start-task — for Claude Code start-task, Glass-Atlas, community, for Claude Code, ide skills, CLAUDE.md, query-workboard, todo -> in_progress, docs[], files[]

v1.0.0

À propos de ce Skill

Scenario recommande : Ideal for AI agents that need use this skill to complete exactly one task from docs/workboard.json. Resume localise : Blog/editorial site with RAG chat over personal knowledge notes # Start Task Use this skill to complete exactly one task from docs/workboard.json. This AI agent skill supports Claude Code, Cursor, and Windsurf workflows.

Fonctionnalités

Use this skill to complete exactly one task from docs/workboard.json.
Read the repo instruction dispatcher first (CLAUDE.md).
Select next startable task using the same startability rules as /query-workboard.
If no startable task exists, report blocked items and stop.
Load only the chosen task record.

# Sujets clés

AG9898 AG9898
[0]
[0]
Mis à jour: 5/1/2026

Skill Overview

Start with fit, limitations, and setup before diving into the repository.

Scenario recommande : Ideal for AI agents that need use this skill to complete exactly one task from docs/workboard.json. Resume localise : Blog/editorial site with RAG chat over personal knowledge notes # Start Task Use this skill to complete exactly one task from docs/workboard.json. This AI agent skill supports Claude Code, Cursor, and Windsurf workflows.

Pourquoi utiliser cette compétence

Recommandation : start-task helps agents use this skill to complete exactly one task from docs/workboard.json. Blog/editorial site with RAG chat over personal knowledge notes # Start Task Use this skill to complete

Meilleur pour

Scenario recommande : Ideal for AI agents that need use this skill to complete exactly one task from docs/workboard.json.

Cas d'utilisation exploitables for start-task

Cas d'usage : Applying Use this skill to complete exactly one task from docs/workboard.json
Cas d'usage : Applying Read the repo instruction dispatcher first (CLAUDE.md)
Cas d'usage : Applying Select next startable task using the same startability rules as /query-workboard

! Sécurité et Limitations

  • Limitation : Load only the chosen task record.
  • Limitation : Implement only what description + acceptance criteria require.
  • Limitation : Mark task done only after verification passes.

About The Source

The section below comes from the upstream repository. Use it as supporting material alongside the fit, use-case, and installation summary on this page.

Démo Labs

Browser Sandbox Environment

⚡️ Ready to unleash?

Experience this Agent in a zero-setup browser environment powered by WebContainers. No installation required.

Boot Container Sandbox

FAQ et étapes d’installation

These questions and steps mirror the structured data on this page for better search understanding.

? Questions fréquentes

Qu’est-ce que start-task ?

Scenario recommande : Ideal for AI agents that need use this skill to complete exactly one task from docs/workboard.json. Resume localise : Blog/editorial site with RAG chat over personal knowledge notes # Start Task Use this skill to complete exactly one task from docs/workboard.json. This AI agent skill supports Claude Code, Cursor, and Windsurf workflows.

Comment installer start-task ?

Exécutez la commande : npx killer-skills add AG9898/Glass-Atlas. Elle fonctionne avec Cursor, Windsurf, VS Code, Claude Code et plus de 19 autres IDE.

Quels sont les cas d’usage de start-task ?

Les principaux cas d’usage incluent : Cas d'usage : Applying Use this skill to complete exactly one task from docs/workboard.json, Cas d'usage : Applying Read the repo instruction dispatcher first (CLAUDE.md), Cas d'usage : Applying Select next startable task using the same startability rules as /query-workboard.

Quels IDE sont compatibles avec start-task ?

Cette skill est compatible avec Cursor, Windsurf, VS Code, Trae, Claude Code, OpenClaw, Aider, Codex, OpenCode, Goose, Cline, Roo Code, Kiro, Augment Code, Continue, GitHub Copilot, Sourcegraph Cody, and Amazon Q Developer. Utilisez la CLI Killer-Skills pour une installation unifiée.

Y a-t-il des limites pour start-task ?

Limitation : Load only the chosen task record.. Limitation : Implement only what description + acceptance criteria require.. Limitation : Mark task done only after verification passes..

Comment installer ce skill

  1. 1. Ouvrir le terminal

    Ouvrez le terminal ou la ligne de commande dans le dossier du projet.

  2. 2. Lancer la commande d’installation

    Exécutez : npx killer-skills add AG9898/Glass-Atlas. La CLI détectera automatiquement votre IDE ou votre agent et configurera la skill.

