菜单
Azure DevOps work items, queries, and dashboards. Use for "get workitem", "fetch task", "show query", "analyze task", or any Azure DevOps URL. For time logging (Timely sync, Clarity fill), invoke the `/gt:timelog` command instead — this skill defers to it.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Control macOS UI via the Accessibility API and record short screen captures reviewed frame-by-frame. Use when automating native apps — clicking buttons, filling forms, reading element state, finding elements, getting window bounds — or when the user wants a short screen recording of an animation/transition/flicker reviewed via a contact sheet, optionally pushed to a vitrinka board. Triggers on "click button in app", "fill form in native app", "automate macOS app", "list UI elements", "find button", "read text field value", "get window position", "interact with native app", "record the screen", "capture this transition", "Xs recording", "Nfps", "watch this animation", "why does it jump/flicker", "record this app for a few seconds", "push the recording to a board".
Drive a real, extension-loaded Chrome/Brave browser for the GenesisTools YouTube extension via chrome-devtools-mcp as a standalone MCP client — no Claude Code config edits, no restart. Use whenever the user wants to test, debug, screenshot, or click through the YouTube extension's UI (side panel tabs, popup, content script), asks to "load the extension and click X", wants a screenshot of the extension mid-interaction, or needs to find exact pixel coordinates to click something inside a web page (not a native macOS app — for that, use the screen-capture skill instead). Also use this whenever chrome-devtools-mcp's own tools are attempted and fail with "no browser" or launch a blank vanilla browser with no extension loaded — that means the server wasn't wired to a real browser, and this skill is the fix.
Cross-agent messaging protocol via `tools agents`. Use when the main agent is about to spawn subagents that should be able to message each other (or back to the main agent) during their work. Teaches the login (auto-registers) / message / discover / listen pattern, the stream-vs-once receive modes, and the no-loss delivery contract.
Empirically evaluate and trim ANY agent-instruction file — user/project CLAUDE.md, CLAUDE.local.md, AND AGENTS.md (the cross-tool standard read by Codex, Cursor, Gemini CLI) — to find rules a clean model already knows, rules the repo/hooks already enforce, and internal duplication, then propose evidence-backed cuts. Autodetects which files exist and which the harness actually loads, runs headless claude with ALL instruction files EXCLUDED (verified by canary) against MULTIPLE models, and classifies every rule KEEP/SHRINK/CUT with a multi-provider rubric. Use when the user says "improve agents.md", "evaluate agents.md", "trim CLAUDE.md", "improve claude md", "is my CLAUDE.md bloated", or "/improve-agents-md [path]".
Create and maintain a compaction-proof progress file (.claude/plans/<plan>.handoff.md) next to an implementation plan, so any executor - fresh session, post-compact resume, or parallel subagent - knows exactly where the work stands and which plan lines to read next. Invoke when starting to execute a plan, when resuming one, when handing work to another agent, or when the user says "create a handoff" / "prepare the handoff".
Write comprehensive implementation plans (subsumes writing-plans - bite-sized TDD tasks, exact paths, complete code) hardened with an executor-proof contract - all judgment spent at plan time, REASON/VERIFY/ON-FAIL per task, no-improvisation covenant - so even the weakest model executes to strong-model-quality output. Invoke INSTEAD of writing-plans whenever the plan may be executed by another session, subagent, or weaker model.
| name | gt:azure-devops |
| description | Azure DevOps work items, queries, and dashboards. Use for "get workitem", "fetch task", "show query", "analyze task", or any Azure DevOps URL. For time logging (Timely sync, Clarity fill), invoke the `/gt:timelog` command instead — this skill defers to it. |
Fetch, manage, and analyze Azure DevOps work items using tools azure-devops.
Time logging: If the user wants to log time, sync Timely, or fill Clarity timesheets, stop here and invoke the
/gt:timelogcommand. This skill only covers raw work-item operations.
