@torus-engineering/tas-kit 1.14.0 → 2.1.0

package/.tas/tools/tas-ado-readme.md

@@ -1,15 +1,17 @@
 # TAS Kit Skill Specification: ado-integration
 
 **Skill name**: ado-integration
-**Version**: 0.2 (concept & spec date: March 17, 2026)
+**Version**: 0.3 (kit v3 — Feature is the only TAS work unit; Epic / User Story removed)
 **Purpose**:
-Allows AI agent in coding environment (Claude Code, Cursor, VS Code + terminal) to perform **CRUD** and two-way sync between Markdown (.md) files in repository and work items on Azure DevOps (Epic, Feature, User Story, Bug), including automatic hierarchy management (parent-child relations).
+Allows AI agent in coding environment (Claude Code, Cursor, VS Code + terminal) to perform **CRUD** and two-way sync between Markdown (.md) files in repository and work items on Azure DevOps (Feature, Bug).
 
 Main objectives:
 - Transform AI into "team member" capable of self-managing backlog without leaving editor
 - Ensure .md file is single source of truth (local ↔ ADO sync)
 - Record sync timestamp so agent knows data freshness
 
+> **Migration note (v3):** Epic and User Story are no longer kit-managed. If your ADO project still uses Epic / User Story templates, treat each TAS Feature as the ADO `Feature` work item; legacy work items pulled with `get` are converted to local `feature-*.md` files.
+
 ## 1. Environment Requirements & Dependencies
 
 **Required**:
@@ -45,16 +47,14 @@ python tools/tas-ado.py <command> [arguments]
 ### Create commands (create new + auto rename file + optional parent)
 
 ```bash
-create-epic   <temp-id> [--parent-id <id>]
 create-feature <temp-id> [--parent-id <id>]
-create-story   <temp-id> [--parent-id <id>]
 create-bug     <temp-id> [--parent-id <id>]
 ```
 
 Example:
 ```bash
-python tools/tas-ado.py create-story 789 --parent-id 456
-# → reads epic-789-*.md → creates User Story → renames to story-1234-*.md → adds parent relation to 456
+python tools/tas-ado.py create-feature 042
+# → reads feature-042-*.md → creates Feature on ADO → renames file to feature-1234-*.md
 ```
 
 ### Pull / Get commands
@@ -67,33 +67,30 @@ pull <ado-id>          # alias
 Example:
 ```bash
 python tools/tas-ado.py get 5345
-# → creates/recreates file epic-5345-slug-title.md with frontmatter + content
+# → creates/updates file feature-5345-slug-title.md
+# → if ADO type is Epic or User Story (legacy), saves as feature-*.md anyway
 ```
 
 ### Update commands
 
 ```bash
-update-epic   <ado-id> [--assign <name/email>] [--status <state>]
-update-feature <ado-id> [--assign <...>] [--status <...>]
-update-story   <ado-id> [--assign <...>] [--status <...>]
+update-feature <ado-id> [--assign <name/email>] [--status <state>]
 update-bug     <ado-id> [--assign <...>] [--status <...>]
 
-update-status <ado-id> --status <state> [--assign <name/email>]
+update-status  <ado-id> --status <state> [--assign <name/email>]
 ```
 
 ### Delete commands
 
 ```bash
-delete-epic   <ado-id>
 delete-feature <ado-id>
-delete-story   <ado-id>
 delete-bug     <ado-id>
 ```
 
 ## 4. File .md Convention
 
 - File name pattern: `{type}-{ado_id or temp_id}-{slug-title}.md`
-  - type: epic / feature / story / bug
+  - type: feature / bug
   - slug-title: lowercase, replace spaces with -, remove special characters
 
