prjct-cli 1.45.4 → 1.45.6

package/templates/commands/git.md

@@ -0,0 +1,17 @@
+---
+allowed-tools: [Bash, AskUserQuestion]
+---
+
+# p. git $ARGUMENTS
+
+Supports: `commit`, `push`, `sync`, `undo`.
+
+## BLOCKING: Never commit/push to main/master.
+
+```bash
+prjct git $ARGUMENTS --md
+```
+
+Follow the instructions in the CLI output.
+
+Every commit MUST include footer: `Generated with [p/](https://www.prjct.app/)`

package/templates/commands/history.md

@@ -0,0 +1,13 @@
+---
+allowed-tools: [Bash]
+---
+
+# p. history $ARGUMENTS
+
+Supports: `undo`, `redo` (default: show snapshot history).
+
+```bash
+prjct history $ARGUMENTS --md
+```
+
+Follow the instructions in the CLI output.

package/templates/commands/idea.md

@@ -0,0 +1,13 @@
+---
+allowed-tools: [Bash, AskUserQuestion]
+---
+
+# p. idea $ARGUMENTS
+
+If $ARGUMENTS is empty, ASK the user for their idea.
+
+```bash
+prjct idea "$ARGUMENTS" --md
+```
+
+Follow the instructions in the CLI output.

package/templates/commands/impact.md

@@ -0,0 +1,13 @@
+---
+allowed-tools: [Bash, AskUserQuestion]
+---
+
+# p. impact $ARGUMENTS
+
+Supports: `list`, `summary`, or specific feature ID (default: most recent ship).
+
+```bash
+prjct impact $ARGUMENTS --md
+```
+
+Follow the instructions in the CLI output. When collecting effort data, success metrics, and learnings, ask the user for input.

package/templates/commands/init.md

@@ -0,0 +1,11 @@
+---
+allowed-tools: [Bash, AskUserQuestion]
+---
+
+# p. init $ARGUMENTS
+
+```bash
+prjct init $ARGUMENTS --md
+```
+
+Follow the instructions in the CLI output.

package/templates/commands/jira.md

@@ -0,0 +1,88 @@
+---
+allowed-tools: ["*"]
+---
+
+# p. jira $ARGUMENTS
+
+Jira is MCP-only — no API tokens, no REST calls.
+
+## Step 0: Check MCP readiness (ALWAYS, except for `setup`)
+
+Look for tools starting with `mcp__jira` or `mcp__atlassian` in your tool list.
+
+**If tools ARE available** → proceed with the requested operation below.
+
+**If NOT available** → run: `prjct jira status --md` and interpret:
+
+| Status | Action |
+|--------|--------|
+| `configured: false` | Run `p. jira setup` |
+| `configured: true, oauthReady: false` | Tell user to run in a terminal: `npx -y mcp-remote@0.1.38 https://mcp.atlassian.com/v1/mcp` then restart their AI client |
+| `configured: true, oauthReady: true` | Tell user to restart their AI client |
+
+**Do NOT attempt MCP tool calls if Jira tools are not in your tool list.**
+
+---
+
+## Setup (`p. jira setup`)
+
+Run step by step:
+
+### Step 1: Write MCP config
+```bash
+prjct jira setup --md
+```
+
+### Step 2: Complete OAuth in terminal (REQUIRED before restarting)
+
+Tell the user to open a NEW terminal and run this **exact** command (version pinned to match mcp.json):
+```
+npx -y mcp-remote@0.1.38 https://mcp.atlassian.com/v1/mcp
+```
+
+This will:
+1. Print an OAuth URL
+2. Try to open the browser automatically
+3. If browser doesn't open → copy-paste the URL manually
+
+Tell the user: **Complete the authorization in the browser, then come back here.**
+
+Wait for the user to confirm they completed OAuth before continuing.
+
+### Step 3: Restart Claude Code
+
+Tell the user: "Close and reopen Claude Code. The Jira MCP tools will be ready."
+
+After restart, Jira MCP tools are available — no more auth needed.
+
+---
+
+## Status (`p. jira status`)
+
+```bash
+prjct jira status --md
+```
+
+---
+
+## Sprint / Backlog
+
+```bash
+prjct jira sprint --md   # → JQL for active sprint
+prjct jira backlog --md  # → JQL for backlog
+```
+
+Use the returned JQL with the Jira MCP search tool (available after setup + restart).
+Show sprint and backlog issues **separately**:
+- `## 🏃 Active Sprint` for sprint issues
+- `## 📋 Backlog` for backlog issues
+
+---
+
+## Issue Operations (list / get / create / update / start / done)
+
+Use Jira MCP tools directly. No REST API, no API tokens.
+
+- `start <KEY>`: transition to In Progress via MCP → `prjct task "<title>" --md`
+- `done <KEY>`: transition to Done via MCP → `prjct done --md`
+- `list`: fetch assigned issues via MCP → show as table