tools azure-devops workitem <id|ids> # Fetch work item(s)
tools azure-devops query <id|url|name> # Fetch query results (supports name matching)
tools azure-devops query <id> --download-workitems # Download all to files
tools azure-devops dashboard <id|url> # Get dashboard queries
tools azure-devops list # List cached items
tools azure-devops workitem-create # Create work item
tools azure-devops timelog configure # Interactive: setup API key, user, allowed types
tools azure-devops timelog types # List available time types
tools azure-devops timelog list -w <id> # List time logs for work item
tools azure-devops timelog add -w <id> -h <hrs> # Log time entry (with precheck)
tools azure-devops timelog prepare-import add # Stage entries for review before import
tools azure-devops timelog prepare-import list # Review staged entries
tools azure-devops timelog prepare-import remove # Remove staged entry
tools azure-devops timelog prepare-import clear # Clear all staged entries
tools azure-devops timelog import <file> # Bulk import time logs (with precheck)
| Option | Description |
|---|---|
--format ai|md|json | Output format (default: ai) |
--force, --refresh | Bypass cache |
--state <states> | Filter by state (comma-separated) |
--severity <sev> | Filter by severity (comma-separated) |
--download-workitems | Download all items from query |
--category <name> | Save to tasks// |
--task-folders | Save in tasks// subfolder |
--attachments-from <datetime> | Download attachments created after this date |
--attachments-to <datetime> | Download attachments created before this date (default: now) |
--attachments-prefix <prefix> | Only attachments starting with this name |
--attachments-suffix <suffix> | Only attachments ending with this (e.g. .har) |
--output-dir <path> | Custom directory for downloaded attachments |
--images | Download inline images from description/comments |
.claude/azure/tasks/ → <id>-<Slug-Title>.md--category react19: .claude/azure/tasks/react19/<id>-<Slug>.md--task-folders: .claude/azure/tasks/<id>/<id>-<Slug>.mdAttachments are downloaded when any --attachments-* filter flag is provided. Without filters, attachments are listed in output with a suggested download command.
.claude/azure/tasks/<taskid>-<attachment-name>--task-folders: .claude/azure/tasks/<id>/<taskid>-<attachment-name>--output-dir /custom/path: /custom/path/<taskid>-<attachment-name>Inline images (screenshots embedded in description/comments HTML) are downloaded when --images is provided.
.claude/azure/tasks/<taskid>-<imagename>.png--task-folders: .claude/azure/tasks/<id>/<taskid>-<imagename>.png.md file with relative pathsRecommended: Use --task-folders --images together to keep each work item's files organized in its own directory.
tools azure-devops workitem 261575
tools azure-devops workitem 261575,261576,261577
tools azure-devops workitem 261575 --category react19
tools azure-devops workitem 261575 --force
The --query option supports three input formats:
d6e14134-9d22-4cbb-b897-b1514f888667https://dev.azure.com/org/project/_queries/query/abc123"Otevřené bugy" (fuzzy matching supported)# By ID
tools azure-devops query d6e14134-9d22-4cbb-b897-b1514f888667
# By name (uses fuzzy matching to find the query)
tools azure-devops query "Open Bugs"
tools azure-devops query "Otevřené bugy"
# With filters
tools azure-devops query <id> --state Active,Development
tools azure-devops query "Active Tasks" --download-workitems --category react19
Query Name Matching:
When user says "analyze workitem/task X" or "analyze tasks from query Y":
Fetch work item(s) with images:
tools azure-devops workitem <ids> --category <cat> --task-folders --images
Read the generated .md file for each work item
Read inline images - Check for image files next to the work item's .md file:
ls $(dirname <path-to-workitem.md>)/<id>-*.{png,jpg,gif,jpeg} 2>/dev/null
If images exist, use the Read tool to view each image file. This gives visual context for:
Spawn Explore agent (Task tool with subagent_type: "Explore") for each:
Analyze codebase for Azure DevOps work item:
**#{id}: {title}**
State: {state} | Severity: {severity}
**Description:** {description}
**Visual Context:** {describe what the inline images show, if any}
**Comments:** {comments}
Find:
1. Relevant code files/components for this issue
2. Current implementation and data flow
3. Required changes
4. Dependencies and related systems
5. Complexity assessment
Return: files found, current implementation, recommended approach, considerations, complexity (Low/Medium/High)
Write .analysis.md next to the work item file:
.claude/azure/tasks/261575-Title.md.claude/azure/tasks/261575-Title.analysis.md# Analysis: #{id} - {title}
**Analyzed**: {timestamp}
**Work Item**: {path to .md file}
## Summary
{1-2 sentence findings summary}
## Relevant Code
- `path/file.ts` - {purpose}
## Current Implementation
{How current code works}
## Recommended Approach
{Step-by-step plan}
## Considerations
- {Risks/considerations}
## Complexity: {Low|Medium|High}
{Reasoning}
| User Request | Action |
|---|---|
| "Get workitem 261575" | tools azure-devops workitem 261575 |
| "Show query results for X" | tools azure-devops query X |
| "Show Open Bugs query" | tools azure-devops query "Open Bugs" |
| "Fetch Otevřené bugy" | tools azure-devops query "Otevřené bugy" |
| "Download React19 bugs" | tools azure-devops query "React19 Bugs" --download-workitems --category react19 |
| "Analyze task 261575" | Fetch → Explore agent → Write .analysis.md |
| "Analyze all active bugs" | Fetch query with --download-workitems → Parallel Explore agents → Write .analysis.md files |
| "Download .har files from task 12345" | tools azure-devops workitem 12345 --attachments-suffix .har |
| "Get attachments from last hour for 12345" | Compute datetime 1h ago, then tools azure-devops workitem 12345 --attachments-from "2026-02-12T10:00:00" |
| "Download all attachments for task 12345" | tools azure-devops workitem 12345 --attachments-from 2000-01-01 |
| "Get task 261575 with screenshots" | tools azure-devops workitem 261575 --task-folders --images |
| "Analyze bug with images" | Fetch with --images → Read images → Explore agent with visual context |
The --create command supports multiple modes for creating new work items.