 - Frontmatter YAML (always at file start):
@@ -101,21 +98,21 @@ delete-bug     <ado-id>
 ```yaml
 ---
 ado_id: 5345
-ado_type: Epic           # Epic / Feature / User Story / Bug
-ado_title: Landing Page Redesign
+ado_type: Feature        # Feature / Bug (legacy: Epic / User Story still accepted on get)
+ado_title: Checkout Refactor
 ado_state: Active
 ado_assigned_to: Nguyen Van A <email@domain.com>
 ado_created: 2026-03-10
 last_ado_sync: 2026-03-17 15:42:08
-# optional: parent_ado_id: 100   (if want to support auto detect parent later)
+# optional: parent_ado_id: 100
 ---
 # Title (must match ado_title)
 
-Content description, acceptance criteria, steps to reproduce, etc.
+Content description, acceptance criteria, etc.
 ```
 
 **Important**:
-- Every time **create / pull / update / update-status** succeeds → **update** (or add) `last_ado_sync` field with current time (format: YYYY-MM-DD HH:MM:SS)
+- Every successful create / pull / update / update-status → **update** `last_ado_sync` (format: YYYY-MM-DD HH:MM:SS)
 
 ## 5. Detailed Logic Requirements
 
@@ -125,15 +122,16 @@ Content description, acceptance criteria, steps to reproduce, etc.
 3. Create work item using `az boards work-item create`
 4. Get new ID → rename file to `{type}-{new_id}-*.md`
 5. If `--parent-id` → run `az boards work-item relation add --id <new_id> --relation-type parent --target-id <parent_id>`
-6. Auto detect parent_id from frontmatter `parent_ado_id` or file name
+6. Auto detect parent_id from frontmatter `parent_ado_id`
 7. Update `last_ado_sync` in frontmatter
 
 ### When pull/get:
 1. Get work item using `az boards work-item show --id <id> --expand fields`
 2. Convert description (HTML → basic Markdown)
-3. Create file `{type}-{id}-{slug}.md`
-4. Write frontmatter + content
-5. Update `last_ado_sync`
+3. Map ADO type → local file type via `TYPE_REVERSE` (Epic / User Story → feature)
+4. Create file `{type}-{id}-{slug}.md`
+5. Write frontmatter + content
+6. Update `last_ado_sync`
 
 ### When update:
 1. Find file by pattern `*-<id>-*.md`
@@ -149,7 +147,7 @@ Content description, acceptance criteria, steps to reproduce, etc.
 ### When delete:
 - Only delete on ADO, don't touch local files
 
-## 6. Extension Suggestions (not required for v0.2)
+## 6. Extension Suggestions (not required for v0.3)
 
 - Support additional fields: priority, effort, labels, acceptance criteria
 - Auto git add/commit/push after sync
@@ -160,10 +158,9 @@ Content description, acceptance criteria, steps to reproduce, etc.
 Example prompt template for AI agent:
 
 ```
-/ado-create-feature 042 --parent-id 100
-/ado-create-story 789 --parent-id <feature-id just created>
+/ado-create feature 042 --parent-id 100
 /ado-get 1234          # fetch new bug to fix
-/ado-update-bug 1234 --status Resolved
+/ado-update bug 1234 --status Resolved
 ```
 
 This skill is designed to integrate smoothly into agentic frameworks like TAS Kit, Superpowers, BMAD.

package/.tas/tools/tas-ado.py

@@ -3,25 +3,28 @@
 TAS ADO Integration Script
 Two-way sync between local .md files and Azure DevOps work items.
 
+Kit v3: Epic and User Story types are removed — Feature is the only TAS work unit.
+
 Usage: python tools/tas-ado.py <command> [arguments]
 
 Commands:
-  create-epic    <temp-id> [--parent-id <id>]
   create-feature <temp-id> [--parent-id <id>]
-  create-story   <temp-id> [--parent-id <id>]
   create-bug     <temp-id> [--parent-id <id>]
   get            <ado-id>
   pull           <ado-id>   (alias for get)
-  update-epic    <ado-id> [--assign <name>] [--status <state>]
   update-feature <ado-id> [--assign <name>] [--status <state>]
-  update-story   <ado-id> [--assign <name>] [--status <state>]
   update-bug     <ado-id> [--assign <name>] [--status <state>]
   update-status  <ado-id> --status <state> [--assign <name>]
-  delete-epic    <ado-id>
   delete-feature <ado-id>
-  delete-story   <ado-id>
   delete-bug     <ado-id>
 