package/templates/commands/learnings.md

@@ -0,0 +1,11 @@
+---
+allowed-tools: [Bash]
+---
+
+# p. learnings
+
+```bash
+prjct learnings --md
+```
+
+Follow the instructions in the CLI output.

package/templates/commands/linear.md

@@ -0,0 +1,82 @@
+---
+allowed-tools: ["*"]
+---
+
+# p. linear $ARGUMENTS
+
+Linear is MCP-only — no SDK, no API tokens.
+
+## Step 0: Check MCP readiness (ALWAYS, except for `setup`)
+
+Look for tools starting with `mcp__linear` in your tool list.
+
+**If tools ARE available** → proceed with the requested operation below.
+
+**If NOT available** → run: `prjct linear status --md` and interpret:
+
+| Status | Action |
+|--------|--------|
+| `configured: false` | Run `p. linear setup` |
+| `configured: true, oauthReady: false` | Tell user to run in a terminal: `npx -y mcp-remote@0.1.38 https://mcp.linear.app/mcp` then restart their AI client |
+| `configured: true, oauthReady: true` | Tell user to restart their AI client |
+
+**Do NOT attempt MCP tool calls if Linear tools are not in your tool list.**
+
+---
+
+## Setup (`p. linear setup`)
+
+Run step by step:
+
+### Step 1: Write MCP config
+```bash
+prjct linear setup --md
+```
+
+### Step 2: Complete OAuth in terminal (REQUIRED before restarting)
+
+Tell the user to open a NEW terminal and run this **exact** command (version pinned to match mcp.json):
+```
+npx -y mcp-remote@0.1.38 https://mcp.linear.app/mcp
+```
+
+This will:
+1. Print an OAuth URL
+2. Try to open the browser automatically
+3. If browser doesn't open → copy-paste the URL manually
+
+Tell the user: **Complete the authorization in the browser, then come back here.**
+
+Wait for the user to confirm they completed OAuth before continuing.
+
+### Step 3: Restart Claude Code
+
+Tell the user: "Close and reopen Claude Code. The Linear MCP tools will be ready."
+
+After restart, Linear MCP tools are available — no more auth needed.
+
+---
+
+## Status (`p. linear status`)
+
+```bash
+prjct linear status --md
+```
+
+---
+
+## Issue Operations (list / get / start / done / update / comment / create)
+
+Use Linear MCP tools directly. No SDK, no API tokens.
+
+- `start <ID>`: move to In Progress via MCP → `prjct task "<title>" --md`
+- `done <ID>`: move to Done via MCP → `prjct done --md`
+- `list`: fetch assigned issues via MCP → show as table with ID, title, status, priority
+
+---
+
+## Sync (`p. linear sync`)
+
+1. Fetch assigned issues via Linear MCP tools
+2. For each untracked issue: `prjct task "<title>" --md`
+3. Show sync summary

package/templates/commands/merge.md

@@ -0,0 +1,25 @@
+---
+allowed-tools: [Bash, AskUserQuestion]
+---
+
+# p. merge
+
+## Pre-flight (BLOCKING)
+Verify: active task exists, PR exists, PR is approved, CI passes, no conflicts.
+
+## Step 1: Get merge plan
+```bash
+prjct merge --md
+```
+
+## Step 2: Get approval (BLOCKING)
+ASK: "Merge this PR?" Yes / No
+
+## Step 3: Execute
+```bash
+gh pr merge {prNumber} --squash --delete-branch
+git checkout main && git pull origin main
+```
+
+## Step 4: Update issue tracker
+If linked to Linear/JIRA, mark as Done via CLI.

