New: 98,000+ skills indexed from GitHub repos with 100+ stars · every skill safety-checkedBrowse top skillsTop skills
Esc to closeBrowse all skills
Submit a skillBrowse skills
Submit a skillBrowse skills

Agent skill

Asw Start Work

by wjgoarxiv in wjgoarxiv/antigravity-swarm

Execute an ASW plan after asw-plan, with subagents, verification, real-surface QA, continuation state, and cleanup receipts.

MITAuto-check passedAgent Workflows

Install Asw Start Work

skills CLI
$ npx skills add wjgoarxiv/antigravity-swarm --skill asw-start-work -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install wjgoarxiv/antigravity-swarm asw-start-work --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/wjgoarxiv/antigravity-swarm.git skills-src && mkdir -p .claude/skills && cp -r skills-src/plugins/antigravity-swarm/skills/asw-start-work .claude/skills/asw-start-work && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
asw-start-work
GitHub stars
166
Token cost
~1.3k tokens
SKILL.md length
697 words
Files
1
Skills in repo
15
Repo updated
First seen
Licence
MIT

At a glance

Execute an ASW plan after asw-plan, with subagents, verification, real-surface QA, continuation state, and cleanup receipts.

  • Works in 8 steps: Select the named plan. If no name is… → Create or resume… → Re-read the selected plan before each… → …
  • Tasks that involve Subagents
  • SKILL.md covers Contract, Execution Rules, Resume Protocol and Evidence Ledger, plus 6 more sections
  • Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md

What it does

Asw Start Work is an agent skill from wjgoarxiv/antigravity-swarm. Execute an ASW plan after asw-plan, with subagents, verification, real-surface QA, continuation state, and cleanup receipts.

Its SKILL.md is about 1.3k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

It sits in Agent Workflows, covering Subagents. The repository describes itself as: ::A team of AI agents to code for you.::. The licence is MIT.

When your agent uses it

  • Tasks that involve Subagents

Example prompts

  • “/asw-start-work”

Workflow steps

8 steps, taken from the first numbered list in SKILL.md.

  1. Select the named plan. If no name is provided, inspect .asw/plans/ and choose the only active plan; ask one focused question only when…
  2. Create or resume .asw/start-work/state.json before editing. Track the selected plan path, active checkbox, spawned subagents, artifacts…
  3. Re-read the selected plan before each step.
  4. Execute the first unchecked top-level checkbox in ## TODOs or ## Final Verification Wave.
  5. Split independent work across Antigravity subagents. Keep dependent file edits serialized.
  6. For each checkbox, capture
  7. Mark the checkbox done only after the evidence exists.
  8. Continue to the next unchecked checkbox without asking whether to proceed.

What it can do on your machine

Read from SKILL.md and the folder at commit a949cb8. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    No scripts in the folder and no shell commands in SKILL.md (its code samples are json).

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Asw Start Work loads about 1.3k tokens when it runs. Until then it costs about 35 tokens; SKILL.md has 697 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~35
When it runs · the whole SKILL.md, loaded when a task matches
~1.3k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from wjgoarxiv/antigravity-swarm at commit a949cb8, republished under its MIT licence (© wjgoarxiv). 697 words, ~1,336 tokens.

Download SKILL.mdSave it as .claude/skills/asw-start-work/SKILL.md (or your agent's skills folder).
name
asw-start-work
description
Execute an ASW plan after asw-plan, with subagents, verification, real-surface QA, continuation state, and cleanup receipts.

Antigravity Swarm Start Work

Use this skill after asw-plan when the user says start-work, asw-start-work, execute the plan, continue the plan, or resume a plan under .asw/plans/.

Contract

  1. Select the named plan. If no name is provided, inspect .asw/plans/ and choose the only active plan; ask one focused question only when multiple plans match.
  2. Create or resume .asw/start-work/state.json before editing. Track the selected plan path, active checkbox, spawned subagents, artifacts, and cleanup items.
  3. Re-read the selected plan before each step.
  4. Execute the first unchecked top-level checkbox in ## TODOs or ## Final Verification Wave.
  5. Split independent work across Antigravity subagents. Keep dependent file edits serialized.
  6. For each checkbox, capture:
    • failing test or reproduction before production changes,
    • passing automated verification after the fix,
    • real-surface QA through tmux, browser, HTTP, or desktop automation,
    • applicable adversarial checks,
    • cleanup receipt for every spawned QA resource.
  7. Mark the checkbox done only after the evidence exists.
  8. Continue to the next unchecked checkbox without asking whether to proceed.

Execution Rules

  • If the selected checkbox is too broad, split it inside the plan before editing.
  • Keep one status ledger entry per checkbox: started, RED evidence, implementation, GREEN evidence, real-surface QA, cleanup, done.
  • Do not mark a checkbox complete until the evidence named in the plan exists.
  • If a command fails for environmental reasons, capture the exact failure and choose the nearest authoritative fallback.
  • If user instructions conflict with the plan, follow the newest user instruction and update the plan ledger.
  • If unrelated local changes exist, preserve them and work around them.