+  PR commands:
+  pr-get    <pr-id>
+  pr-diff   <pr-id>
+  pr-comment <pr-id> --comment <text>
+  pr-inline  <pr-id> --repo-id <guid> --file <path> --line <n> --comment <text>
+  pr-vote    <pr-id> --vote <approve|reject|wait-for-author|reset>
+
 Prerequisites:
   - Azure CLI + azure-devops extension
   - Python 3.8+ with pyyaml
@@ -29,12 +32,15 @@ Prerequisites:
 """
 
 import argparse
+import base64
 import json
 import os
 import re
 import subprocess
 import sys
 import tempfile
+import urllib.error
+import urllib.request
 from datetime import datetime
 from pathlib import Path
 
@@ -59,15 +65,21 @@ def find_repo_root():
 
 
 def load_dotenv(root):
-    """Load .env file from repo root into os.environ."""
-    env_file = root / ".env"
-    if env_file.exists():
+    """Load .env and .env.local from repo root into os.environ.
+
+    .env.local takes precedence (loaded first; setdefault keeps first value).
+    """
+    for filename in (".env.local", ".env"):
+        env_file = root / filename
+        if not env_file.exists():
+            continue
         with open(env_file, "r", encoding="utf-8") as f:
             for line in f:
                 line = line.strip()
                 if line and not line.startswith("#") and "=" in line:
                     key, _, value = line.partition("=")
-                    os.environ.setdefault(key.strip(), value.strip())
+                    value = value.strip().strip('"').strip("'")
+                    os.environ.setdefault(key.strip(), value)
 
 
 def load_config():
@@ -173,12 +185,18 @@ def az_cmd(args, config):
 # --- File helpers ---
 
 TYPE_MAP = {
-    "epic": "Epic",
     "feature": "Feature",
-    "story": "User Story",
     "bug": "Bug",
 }
 
+# Reverse map for cmd_get — accept legacy ADO types but map them to local Feature.
+TYPE_REVERSE = {
+    "Feature": "feature",
+    "Bug": "bug",
+    "Epic": "feature",       # legacy: any Epic pulled from ADO becomes a local Feature
+    "User Story": "feature", # legacy: any Story pulled from ADO becomes a local Feature
+}
+
 
 def slugify(text):
     text = text.lower().strip()
@@ -304,7 +322,7 @@ def cmd_create(args, config):
     temp_id = args.temp_id
     ado_type = TYPE_MAP.get(file_type)
     if not ado_type:
-        print(f"ERROR: Unknown type '{file_type}'. Use: epic, feature, story, bug")
+        print(f"ERROR: Unknown type '{file_type}'. Use: feature, bug")
         sys.exit(1)
 
     filepath = find_file(config, file_type, temp_id)
@@ -315,7 +333,8 @@ def cmd_create(args, config):
     fm, body = read_md_file(filepath)
     raw_title = fm.get("ado_title") or extract_title(body)
 
-    # Strip "Type-NNN: " prefix (e.g. "Epic-001: Foo" -> "Foo")
+    # Strip "Type-NNN: " prefix (e.g. "Feature-001: Foo" -> "Foo")
+    # Accept legacy Epic/Story prefixes too in case user pulled in old files.
     core_title = re.sub(
         r"^(?:Epic|Feature|Story|User Story|Bug)-\d+[:\s\-]+\s*",
         "",
@@ -331,9 +350,9 @@ def cmd_create(args, config):
 
     description = extract_description(body).replace("\n", "<br>")
 
-    # Default assignee: PE for epic, SE for everything else
+    # Default assignee: PE for feature, SE for bug
     assignee_by_role = config.get("assignee_by_role", {})
-    default_assignee = assignee_by_role.get("pe" if file_type == "epic" else "se", "")
+    default_assignee = assignee_by_role.get("pe" if file_type == "feature" else "se", "")
 
     status = extract_status(fm, body)
 