package/templates/commands/next.md

@@ -0,0 +1,12 @@
+---
+allowed-tools: [Bash, AskUserQuestion]
+---
+
+# p. next $ARGUMENTS
+
+```bash
+prjct next $ARGUMENTS --md
+```
+If CLI output is JSON with `options`, present the options to the user and execute the chosen command.
+
+Follow the instructions in the CLI output.

package/templates/commands/p.md

@@ -0,0 +1,62 @@
+---
+description: 'prjct CLI - Context layer for AI agents'
+allowed-tools: [Read, Write, Edit, Bash, Glob, Grep, Task, AskUserQuestion, TodoWrite, WebFetch]
+---
+
+# prjct Command Router
+
+**ARGUMENTS**: $ARGUMENTS
+
+All commands use the `p.` prefix.
+
+## Quick Reference
+
+| Command | Description |
+|---------|-------------|
+| `p. task <desc>` | Start a task |
+| `p. done` | Complete current subtask |
+| `p. ship [name]` | Ship feature with PR + version bump |
+| `p. sync` | Analyze project, regenerate agents |
+| `p. pause` | Pause current task |
+| `p. resume` | Resume paused task |
+| `p. next` | Show priority queue |
+| `p. idea <desc>` | Quick idea capture |
+| `p. bug <desc>` | Report bug with auto-priority |
+| `p. linear` | Linear integration (via MCP) |
+| `p. jira` | JIRA integration (via MCP) |
+
+## Execution
+
+```
+1. PARSE: $ARGUMENTS → extract command (first word)
+2. GET npm root: npm root -g
+3. LOAD template: {npmRoot}/prjct-cli/templates/commands/{command}.md
+4. EXECUTE template
+```
+
+## Command Aliases
+
+| Input | Redirects To |
+|-------|--------------|
+| `p. undo` | `p. history undo` |
+| `p. redo` | `p. history redo` |
+
+## State Context
+
+All state is managed by the `prjct` CLI via SQLite (prjct.db).
+Templates should use CLI commands for data operations — never read/write JSON storage files directly.
+
+## Error Handling
+
+| Error | Action |
+|-------|--------|
+| Unknown command | "Unknown command: {command}. Run `p. help` for available commands." |
+| No project | "No prjct project. Run `p. init` first." |
+| Template not found | "Template not found: {command}.md" |
+
+## NOW: Execute
+
+1. Parse command from $ARGUMENTS
+2. Handle aliases (undo → history undo, redo → history redo)
+3. Run `npm root -g` to get template path
+4. Load and execute command template

package/templates/commands/p.toml

@@ -0,0 +1,37 @@
+# prjct Command Router for Gemini CLI
+description = "prjct - Context layer for AI coding agents"
+
+prompt = """
+# prjct Command Router
+
+You are using prjct, a context layer for AI coding agents.
+
+**ARGUMENTS**: {{args}}
+
+## Instructions
+
+1. Parse arguments: first word = `command`, rest = `commandArgs`
+2. Get npm global root by running: `npm root -g`
+3. Read the command template from:
+   `{npmRoot}/prjct-cli/templates/commands/{command}.md`
+4. Execute the template with `commandArgs` as input
+
+## Example
+
+If arguments = "task fix the login bug":
+- command = "task"
+- commandArgs = "fix the login bug"
+- npm root -g → `/opt/homebrew/lib/node_modules`
+- Read: `/opt/homebrew/lib/node_modules/prjct-cli/templates/commands/task.md`
+- Execute template with: "fix the login bug"
+
+## Available Commands
+
+task, done, ship, sync, init, idea, dash, next, pause, resume, bug,
+linear, jira, feature, prd, plan, review, merge, git, test, cleanup,
+design, analyze, history, enrich, update
+
+## Action
+
+NOW run `npm root -g` and read the appropriate command template.
+"""

