude/projects/` and find **related folders** -- any folder whose name contains the suffix (case-insensitive), excluding the primary folder itself.
+- Run this command to discover folders:
+  ```bash
+  ls -d ~/.claude/projects/*/ | grep -i "{suffix}"
+  ```
+- Store the primary folder path and list of related folder paths.
+- Report to the user:
+  > Searching primary: `{primary-folder}`
+  > Related folders found: `{list}` (or "none")
+</action>
+<verify>
+- Primary project folder exists and is accessible.
+</verify>
+<on-fail>
+**ERROR** -- "Cannot locate project folder at ~/.claude/projects/. Verify Claude Code has been used in this project directory."
+Stop execution.
+</on-fail>
+</step>
+
+<step id="3" name="Search Session Index">
+<action>
+Skip this step if `--memory-only` flag is set.
+
+For each project folder (primary first, then related):
+1. Read `sessions-index.json` if it exists.
+2. Search these fields for `SEARCH_TERMS` (case-insensitive):
+   - `firstPrompt` -- the opening message of each session
+   - `summary` -- Claude's summary of the session
+   - `gitBranch` -- the branch name
+3. Collect matching entries with: `sessionId`, `summary`, `firstPrompt`, `gitBranch`, `created`, `modified`, `messageCount`.
+4. Sort matches by `modified` date (most recent first).
+5. Cap results at **15 matches per folder**.
+</action>
+<verify>
+- Sessions-index was read successfully (or does not exist, which is acceptable).
+</verify>
+<on-fail>
+Warn: "Could not read sessions-index.json in {folder}. Skipping session search for this folder."
+Continue to next folder or next step.
+</on-fail>
+</step>
+
+<step id="4" name="Search Memory Files">
+<action>
+Skip this step if `--sessions-only` flag is set.
+
+For each project folder (primary first, then related):
+1. Read `memory/MEMORY.md` if it exists -- search its content for `SEARCH_TERMS`.
+2. Use Grep to search all `.md` files in `memory/` for `SEARCH_TERMS`:
+   ```
+   Grep pattern: {SEARCH_TERMS} path: {folder}/memory/ glob: "*.md"
+   ```
+3. For each match, read the file's frontmatter to get `name`, `description`, and `type`.
+4. Collect: file name, memory type, description, and matching line(s).
+</action>
+<verify>
+- Memory folder was searched (or does not exist, which is acceptable).
+</verify>
+<on-fail>
+Warn: "No memory folder found in {folder}. Skipping memory search for this folder."
+Continue to next folder or next step.
+</on-fail>
+</step>
+
+<step id="5" name="Deep Search Conversation Transcripts" condition="--deep flag is present">
+<action>
+For each project folder (primary first, then related):
+1. Get the 20 most recently modified `.jsonl` files:
+   ```bash
+   ls -t {folder}/*.jsonl 2>/dev/null | head -20
+   ```
+2. For each `.jsonl` file, use Grep to search for `SEARCH_TERMS`:
+   ```
+   Grep pattern: {SEARCH_TERMS} path: {jsonl-file}
+   ```
+3. For matching files, cross-reference with `sessions-index.json` to get the session summary and date.
+4. Collect: session ID, summary, date, and a short snippet of the matching line (truncated to 200 chars).
+5. Cap results at **10 matches per folder**.
+
+**Performance note**: `.jsonl` files can be large. Search only the raw text content, do not parse JSON structure.
+</action>
+<verify>
+- At least one `.jsonl` file was searchable (or none exist).
+</verify>
+<on-fail>
+Warn: "Could not search conversation transcripts in {folder}."
+Continue to next step.
+</on-fail>
+</step>
+
+<step id="6" name="Present Results">
+<action>
+Format and present all results grouped by source, most relevant first.
+
+**If matches found**, use this format:
+
+```
+## Search Results for: "{SEARCH_TERMS}"
+
+### Primary Project: {primary-folder-name}
+
+#### Session Matches
+| # | Date | Branch | Summary | First Prompt |
+|---|------|--------|---------|--------------|
+| 1 | {date} | {branch} | {summary} | {first-prompt-truncated} |
+
+#### Memory Matches
+| File | Type | Description | Match |
+|------|------|-------------|-------|
+| {file} | {type} | {description} | {matching-line} |
+
+#### Transcript Matches (deep search)
+| # | Date | Summary | Snippet |
+|---|------|---------|---------|
+| 1 | {date} | {summary} | {snippet} |
+
+### Related: {related-folder-name}
+{same structure}
+```
+
+**If no matches found**:
+
+> No results found for "{SEARCH_TERMS}" across {N} project folder(s).
+>
+> Suggestions:
+> - Try broader keywords
+> - Use `--deep` flag to search inside conversation transcripts
+> - Check if you were working in a different project directory
+
+**After presenting results**, offer these follow-up actions:
+- "To read a specific session transcript, provide the session ID and I will search it for more detail."
+- "To read a memory file, I can open it directly."
+</action>
+<verify>
+- Results were presented to the user.
+</verify>
+<on-fail>
+Report: "Search completed but encountered an error formatting results. Raw match count: {count}."
+</on-fail>
+</step>
+
+## How Folder Discovery Works
+
+Claude Code stores project data in `~/.claude/projects/` using path-encoded folder names:
+
+```
+~/.claude/projects/
+  C--dev-GitHub-SixSevenAI/              <-- primary (this project)
+  C--dev-GitHub-SixSevenAI-src-installer/ <-- related (contains "SixSevenAI")
+  C--dev-workflows-SixSeven-SixSevenAI/  <-- related (contains "SixSevenAI")
+  C--dev-Tools-SixSevenBots/             <-- related (contains "SixSeven")
+  C--dev-SuiteFYI-ConnectFYI/            <-- unrelated (different project)
+```
+
+The suffix match is case-insensitive and uses the last meaningful path segment. Multiple Claude sessions against the same codebase (from different working directories or worktrees) will be discovered and searched.
+
+## What Gets Searched
+
+| Source | File | Searched Fields | Default | --deep |
+|--------|------|-----------------|---------|--------|
+| Session index | `sessions-index.json` | `summary`, `firstPrompt`, `gitBranch` | Yes | Yes |
+| Memory index | `memory/MEMORY.md` | Full content | Yes | Yes |
+| Memory files | `memory/*.md` | Full content + frontmatter | Yes | Yes |
+| Transcripts | `*.jsonl` | Raw text (last 20 sessions) | No | Yes |
+
+## Examples
+
+```bash
+# Find sessions where you worked on the installer
+/im-lost installer publish npm
+
+# Find where you discussed XML conversion
+/im-lost xml-markdown conversion runbook
+
+# Search only memory files for feedback about testing
+/im-lost testing patterns --memory-only
+
+# Deep search transcripts for a specific error
+/im-lost "MISSING_PREREQUISITE" --deep
+
+# Find sessions related to a specific branch
+/im-lost features/v-next --sessions-only
+```
+
+## Related Commands
+
+- `/prime` - Prime context by examining directories, files, or URLs
+- `/ai-dlc:status` - Show current AI-DLC phase and progress