tools azure-devops workitem-create -i # Interactive mode
tools azure-devops workitem-create --from-file <path> # From template file
tools azure-devops workitem-create <query-url> --type Bug # Generate template from query
tools azure-devops workitem-create <workitem-url> # Generate template from work item
tools azure-devops workitem-create --type Task --title X # Quick creation
| Option | Description |
|---|---|
-i, --interactive | Interactive mode with step-by-step prompts |
--from-file <path> | Create from template JSON file |
--type <type> | Work item type (Bug, Task, User Story, etc.) |
--title <text> | Work item title (required for quick mode) |
--severity <sev> | Severity level |
--tags <tags> | Tags (comma-separated) |
--assignee <email> | Assignee email |
-i)Best for: Manual creation with full control over all fields.
tools azure-devops workitem-create -i
Prompts for: type, title, description (via editor), severity, state, tags, assignee, parent link.
Best for: Creating work items that match patterns from existing items.
tools azure-devops workitem-create "https://dev.azure.com/.../_queries/query/abc" --type Bug
This:
.claude/azure/tasks/created/template-<timestamp>.jsonBest for: Cloning or creating similar work items.
tools azure-devops workitem-create "https://dev.azure.com/.../_workitems/edit/12345"
This:
.claude/azure/tasks/created/template-<timestamp>.jsonBest for: LLM workflows where templates are prepared programmatically.
tools azure-devops workitem-create --from-file ".claude/azure/tasks/created/template.json"
Template format:
{
"$schema": "azure-devops-workitem-v1",
"type": "Bug",
"fields": {
"title": "Error in checkout flow",
"description": "<p>Description here</p>",
"severity": "A - critical",
"tags": ["frontend", "checkout"],
"assignedTo": "user@example.com",
"areaPath": "Project\\Area",
"iterationPath": "Project\\Sprint1"
},
"relations": {
"parent": 12345
}
}
Best for: Simple work items created from command line.
tools azure-devops workitem-create --type Task --title "Fix login bug"
tools azure-devops workitem-create --type Bug --title "Error" --severity "A - critical" --tags "frontend,urgent"
When user asks to "create a work item" or "file a bug":
Gather information - Ask for: type, title, description, severity (if bug)
Choose mode based on context:
--from-file--type X --title "Y"Create the work item:
# Quick creation
tools azure-devops workitem-create --type Bug --title "Error in checkout" --severity "B - high"
# Or from template
tools azure-devops workitem-create --from-file template.json
Report the result - Include the work item ID and URL in your response.
| User Request | Action |
|---|---|
| "Create a bug for the login issue" | --create --type Bug --title "Login issue" --severity "B - high" |
| "File a task to update docs" | --create --type Task --title "Update documentation" |
| "Create a bug like #12345" | --create <workitem-url> then --from-file template.json |
| "Help me create a detailed work item" | --create -i (interactive) |
Track work item history: who changed what, when, and how long items spent in each state.