@@ -374,10 +393,10 @@ def cmd_create(args, config):
             args.parent_id = fm_parent
             print(f"  Auto-detected parent from frontmatter: #{args.parent_id}")
         else:
-            # Check immediate parent dir first, then grandparent
-            # (stories live directly in feature folder; features live in their own subfolder inside epic)
+            # Check immediate parent dir for a numeric ID match (legacy/optional).
+            # Kit v3 features live flat in docs/features/ — usually no parent to detect.
             for parent_dir in [Path(filepath).parent, Path(filepath).parent.parent]:
-                m = re.match(r"^(?:epic|feature|story)-(\d+)", parent_dir.name, re.IGNORECASE)
+                m = re.match(r"^(?:epic|feature)-(\d+)", parent_dir.name, re.IGNORECASE)
                 if m:
                     args.parent_id = m.group(1)
                     print(f"  Auto-detected parent from directory: #{args.parent_id}")
@@ -401,7 +420,7 @@ def cmd_create(args, config):
 
     old_folder = old_path.parent
     new_folder = old_folder
-    is_work_item_folder = file_type in ("epic", "feature") and bool(
+    is_work_item_folder = file_type == "feature" and bool(
         re.match(rf"(?i)^{re.escape(file_type)}-", old_folder.name)
     )
     if is_work_item_folder:
@@ -452,8 +471,7 @@ def cmd_get(args, config):
     description = fields.get("System.Description", "")
     created = fields.get("System.CreatedDate", "")[:10]
 
-    type_reverse = {v: k for k, v in TYPE_MAP.items()}
-    file_type = type_reverse.get(work_item_type, "story")
+    file_type = TYPE_REVERSE.get(work_item_type, "feature")
 
     slug = slugify(title)
     filename = f"{file_type}-{ado_id}-{slug}.md"
@@ -563,13 +581,265 @@ def cmd_delete(args, config):
         print(f"  Updated local file status to Removed (file preserved)")
 