  3. 3. Commencer à utiliser le skill

    Le skill est maintenant actif. Votre agent IA peut utiliser start-task immédiatement dans le projet.

! Source Notes

This page is still useful for installation and source reference. Before using it, compare the fit, limitations, and upstream repository notes above.

Upstream Repository Material

The section below comes from the upstream repository. Use it as supporting material alongside the fit, use-case, and installation summary on this page.

Upstream Source

start-task

Install start-task, an AI agent skill for AI agent workflows and automation. Explore features, use cases, limitations, and setup guidance.

SKILL.md
Readonly
Upstream Repository Material
The section below comes from the upstream repository. Use it as supporting material alongside the fit, use-case, and installation summary on this page.
Upstream Source

Start Task

Use this skill to complete exactly one task from docs/workboard.json.

Workflow

  1. Read the repo instruction dispatcher first (CLAUDE.md).
  2. Select next startable task using the same startability rules as /query-workboard.
  3. If no startable task exists, report blocked items and stop.
  4. Load only the chosen task record.
  5. Claim the task by updating status todo -> in_progress with a targeted edit.
  6. Read every existing file in task docs[] and files[] before editing code.
  7. Implement only what description + acceptance_criteria require.
  8. Run task commands[] as preferred verification.
  9. Add minimum extra validation needed for changed surface if commands are absent/incomplete.
  10. Update authoritative docs affected by behavior changes.
  11. Mark task done only after verification passes.
  12. Commit the completed task after all verification passes and the board is updated.
  13. Summarize: task, commit, files changed, validations, docs updated, next startable tasks.

Startability Query

bash
1jq ' 2 (.tasks | map({(.id): .}) | add) as $byId | 3 [.tasks[] | 4 select(.status == "todo") | 5 select((.blocked_by // []) | length == 0) | 6 select((.depends_on // []) | map($byId[.].status == "done") | all) 7 ] | 8 sort_by(if .priority == "critical" then 0 elif .priority == "high" then 1 elif .priority == "medium" then 2 else 3 end) | 9 .[0] | {id, title, priority, group_id, depends_on, blocked_by} 10' docs/workboard.json

Targeted Board Edit Rules

  • Never rewrite full docs/workboard.json.
  • Never mutate tasks other than selected task.
  • Use precise patch/edit around selected task block only.
  • If blocked mid-task and unresolved, set selected task back to todo and stop cleanly.

Verification Rules

  • Run every command in task commands[] unless impossible in current environment; if skipped, state why.
  • Do not claim completion with failing checks.
  • Prefer targeted tests/build checks over full-suite runs unless task requires full regression.

Documentation Concurrency

  • Update each doc listed in task docs[] when truth changed.
  • Also update canonical repo docs impacted by API/schema/architecture/UX/test changes.
  • Do not append active execution notes to archive-only logs.

Guardrails

  • One task per run.
  • Do not bypass dependency checks.
  • Do not invent deprecated board fields from older repo variants.
  • Do not mark done before checks pass.
  • Do not push; stop after creating the local commit.

Commit Phase

  • Create one local git commit for the completed task after verification passes and docs/workboard.json is updated.
  • Include only the selected task's related code, docs, tests, and targeted board status changes.
  • Do not push the branch or publish the commit.

Compétences associées

Looking for an alternative to start-task or another community skill for your workflow? Explore these related open-source skills.

Voir tout

openclaw-release-maintainer

Logo of openclaw
openclaw

Resume localise : 🦞 # OpenClaw Release Maintainer Use this skill for release and publish-time workflow. It covers ai, assistant, crustacean workflows. This AI agent skill supports Claude Code, Cursor, and Windsurf workflows.

widget-generator

Logo of f
f

Resume localise : Generate customizable widget plugins for the prompts.chat feed system # Widget Generator Skill This skill guides creation of widget plugins for prompts.chat . It covers ai, artificial-intelligence, awesome-list workflows. This AI agent skill supports Claude Code, Cursor, and

flags

Logo of vercel
vercel

Resume localise : The React Framework # Feature Flags Use this skill when adding or changing framework feature flags in Next.js internals. It covers blog, browser, compiler workflows. This AI agent skill supports Claude Code, Cursor, and Windsurf workflows.

138.4k
0
Navigateur

pr-review

Logo of pytorch
pytorch

Resume localise : Usage Modes No Argument If the user invokes /pr-review with no arguments, do not perform a review . It covers autograd, deep-learning, gpu workflows. This AI agent skill supports Claude Code, Cursor, and Windsurf workflows.

98.6k
0
Développeur