tools azure-devops history show <id> # Summary view (assignment/state periods, time-in-state)
tools azure-devops history show <id> -f timeline # Chronological events
tools azure-devops history show <id> -f json # JSON output
tools azure-devops history show <id> --force # Force refresh from API
tools azure-devops history show <id> --assigned-to "X" # Filter by assignee
tools azure-devops history show <id> --state Active # Filter by state
tools azure-devops history search --assigned-to-me --wiql # Currently assigned to me (WIQL @Me)
tools azure-devops history search --assigned-to "Martin" --wiql # Ever assigned to user (server-side)
tools azure-devops history search --assigned-to "Martin" --wiql --current # Currently assigned
tools azure-devops history search --assigned-to "Martin" # Local cached history search
tools azure-devops history search --assigned-to "Martin" --min-time 2h # Min time filter
tools azure-devops history search --state Active --since 2024-12-01 --wiql # State + date range (--since/--until aliases for --from/--to)
tools azure-devops history sync # Bulk sync history for cached work items (per-item mode)
tools azure-devops history sync --force # Force re-sync all
tools azure-devops history sync --dry-run # Show what would be synced
tools azure-devops history sync --batch # Use batch reporting API instead
| User says | Command |
|---|---|
| "tasks assigned to me" | history search --assigned-to-me --wiql |
| "tasks ever assigned to Martin" | history search --assigned-to "Martin" --wiql |
| "how long was #123 in Active" | history show 123 --state Active |
| "time Martin spent on #456" | history show 456 --assigned-to Martin |
| "all work in last 2 months" | history search --assigned-to "Martin" --from 2024-12-01 --wiql |
--assigned-to @me or --assigned-to-me uses WIQL @Me macro (auto-enables WIQL)= instead of EVER for current assignment--batch): Uses reporting API, better for 500+ itemsTime logging for Azure DevOps work items using the third-party TimeLog extension.
# Interactive configuration (recommended)
tools azure-devops timelog configure
This launches an interactive prompt (using clack) that configures:
functionsKey: TimeLog API key (auto-fetched from Azure DevOps)defaultUser: Your user email/name for time loggingallowedWorkItemTypes: Work item types that can be logged to (e.g., "Bug,Task")allowedStatesPerType: Required states per type (e.g., "Task:In Progress")The configuration is saved to .claude/azure/config.json.
Note for LLM agents: Since configure uses interactive clack prompts, you cannot drive it directly. Use AskUserQuestion to suggest the user run it themselves, or use non-interactive flags:
# Non-interactive mode (for scripting)
tools azure-devops timelog configure --allowed-work-item-types "Bug,Task" --allowed-states-for-type "Task:In Progress"
tools azure-devops timelog types # AI-friendly list
tools azure-devops timelog types --format json # JSON output
tools azure-devops timelog list -w <workItemId>
tools azure-devops timelog list -w 268935 --format md
tools azure-devops timelog list --from 2026-02-01 --to 2026-02-08 --format json
tools azure-devops timelog list --from 2026-02-01 --to 2026-02-08 --user @me --format json
The --user @me resolves to the configured default username. Use --from/--to for date ranges (--since/--upto also accepted as aliases).
# Quick mode (all options on CLI)
tools azure-devops timelog add -w <id> -h <hours> -t <type>
tools azure-devops timelog add -w 268935 -h 2 -t "Development"
tools azure-devops timelog add -w 268935 -h 1 -m 30 -t "Code Review" -c "PR review"
# Interactive mode
tools azure-devops timelog add -i
tools azure-devops timelog add -w 268935 -i
Before creating the entry, the command runs a workitem type precheck (see Workitem Type Validation below).
The prepare-import workflow allows you to stage, review, and validate entries before committing them to Azure DevOps.
# Stage entries for review
tools azure-devops timelog prepare-import add --from 2026-02-01 --to 2026-02-08 --entry '{
"workItemId": 268935,
"date": "2026-02-04",
"hours": 2,
"timeType": "Development",
"comment": "Implemented feature X"
}'
# Add another entry (same date range)
tools azure-devops timelog prepare-import add --from 2026-02-01 --to 2026-02-08 --entry '{
"workItemId": 262042,
"date": "2026-02-04",
"hours": 0.5,
"timeType": "Ceremonie",
"comment": "Daily standup"
}'
# Review all staged entries
tools azure-devops timelog prepare-import list --name 2026-02-01.2026-02-08 --format table
# Remove a specific entry if needed
tools azure-devops timelog prepare-import remove --name 2026-02-01.2026-02-08 --id <uuid>
# Clear all entries for this date range
tools azure-devops timelog prepare-import clear --name 2026-02-01.2026-02-08
The name is auto-generated from --from and --to as <from>.<to> (e.g., 2026-02-01.2026-02-08). Each entry is validated with Zod schema and runs workitem type precheck before being added.
Do not include workItemTitle in staged --entry JSON unless you already have the exact ADO title. Precheck backfills it on prepare-import add; prepare-import list and import --dry-run fill any still-missing titles. Run from the project directory that has .claude/azure/config.json (e.g. col-fe/).
# Import from prepare-import staging file
tools azure-devops timelog import .genesis-tools/azure-devops/cache/prepare-import/2026-02-01.2026-02-08.json
# Or import from custom JSON file
tools azure-devops timelog import entries.json
# Dry run — validates, backfills workItemTitle into the file, does not create entries
tools azure-devops timelog import entries.json --dry-run
Omit workItemTitle when authoring import JSON. Precheck resolves titles (single ADO fetch per work item) and --dry-run writes them back to the source file.