 
+# --- PR helpers ---
+
+def get_org_name(org):
+    """Extract bare org name from org URL or name."""
+    org = org.rstrip("/")
+    if "dev.azure.com/" in org:
+        return org.split("dev.azure.com/")[-1]
+    if "visualstudio.com" in org:
+        return org.split(".visualstudio.com")[0].split("/")[-1]
+    return org
+
+
+def az_repos_cmd(args, config):
+    """Run az repos command with explicit org/project from config."""
+    az_bin = "az.cmd" if sys.platform == "win32" else "az"
+
+    tmp_files = []
+    safe_args = []
+    i = 0
+    while i < len(args):
+        if (len(str(args[i])) > 200 or "\n" in str(args[i])) and i > 0 and str(args[i - 1]).startswith("--"):
+            tmp = tempfile.NamedTemporaryFile(mode="wb", suffix=".txt", delete=False)
+            tmp.write(str(args[i]).encode("utf-8"))
+            tmp.close()
+            tmp_files.append(tmp.name)
+            safe_args.append(f"@{tmp.name}")
+        else:
+            safe_args.append(args[i])
+        i += 1
+
+    # az repos pr show/list/comment/vote use globally-unique PR id; --project not accepted
+    is_repos_pr = len(safe_args) >= 2 and safe_args[0] == "repos" and safe_args[1] == "pr"
+    base_cmd = [az_bin] + safe_args + ["--org", config["org"]]
+    if not is_repos_pr:
+        base_cmd += ["--project", config["project"]]
+    cmd = base_cmd + ["--output", "json"]
+    env = os.environ.copy()
+    if config["pat"]:
+        env["AZURE_DEVOPS_EXT_PAT"] = config["pat"]
+
+    try:
+        result = subprocess.run(cmd, capture_output=True, env=env)
+    finally:
+        for f in tmp_files:
+            try:
+                os.unlink(f)
+            except OSError:
+                pass
+
+    def _decode(b):
+        try:
+            return (b or b"").decode("utf-8")
+        except UnicodeDecodeError:
+            return (b or b"").decode("cp1252", errors="replace")
+
+    if result.returncode != 0:
+        print(f"ERROR: az command failed:\n{_decode(result.stderr)}")
+        sys.exit(1)
+    stdout = _decode(result.stdout).strip()
+    if not stdout:
+        return {}
+    json_start = re.search(r"[{\[]", stdout)
+    if json_start:
+        return json.loads(stdout[json_start.start():])
+    return {}
+
+
+def ado_rest(method, url, data, config):
+    """Call ADO REST API using PAT Basic auth."""
+    token = base64.b64encode(f":{config['pat']}".encode("utf-8")).decode("ascii")
+    body = json.dumps(data).encode("utf-8")
+    req = urllib.request.Request(url, data=body, method=method)
+    req.add_header("Authorization", f"Basic {token}")
+    req.add_header("Content-Type", "application/json")
+    req.add_header("Accept", "application/json")
+    try:
+        with urllib.request.urlopen(req) as resp:
+            return json.loads(resp.read().decode("utf-8"))
+    except urllib.error.HTTPError as e:
+        err_body = e.read().decode("utf-8")
+        print(f"ERROR: ADO REST API {method} {url} failed ({e.code}): {err_body}")
+        sys.exit(1)
+
+
+# --- PR Commands ---
+
+def cmd_pr_get(args, config):
+    pr_id = args.pr_id
+    result = az_repos_cmd([
+        "repos", "pr", "show",
+        "--id", str(pr_id),
+    ], config)
+
+    source_ref = result.get("sourceRefName", "")
+    target_ref = result.get("targetRefName", "")
+    source_branch = source_ref.replace("refs/heads/", "")
+    target_branch = target_ref.replace("refs/heads/", "")
+    created_by = result.get("createdBy", {})
+    creator = created_by.get("displayName", "") if isinstance(created_by, dict) else str(created_by)
+    repo = result.get("repository", {})
+    repo_id = repo.get("id", "")
+    repo_name = repo.get("name", "")
+    work_item_refs = result.get("workItemRefs") or []
+    work_items = [str(wi.get("id", "")) for wi in work_item_refs if wi.get("id")]
+
+    print(f"PR_ID: {pr_id}")
+    print(f"TITLE: {result.get('title', '')}")
+    print(f"DESCRIPTION: {result.get('description', '')}")
+    print(f"SOURCE_BRANCH: {source_branch}")
+    print(f"TARGET_BRANCH: {target_branch}")
+    print(f"CREATOR: {creator}")
+    print(f"STATUS: {result.get('status', '')}")
+    print(f"REPO_ID: {repo_id}")
+    print(f"REPO_NAME: {repo_name}")
+    print(f"WORK_ITEMS: {','.join(work_items)}")
+
+
+def cmd_pr_diff(args, config):
+    pr_id = args.pr_id
+    result = az_repos_cmd([
+        "repos", "pr", "show",
+        "--id", str(pr_id),
+    ], config)
+
+    source_ref = result.get("sourceRefName", "")
+    target_ref = result.get("targetRefName", "")
+    source_branch = source_ref.replace("refs/heads/", "")