Resume Protocol

On start:

  1. Read the selected plan.
  2. Read .asw/start-work/state.json if present.
  3. Confirm the last completed checkbox.
  4. Confirm any in-progress checkbox.
  5. Verify artifacts listed in state still exist or mark them stale.
  6. Inspect git status.
  7. Continue from the first unchecked checkbox.

State shape:

json
{
  "plan": ".asw/plans/name.md",
  "activeTask": "Task 3",
  "completed": [],
  "artifacts": [],
  "cleanup": [],
  "blockers": []
}

Do not restart from scratch when state exists. Reconcile state with the plan.

Evidence Ledger

For each checkbox, append:

text
Task:
RED:
Implementation:
GREEN:
Manual QA:
Reviewer:
Cleanup:
Status:

If a plan item has no RED test because it is docs-only or inventory-only, record the exemption and the replacement check. Do not silently skip evidence.

Manual QA Channels

Run the channel named by the plan:

  • HTTP call for API behavior,
  • tmux for CLI, installer, hook, HUD, and package smoke,
  • browser for web UI,
  • computer use for desktop or IDE flows.

If the named channel is impossible in the environment, record:

  • why it is impossible,
  • nearest authoritative fallback,
  • what risk remains.

Do not substitute unit tests for manual QA.

Show full SKILL.md (280 more words)Show less

Reviewer Gate

Run a reviewer before completion when:

  • the plan's final verification wave asks for it,
  • 3 or more files changed,
  • package or public docs changed,
  • hooks, agents, skills, installer, or status line changed,
  • private/runtime exclusion is part of the goal.

Reviewer packet:

  • selected plan,
  • evidence ledger,
  • diff,
  • automated test outputs,
  • manual QA artifacts,
  • package/private scans,
  • cleanup receipts,
  • remaining risk.

Blockers must be fixed and rechecked before final status.

Cleanup Receipts

Every task must close its own resources:

  • temp dirs removed,
  • tmux sessions killed,
  • spawned servers stopped,
  • browser contexts closed,
  • generated scratch files deleted,
  • package tarballs removed,
  • config backups restored or documented.

Append one receipt per resource. A missing cleanup receipt keeps the task open.

Failure Handling

When a task fails:

  1. Capture the failing command and output.
  2. Identify whether the failure is code, test, environment, or plan quality.
  3. Fix code/test issues with the smallest change.
  4. Update the plan if the task was underspecified.
  5. Retry once with a clearer check.
  6. If the same blocker repeats, stop and report with evidence.

Do not mark the checkbox done because a fallback passed unless the fallback proves the same user-visible criterion.

Final Verification

Before printing completion:

  • all top-level checkboxes are marked complete,
  • evidence ledger has no empty RED/GREEN/manual QA rows except documented exemptions,
  • final verification wave ran,
  • package/private surfaces were checked when shipping changed,
  • reviewer gate passed when triggered,
  • cleanup list is closed,
  • git status is understood.

Completion

When every top-level checkbox is complete, run the plan's final verification commands and print:

text
ORCHESTRATION COMPLETE
Plan: <path>
Verification: <commands>
Artifacts: <paths>
Cleanup: <receipts>

Never treat green tests alone as completion. The plan is done only when the observable user-facing surface has been exercised and all continuation state is closed.

© wjgoarxiv, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in plugins/antigravity-swarm/skills/asw-start-work of wjgoarxiv/antigravity-swarm.

Open the folder on GitHubat commit a949cb8

Compare with similar skills

Asw Start Work next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Asw Start Work compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Asw Start Work this skillwjgoarxiv/antigravity-swarm166—~1.3kAutomated safety check: PassMIT
Claude Code Agent Developmentanthropics/claude-plugins-official38k7 repos~2.8kAutomated safety check: PassApache-2.0
Subagent Driven DevelopmentAsvarox/allkaraoke26138 repos~1.2kAutomated safety check: PassNone
Dispatching Parallel Agentsultralisp/ultralisp25841 repos~1.5kAutomated safety check: PassNone
Reflect on Session Learningscursor/plugins11k5 repos~1.2kAutomated safety check: PassNone
Paseo Advisor Second Opiniongetpaseo/paseo20k1 repos~756Automated safety check: PassCustom licence

Similar skills

  • Claude Code Agent Development

    anthropics/claude-plugins-official

    Official

    Explains how to write agents for Claude Code plugins: the markdown file with YAML frontmatter, trigger descriptions, model and color settings, and system prompt design.

    38k GitHub starsUsed in 7 repos~2.8k tokens
    Agent WorkflowsAuto-check passed

  • Subagent Driven Development

    Asvarox/allkaraoke

    A skill your agent uses when executing implementation plans with independent tasks in the current session

    261 GitHub starsUsed in 38 repos~1.2k tokens
    Agent WorkflowsAuto-check passed

  • Dispatching Parallel Agents

    ultralisp/ultralisp

    A skill your agent uses when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies

    258 GitHub starsUsed in 41 repos~1.5k tokens
    Agent WorkflowsAuto-check passed

  • Reflect on Session Learnings

    cursor/plugins

    Official

    Starts three parallel reviewer subagents over the current conversation transcript, then turns their findings into concrete edits to existing skills.

