add-task
/add-task
Create a new task file following the taskmd specification. Use when the user wants to add a new task to the project.
- Author
- driangle ↗
- Repository
- taskmd ↗
- Licence
- MIT
- Last repository change
- 02.10.2026
Assessment and security
- Files in this package
- 1
- Scripts
- 0
- HTML comments
- 1
- URLs in SKILL.md
- 0
No signals detected. Repository changed on 02.10.2026, without references to retired models or APIs.
MIT, open: use and modify while retaining author attribution.
- SKILL.md zawiera komentarze HTML (1). Nie widać ich w podglądzie Markdown, a model je czyta.
What it does and when to use it
Create a new task file following the taskmd specification. Use when the user wants to add a new task to the project.
Original author description
Create a new task file following the taskmd specification. Use when the user wants to add a new task to the project.
How it works
- Instructions
SKILL.md preview
---
name: add-task
description: Create a new task file following the taskmd specification. Use when the user wants to add a new task to the project.
allowed-tools: Glob, Read, Write
---
# Add Task
Create a new task file — no CLI required.
## Instructions
The user's task description is in `$ARGUMENTS`.
1. **Parse the user's input** from `$ARGUMENTS` to extract:
- The task **title** (required)
- Any optional metadata: priority, effort, type, tags, group, dependencies, parent, owner, phase
2. **Read configuration**:
- Read `.taskmd.yaml` if it exists for: task `dir` (default: `tasks`), `id` config (strategy, prefix, padding, length), and `phases`
- If the user mentions a phase/milestone/sprint, check `.taskmd.yaml` for configured `phases` and use the matching `id`. If the phase doesn't exist yet, add it to the `phases` array in `.taskmd.yaml` (see `SPEC_REFERENCE.md` for the phases config format)
3. **Determine the group** based on the task's domain:
- If the user specified `--group`, use that
- Otherwise infer it from the task's domain:
- `cli` — CLI commands, terminal features, backend/server code
- `web` — web frontend, UI, components
- no group — cross-cutting, infrastructure, documentation, or genuinely unclear domain
- `Glob` for `<task-dir>/*/` to see which groups the project already uses, and prefer an
existing name over a synonym (`web` over `frontend`). A project with no group directories
yet is not a reason to skip grouping — it just means yours is the first.
- The group is a **subdirectory**, so it changes where you write the file:
`tasks/cli/007-fix-search-crash.md`, not `tasks/007-fix-search-crash.md`. Create the
directory if it doesn't exist, and don't also add a `group:` frontmatter field — the
directory name is what defines the group.
4. **Pick the task body shape**. Different kinds of task need different sections; a bug filed
with a generic "Objective" body loses the information that makes it actionable.
- Look for project templates with `Glob` on `.taskmd/templates/*.md`. If one matches the
kind of task being filed (a bug report → `bug.md`, a feature → `feature.md`, routine work
→ `chore.md`), `Read` it and use it as the basis for the new file:
- Drop the `_template:` block from the frontmatter — it describes the template, not the task
- Substitute `{{title}}`, `{{id}}`, and `{{date}}`
- Keep the template's field defaults (e.g. `type: bug`, `priority: high`) unless the user
specified otherwise — explicit user values always win
- Keep the template's section headings, and **fill every one of them in**
- If no template matches, choose the sections yourself:
- **Bug**: `## Steps to Reproduce` (numbered, concrete), `## Expected Behavior`,
`## Actual Behavior`, `## Environment`
- **Anything else**: `## Objective`, `## Tasks`, `## Acceptance Criteria`
Never leave placeholder content behind — no `<!-- ... -->` comments, no bare `TODO`, no
`1. ...`. Infer concrete content from the user's description; if a detail is genuinely
unknown (an exact version number, say), write what is known rather than a stub.
5. **Generate the task ID**:
- Read the ID strategy from `.taskmd.yaml` (default: `sequential`)
- Scan existing files with `Glob` for `<task-dir>/**/*.md` to determine used IDs —
this includes `archive/`, and it must: an archived task's ID is still taken and
must never be handed out again
- **Sequential** (default): Find the highest numeric ID, add 1, zero-pad to `padding` width (default 3). E.g., if highest is `042`, next is `043`
- **Prefixed**: Find highest number with the configured prefix. E.g., `dr-001`, `dr-002`
- **Random**: Generate a random alphanumeric string (lowercase letters and digits, i.e. base36) of configured `length` (default 6)
- **ULID**: Generate a ULID-like ID — use current timestamp in Crockford Base32 + random chars
6. **Create the task file** using `Write`:
- Path: `<task-dir>/<group>/<ID>-<slug-title>.md` — use the group decided in step 3, creating
the directory if needed. Write to `<task-dir>/<ID>-<slug-title>.md` only when that decision
was "no group"
- Slug: lowercase, hyphenated version of the title (max ~50 chars)
- Content:
```markdown
---
id: "<ID>"
title: "<title>"
status: pending
priority: <priority if provided>
effort: <effort if provided>
type: <type if provided>
tags: [<tags if provided>]
dependencies: [<deps if provided>]
parent: "<parent if provided>"
owner: "<owner if provided>"
phase: "<phase if provided>"
created_at: <today's date YYYY-MM-DD>
---
# <Title>
<the sections chosen in step 4, each filled in>
```
The default body, when no template applies and the task isn't a bug report:
```markdown
## Objective
<Description derived from user's input>
## Tasks
- [ ] <Subtask 1>
- [ ] <Subtask 2>
## Acceptance Criteria
- <Criterion derived from the task>
```
Only include optional frontmatter fields that were specified or can be inferred. Don't include empty fields.
7. **Confirm** the created file path and ID to the user
See `SPEC_REFERENCE.md` (in the plugin root) for valid field values, ID strategies, and frontmatter schema.
Versions
| Repository | Score | Licence |
|---|---|---|
| This page · driangle/taskmd | 100 | MIT |
| microbus-io/fabric | 100 | Apache-2.0 |
| driangle/taskmd | 100 | MIT |
| AdrianV101/obsidian-pkm-plugin | 98 | MIT |
| tbeack/fsad_playbook | 84 | MIT |
Database snapshot: 13.11.2026 · package version 765fa4c77745 (SHA-256 digest of the entire package)
Downloads in the last 30 days: 0 · ranking
Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.
Similar skills
- 81task_create
Create exactly one new task file for one upcoming work item. Use when the user asks to make, add, file, write, or capture a single task or todo. Use task for listing, querying, updating, archiving, linting, or multi task backlog work. - 100list-tasks
List tasks from the project's taskmd files, with optional filters (status, priority, phase, group, owner). Use whenever the user asks what is pending, in progress, done, outstanding, assigned, high priority, or "on the plate" — including phrasings like "my tasks", "my todos" or… - 100list-tasks
List tasks from the project's taskmd files, with optional filters (status, priority, phase, group, owner). Use whenever the user asks what is pending, in progress, done, outstanding, assigned, high priority, or "on the plate" — including phrasings like "my tasks", "my todos" or… - 91add-tool
Scaffold a new MCP tool definition. Use when the user asks to add a tool, create a new tool, or implement a new capability for the server. - 100task-prd-creator
Use when the user says I want to add X, implement Y, create a task for Z, or plan this feature. Writes a task file plus a PRD under .agent/, links them, and waits for approval before any implementation starts. - 91add-tool
Scaffold a new MCP tool definition. Use when the user asks to add a tool, create a new tool, or implement a new capability for the server.