+    target_branch = target_ref.replace("refs/heads/", "")
+
+    print(f"SOURCE_BRANCH: {source_branch}")
+    print(f"TARGET_BRANCH: {target_branch}")
+
+    fetch_result = subprocess.run(
+        ["git", "fetch", "origin", source_branch],
+        capture_output=True,
+        cwd=str(config["root"]),
+    )
+
+    fetch_head = ""
+    diff_base = f"origin/{target_branch}"
+
+    if fetch_result.returncode == 0:
+        head_result = subprocess.run(
+            ["git", "rev-parse", "FETCH_HEAD"],
+            capture_output=True,
+            cwd=str(config["root"]),
+        )
+        fetch_head = head_result.stdout.decode("utf-8").strip() if head_result.returncode == 0 else ""
+    else:
+        # Fallback for merged PRs where source branch was deleted
+        src_commit = (result.get("lastMergeSourceCommit") or {}).get("commitId", "")
+        tgt_commit = (result.get("lastMergeTargetCommit") or {}).get("commitId", "")
+        if src_commit and tgt_commit:
+            subprocess.run(["git", "fetch", "origin", src_commit], capture_output=True, cwd=str(config["root"]))
+            subprocess.run(["git", "fetch", "origin", tgt_commit], capture_output=True, cwd=str(config["root"]))
+            fetch_head = src_commit
+            diff_base = tgt_commit
+        else:
+            warn = (fetch_result.stderr or b"").decode("utf-8", errors="replace")
+            print(f"WARNING: git fetch failed and no merge commits available: {warn}")
+            return
+
+    print(f"FETCH_HEAD: {fetch_head}")
+
+    diff_result = subprocess.run(
+        ["git", "diff", "--name-status", f"{diff_base}...{fetch_head}"],
+        capture_output=True,
+        cwd=str(config["root"]),
+    )
+    if diff_result.returncode == 0:
+        print("CHANGED_FILES:")
+        print(diff_result.stdout.decode("utf-8", errors="replace").strip())
+    else:
+        warn = (diff_result.stderr or b"").decode("utf-8", errors="replace")
+        print(f"WARNING: git diff failed: {warn}")
+
+
+def cmd_pr_comment(args, config):
+    """Post a PR-level (non-inline) comment thread via REST.
+
+    Older azure-devops CLI extensions don't ship `az repos pr comment`, so we
+    talk to the Threads REST endpoint directly. Resolves the repo id from
+    `pr show`.
+    """
+    pr_id = args.pr_id
+    pr_info = az_repos_cmd([
+        "repos", "pr", "show",
+        "--id", str(pr_id),
+    ], config)
+    repo_id = (pr_info.get("repository") or {}).get("id", "")
+    if not repo_id:
+        print("ERROR: could not resolve repo id for PR")
+        sys.exit(1)
+
+    org_name = get_org_name(config["org"])
+    project = urllib.request.quote(config["project"], safe="")
+    url = (
+        f"https://dev.azure.com/{org_name}/{project}/_apis/git/repositories"
+        f"/{repo_id}/pullRequests/{pr_id}/threads?api-version=7.1"
+    )
+    data = {
+        "comments": [{"parentCommentId": 0, "content": args.comment, "commentType": "text"}],
+        "status": "active",
+    }
+    result = ado_rest("POST", url, data, config)
+    thread_id = result.get("id", "")
+    print(f"Posted comment thread #{thread_id} on PR #{pr_id}")
+
+
+def cmd_pr_inline(args, config):
+    pr_id = args.pr_id
+    repo_id = args.repo_id
+    file_path = "/" + args.file_path.lstrip("/").replace("\\", "/")
+    line = args.line
+
+    org_name = get_org_name(config["org"])
+    project = urllib.request.quote(config["project"], safe="")
+    url = (
+        f"https://dev.azure.com/{org_name}/{project}/_apis/git/repositories"
+        f"/{repo_id}/pullRequests/{pr_id}/threads?api-version=7.1"
+    )
+    data = {
+        "comments": [{"parentCommentId": 0, "content": args.comment, "commentType": "text"}],
+        "status": "active",
+        "threadContext": {
+            "filePath": file_path,
+            "rightFileStart": {"line": line, "offset": 1},
+            "rightFileEnd": {"line": line, "offset": 1},
+        },
+    }
+    result = ado_rest("POST", url, data, config)
+    thread_id = result.get("id", "")
+    print(f"Posted inline comment on {file_path}:{line} (thread #{thread_id})")
+
+
+def cmd_pr_vote(args, config):
+    vote_map = {
+        "approve": "approve",
+        "reject": "reject",
+        "wait": "wait-for-author",
+        "wait-for-author": "wait-for-author",
+        "reset": "reset",
+    }
+    vote = vote_map.get(args.vote.lower(), args.vote)
+    az_repos_cmd([
+        "repos", "pr", "set-vote",
+        "--id", str(args.pr_id),
+        "--vote", vote,
+    ], config)
+    print(f"Voted '{vote}' on PR #{args.pr_id}")
+
+
 # --- Main ---
 