    11k GitHub starsUsed in 5 repos~1.2k tokens
    Agent WorkflowsAuto-check passed

  • Paseo Advisor Second Opinion

    getpaseo/paseo

    Launches one separate agent through Paseo to give a second opinion on the current task, with a self-contained briefing and no permission to edit files.

    20k GitHub starsUsed in 1 repo~756 tokens
    Agent WorkflowsAuto-check passed

  • Task Observer

    rebelytics/one-skill-to-rule-them-all

    Monitors task execution for skill improvement opportunities.

    3.2k GitHub starsUsed in 1 repo~11k tokens
    Agent WorkflowsAuto-check passed

More from wjgoarxiv/antigravity-swarm

All 15 skills in this repo

  • Asw

    wjgoarxiv/antigravity-swarm

    Antigravity Swarm execution loop for evidence-driven implementation with tests, subagents, hooks, and manual QA.

    166 GitHub stars~1.1k tokensUpdated 3 mo ago
    Auto-check passed

  • Asw Cleanup

    wjgoarxiv/antigravity-swarm

    Remove AI-looking clutter and temporary artifacts without changing behavior.

    166 GitHub stars~1.2k tokensUpdated 3 mo ago
    Auto-check passed

  • Asw Debug

    wjgoarxiv/antigravity-swarm

    Hypothesis-driven Antigravity Swarm debugging for crashes, hangs, wrong output, and runtime drift.

    166 GitHub stars~1k tokensUpdated 3 mo ago
    Auto-check passed

  • Asw Goal

    wjgoarxiv/antigravity-swarm

    Durable Antigravity Swarm goal orchestration with explicit success criteria, evidence ledger, manual QA channels, and completion audit.

    166 GitHub stars~905 tokensUpdated 3 mo ago
    Auto-check passed

  • Asw Loop

    wjgoarxiv/antigravity-swarm

    Antigravity Swarm Loop executes RED to GREEN to real-surface QA with cleanup receipts.

    166 GitHub stars~824 tokensUpdated 3 mo ago
    Auto-check passed

  • Asw Plan

    wjgoarxiv/antigravity-swarm

    Antigravity Swarm Plan creates a decision-complete plan before large or ambiguous work.

    166 GitHub stars~1.4k tokensUpdated 3 mo ago
    Auto-check passed

Categories

Questions about Asw Start Work

What does Asw Start Work do?

Execute an ASW plan after asw-plan, with subagents, verification, real-surface QA, continuation state, and cleanup receipts. Asw Start Work is an agent skill from wjgoarxiv/antigravity-swarm. Execute an ASW plan after asw-plan, with subagents, verification, real-surface QA, continuation state, and cleanup receipts.

When should I use Asw Start Work?

Asw Start Work fits situations like: tasks that involve Subagents.

How do I install Asw Start Work in Claude Code?

Run `npx skills add wjgoarxiv/antigravity-swarm --skill asw-start-work -a claude-code`. Or copy the skill folder (plugins/antigravity-swarm/skills/asw-start-work in wjgoarxiv/antigravity-swarm) into .claude/skills/asw-start-work in your project. Claude Code loads it when a task matches its description.

How do I install Asw Start Work in Codex?

Run `npx skills add wjgoarxiv/antigravity-swarm --skill asw-start-work -a codex`. Or copy the skill folder (plugins/antigravity-swarm/skills/asw-start-work in wjgoarxiv/antigravity-swarm) into .agents/skills/asw-start-work in your project. Codex loads it when a task matches its description.

Can I use Asw Start Work in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add wjgoarxiv/antigravity-swarm --skill asw-start-work -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/asw-start-work, .gemini/skills/asw-start-work, .github/skills/asw-start-work and .opencode/skills/asw-start-work in your project.

What does Asw Start Work need to run?

SKILL.md names no scripts, command-line tools or credentials: Asw Start Work is instructions for the agent only.

Does Asw Start Work access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Asw Start Work safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Asw Start Work use?

Asw Start Work is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Asw Start Work use?

About 1.3k tokens (SKILL.md is roughly 5.3k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Asw Start Work?

Skills that share tags, products or a category with Asw Start Work: Claude Code Agent Development (anthropics/claude-plugins-official, 38k stars), Subagent Driven Development (Asvarox/allkaraoke, 261 stars), Dispatching Parallel Agents (ultralisp/ultralisp, 258 stars) and Reflect on Session Learnings (cursor/plugins, 11k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Asw Start Work?

wjgoarxiv (a GitHub user) maintains it in wjgoarxiv/antigravity-swarm, which has 166 GitHub stars. The repository holds 15 skills in this directory. The repository was last updated on June 19, 2026.

Source: wjgoarxiv/antigravity-swarm on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.