package/templates/commands/pause.md

@@ -0,0 +1,16 @@
+---
+allowed-tools: [Bash, AskUserQuestion]
+---
+
+# p. pause $ARGUMENTS
+
+If no reason provided, ask the user:
+
+Ask the user: "Why are you pausing?" with options: Blocked, Switching task, Break, Researching
+
+```bash
+prjct pause "$ARGUMENTS" --md
+```
+If CLI output is JSON with `options`, present the options to the user and execute the chosen command.
+
+Follow the instructions in the CLI output.

package/templates/commands/plan.md

@@ -0,0 +1,13 @@
+---
+allowed-tools: [Bash, AskUserQuestion]
+---
+
+# p. plan $ARGUMENTS
+
+Supports: `quarter`, `prioritize`, `add <prd-id>`, `capacity` (default: show status).
+
+```bash
+prjct plan $ARGUMENTS --md
+```
+
+Follow the instructions in the CLI output. When selecting features or adjusting capacity, ask the user for input.

package/templates/commands/prd.md

@@ -0,0 +1,21 @@
+---
+allowed-tools: [Bash, Read, Write, AskUserQuestion, Task]
+---
+
+# p. prd $ARGUMENTS
+
+## Step 1: Validate
+If $ARGUMENTS is empty, ASK the user for a feature title.
+
+## Step 2: Create PRD via CLI
+```bash
+prjct prd "$ARGUMENTS" --md
+```
+
+## Step 3: Follow CLI methodology
+The CLI guides through discovery, sizing, and phase execution.
+Search the codebase for architecture patterns.
+
+## Step 4: Get approval
+Show the PRD summary and get explicit approval.
+ASK: "Add to roadmap now?" Yes / No (keep as draft)

package/templates/commands/resume.md

@@ -0,0 +1,12 @@
+---
+allowed-tools: [Bash, AskUserQuestion]
+---
+
+# p. resume $ARGUMENTS
+
+```bash
+prjct resume $ARGUMENTS --md
+```
+If CLI output is JSON with `options`, present them to the user with AskUserQuestion and execute the chosen command.
+
+Follow the instructions in the CLI output. If the CLI says to switch branches, do so.

package/templates/commands/review.md

@@ -0,0 +1,20 @@
+---
+allowed-tools: [Bash, Read, AskUserQuestion]
+---
+
+# p. review $ARGUMENTS
+
+## Step 1: Run review
+```bash
+prjct review $ARGUMENTS --md
+```
+
+## Step 2: Analyze changes
+Read changed files and check for security issues, logic errors, and missing error handling.
+
+## Step 3: Create/check PR
+If no PR exists, create one with `gh pr create`.
+If PR exists, check approval status with `gh pr view`.
+
+## Step 4: Follow CLI next steps
+The CLI output indicates what to do next (fix issues, wait for approval, merge).

package/templates/commands/serve.md

@@ -0,0 +1,11 @@
+---
+allowed-tools: [Bash]
+---
+
+# p. serve $ARGUMENTS
+
+```bash
+prjct serve ${ARGUMENTS || "3478"} --md
+```
+
+Follow the instructions in the CLI output.

package/templates/commands/sessions.md

@@ -0,0 +1,13 @@
+---
+allowed-tools: [Bash, Read, AskUserQuestion]
+---
+
+# p. sessions
+
+## Step 1: Show recent sessions
+```bash
+prjct sessions --md
+```
+
+## Step 2: Offer to resume
+If sessions exist, ask the user which one to resume. Then switch to that project directory and run `prjct resume --md`.

