bladnman/do-work

Overview

This skill provides a unified entry point for capturing tasks or processing a queued backlog. It routes free-form input into either task creation or queue execution, preserving the original payload until the user confirms the chosen action. The design supports a two-phase flow: capture intent and then act, so no user content is ever lost.

How this skill works

The skill parses the text that follows the trigger phrase to decide between two actions: 'do' (capture a request) and 'work' (process pending items). Single-word action verbs like run, start, or go route immediately to processing; descriptive text or explicit add/create cues route to task capture. If routing is ambiguous but content exists, the skill defaults to capturing the payload and prompts a simple confirmation question, accepting minimal replies like 'add' or 'work' to proceed.

When to use it

• Capture feature requests, bug reports, meeting notes, or any descriptive task content.
• Start or resume automated processing of the pending task queue with a verb command.
• Quickly log context-rich payloads such as screenshots, long descriptions, or user stories without losing data.
• When you want a safe two-step flow: stash content first, then confirm execution later.

Best practices

• Use single-word verbs (run, start, go) to immediately trigger queue processing.
• Provide descriptive text or include 'add'/'create' to ensure a new task is captured.
• Rely on the confirmation prompt when unsure—reply with 'add' or 'work' to act on stored payload.
• Always assume the skill preserves your full input; you will not need to retype content after clarification.
• Keep task descriptions concise but include necessary context or attachments to speed processing.

Example use cases

• do work add dark mode — capture a feature request with context for later triage.
• do work the login button is broken — log a bug report containing screenshots or steps.
• do work run — immediately begin processing the queued requests.
• do work — get a confirmation prompt: 'Start the work loop?' before executing.
• do work [500 words of meeting notes] — store long-form context as a request file for follow-up.