 def main():
     parser = argparse.ArgumentParser(description="TAS ADO Integration")
     subparsers = parser.add_subparsers(dest="command", help="Command to run")
 
-    for t in ["epic", "feature", "story", "bug"]:
+    for t in ["feature", "bug"]:
         p = subparsers.add_parser(f"create-{t}", help=f"Create {t} on ADO")
         p.add_argument("temp_id", help="Temporary ID in filename")
         p.add_argument("--parent-id", dest="parent_id", help="Parent work item ID")
@@ -579,7 +849,7 @@ def main():
         p = subparsers.add_parser(cmd_name, help="Pull work item from ADO")
         p.add_argument("ado_id", type=int, help="ADO work item ID")
 
-    for t in ["epic", "feature", "story", "bug"]:
+    for t in ["feature", "bug"]:
         p = subparsers.add_parser(f"update-{t}", help=f"Update {t} on ADO")
         p.add_argument("ado_id", type=int, help="ADO work item ID")
         p.add_argument("--assign", help="Assign to name/email")
@@ -591,11 +861,34 @@ def main():
     p.add_argument("--status", required=True, help="New state")
     p.add_argument("--assign", help="Assign to name/email")
 
-    for t in ["epic", "feature", "story", "bug"]:
+    for t in ["feature", "bug"]:
         p = subparsers.add_parser(f"delete-{t}", help=f"Delete {t} on ADO")
         p.add_argument("ado_id", type=int, help="ADO work item ID")
         p.set_defaults(type=t)
 
+    p = subparsers.add_parser("pr-get", help="Get PR metadata")
+    p.add_argument("pr_id", type=int, help="Pull Request ID")
+
+    p = subparsers.add_parser("pr-diff", help="Fetch PR branch and list changed files")
+    p.add_argument("pr_id", type=int, help="Pull Request ID")
+
+    p = subparsers.add_parser("pr-comment", help="Post summary comment thread on PR")
+    p.add_argument("pr_id", type=int, help="Pull Request ID")
+    p.add_argument("--comment", required=True, help="Comment text (markdown supported)")
+
+    p = subparsers.add_parser("pr-inline", help="Post inline comment on PR diff")
+    p.add_argument("pr_id", type=int, help="Pull Request ID")
+    p.add_argument("--repo-id", dest="repo_id", required=True, help="Repository GUID (from pr-get REPO_ID)")
+    p.add_argument("--file", dest="file_path", required=True, help="File path (e.g. src/api/users.ts)")
+    p.add_argument("--line", type=int, required=True, help="Line number")
+    p.add_argument("--comment", required=True, help="Inline comment text")
+
+    p = subparsers.add_parser("pr-vote", help="Set vote on PR")
+    p.add_argument("pr_id", type=int, help="Pull Request ID")
+    p.add_argument("--vote", required=True,
+                   choices=["approve", "reject", "wait-for-author", "reset"],
+                   help="Vote to cast")
+
     args = parser.parse_args()
     if not args.command:
         parser.print_help()
@@ -613,6 +906,16 @@ def main():
         cmd_update_status(args, config)
     elif args.command.startswith("delete-"):
         cmd_delete(args, config)
+    elif args.command == "pr-get":
+        cmd_pr_get(args, config)
+    elif args.command == "pr-diff":
+        cmd_pr_diff(args, config)
+    elif args.command == "pr-comment":
+        cmd_pr_comment(args, config)
+    elif args.command == "pr-inline":
+        cmd_pr_inline(args, config)
+    elif args.command == "pr-vote":
+        cmd_pr_vote(args, config)
     else:
         parser.print_help()