package/templates/commands/setup.md

@@ -0,0 +1,11 @@
+---
+allowed-tools: [Bash]
+---
+
+# p. setup $ARGUMENTS
+
+```bash
+prjct setup $ARGUMENTS --md
+```
+
+Follow the instructions in the CLI output.

package/templates/commands/ship.md

@@ -0,0 +1,46 @@
+---
+allowed-tools: [Bash, Read, AskUserQuestion]
+---
+
+# p. ship $ARGUMENTS
+
+## Step 0: Complete task (implicit)
+The ship workflow automatically completes the current task before shipping.
+This means `p. done` is implicit — you do NOT need to run it separately before shipping.
+
+## Pre-flight (BLOCKING)
+```bash
+git branch --show-current
+```
+IF on main/master: STOP. Create a feature branch first.
+
+```bash
+gh auth status
+```
+IF not authenticated: STOP. Run `gh auth login`.
+
+## Step 1: Quality checks
+```bash
+prjct ship "$ARGUMENTS" --md
+```
+
+## Step 2: Review changes
+Show the user what will be committed, versioned, and PR'd.
+
+## Step 3: Get approval (BLOCKING)
+ASK: "Ready to ship?" Yes / No / Show diff
+
+## Step 4: Ship
+- Commit with prjct footer: `Generated with [p/](https://www.prjct.app/)`
+- Push and create PR
+- Update issue tracker if linked
+- Every commit MUST include the prjct footer. No exceptions.
+
+
+## Presentation
+Format the ship flow as:
+
+1. `**Shipping**: {feature name}`
+2. Quality checks as a table: | Check | Status |
+3. Show the PR summary
+4. Ask for approval with clear formatting

package/templates/commands/skill.md

@@ -0,0 +1,13 @@
+---
+allowed-tools: [Bash, Read, Glob]
+---
+
+# p. skill $ARGUMENTS
+
+Supports: `list` (default), `search <query>`, `show <id>`, `invoke <id>`, `add <source>`, `remove <name>`, `init <name>`, `check`.
+
+```bash
+prjct skill $ARGUMENTS --md
+```
+
+Follow the instructions in the CLI output.

package/templates/commands/spec.md

@@ -0,0 +1,20 @@
+---
+allowed-tools: [Bash, Read, Write, AskUserQuestion, Task]
+---
+
+# p. spec $ARGUMENTS
+
+## Step 1: Validate
+If $ARGUMENTS is empty, ASK the user for a feature name.
+
+## Step 2: Create spec via CLI
+```bash
+prjct spec "$ARGUMENTS" --md
+```
+
+## Step 3: Follow CLI instructions
+The CLI will guide through requirements, design decisions, and task breakdown.
+Search the codebase for relevant patterns.
+
+## Step 4: Get approval
+Show the spec to the user and get explicit approval before adding tasks to queue.

package/templates/commands/status.md

@@ -0,0 +1,11 @@
+---
+allowed-tools: [Bash]
+---
+
+# p. status $ARGUMENTS
+
+```bash
+prjct status $ARGUMENTS --md
+```
+
+Follow the instructions in the CLI output.

package/templates/commands/sync.md

@@ -0,0 +1,23 @@
+---
+allowed-tools: [Bash, AskUserQuestion]
+---
+
+# p. sync $ARGUMENTS
+
+## Step 1: Run CLI sync
+```bash
+prjct sync $ARGUMENTS --md
+```
+If CLI output is JSON with `options`, present the options to the user and execute the chosen command.
+
+Follow ALL instructions in the CLI output (including LLM Analysis if present).
+
+## Step 2: Present results
+After all steps complete, present the output clearly:
+- Use the tables and sections as-is from CLI markdown
+- If LLM analysis was performed, summarize key findings:
+  - Architecture style and top insights
+  - Critical anti-patterns (high severity)
+  - Top tech debt items
+  - Key conventions discovered
+- Add a brief interpretation of what changed and why