The import command runs workitem type precheck for each entry before creating it. After import, the command shows a precheck summary with counts of passed/redirected/failed entries. The cache is automatically evicted after successful imports.
Before creating time log entries, the tool validates that the workitem type is configured as allowed in allowedWorkItemTypes. This precheck behavior helps prevent errors:
Automatic Redirect for User Stories:
Configuration:
tools azure-devops timelog configure to set allowedWorkItemTypes"Bug,Task" (excludes User Stories, Features, etc.)allowedStatesPerType for additional validationWhere Precheck Applies:
timelog add - Before creating single entrytimelog import - Before importing each entryprepare-import add - When staging entry for review| User Request | Action |
|---|---|
| "Log 2 hours on task 268935" | tools azure-devops timelog add -w 268935 -h 2 -t "Development" |
| "What time types are available?" | tools azure-devops timelog types |
| "Show time logged on 268935" | tools azure-devops timelog list -w 268935 |
| "Help me log time" | tools azure-devops timelog add -i |
| "Stage entries for review" | tools azure-devops timelog prepare-import add --from ... --to ... --entry '{...}' |
| "Review staged entries" | tools azure-devops timelog prepare-import list --name 2026-02-01.2026-02-08 |
| "Import time entries from file" | tools azure-devops timelog import entries.json |
| "Import with validation only" | tools azure-devops timelog import entries.json --dry-run |
When user asks to log time in natural language, parse their request and construct the CLI command:
1. Parse Duration Formats:
-h 1-h 0 -m 30-h 1 -m 30-h 2 -m 152. Extract Work Item IDs:
feature/268935-fix-login → work item 268935feat(#268935): fix login bug → work item 268935To extract from git context:
# Get current branch
git branch --show-current
# Get recent commit messages (look for #NNNNNN patterns)
git log --oneline -5
3. Infer Time Type from Context:
| Context Clues | Time Type |
|---|---|
| "reviewing PR", "code review", "review" | Code Review |
| "implementing", "coding", "development", "fixing" | Development |
| "testing", "writing tests", "QA" | Test |
| "documentation", "docs", "readme" | Dokumentace |
| "meeting", "standup", "planning", "retro" | Ceremonie |
| "analysis", "analyzing", "design" | IT Analýza |
| "configuring", "setup", "deployment" | Konfigurace |
Default to "Development" if no context clues.
4. Use Git Commit Messages as Notes:
When user says "use commit message" or doesn't provide a note:
# Get last commit message
git log -1 --pretty=%B
Use the commit subject line as the time log comment.
| User Request | Parsed Command |
|---|---|
| "log 1h on 268935 for Development" | timelog add -w 268935 -h 1 -t "Development" |
| "spent 30min reviewing PR on task 789" | timelog add -w 789 -h 0 -m 30 -t "Code Review" |
| "log 2 hours, use last commit message" | Get work item from branch/commit, use commit msg as comment |
| "log my work on the current task" | Extract ID from branch, infer type from commits |
| "log 1.5h implementing the fix" | timelog add -w <from-branch> -h 1 -m 30 -t "Development" |
When user says "log time for my work" without explicit details:
Get work item ID:
git branch --show-current # e.g., feature/268935-fix-login
Extract number: 268935
Get commit messages for note:
git log -1 --pretty=%B
Infer time type from commit message keywords
Ask user for duration if not specified (use AskUserQuestion)
Execute:
tools azure-devops timelog add -w 268935 -h <hours> -t "<inferred-type>" -c "<commit-message>"
tools gitFor gathering commit data to correlate with time entries, use the tools git commits command:
# Get commits for a date range with automatic workitem ID extraction
tools git commits --from 2026-02-01 --to 2026-02-08 --format json 2>/dev/null | tools json
This command:
tools git configure authors)The extracted workitem IDs can be used to match commits to Azure DevOps work items for time logging purposes.
When user explicitly asks to download and analyze HAR attachments from a work item:
# 1. Download HAR attachment
tools azure-devops workitem <id> --attachments-suffix .har --output-dir /tmp/har
# 2. Load and analyze with har-analyzer
tools har-analyzer load /tmp/har/<taskid>-capture.har
Do NOT download or analyze HAR files automatically -- only when the user requests it.
For deeper API research beyond this skill:
src/azure-devops/docs/ contains 14 reference files (work items, iterations, PRs, REST API, WIQL syntax, timelog history)/websites/learn_microsoft_en-us_rest_api_azure_devops for detailed REST API specs