#601

Globales Ranking · von 601 Skills

agent-decommissioning AI Agent Skill

Quellcode ansehen: b-open-io/prompts

Critical

Installation

npx skills add b-open-io/prompts --skill agent-decommissioning

9

Installationen

Agent Decommissioning

Safely retire a bOpen team agent — removing it from the plugin, ClawNet, routing, and all references. This is a coordinated two-specialist workflow.

Satchmo (agent-builder) owns the plugin/code side.
Johnny (clawnet-mechanic) owns ClawNet infrastructure teardown.

Neither should proceed without the other completing their side.


Phase 1: Assessment

Before touching anything, confirm the intent and scope.

  1. Confirm retirement vs. update — Is this agent being permanently retired, or does it need an update/rename? If unsure, ask before proceeding.

  2. Check for dependents — Search for references to the agent across the plugin repo:

    cd ~/code/prompts
    grep -r "<agent-name>" agents/ skills/ .claude-plugin/ --include="*.md" --include="*.json" -l
  3. Check Martha's routing rules — Read the front-desk agent to find any routing logic that dispatches to this agent:

    grep -i "<agent-name>" ~/code/prompts/agents/front-desk.md
  4. Check skills-lock.json:

    grep "<agent-name>" ~/code/prompts/skills-lock.json 2>/dev/null
  5. Determine if the agent has a live ClawNet bot — Run clawnet bot list (or ask Johnny) to check. If a bot exists, Johnny must handle teardown before Satchmo removes the plugin file.

Document findings before proceeding. If other agents or skills reference the retiring agent, plan those updates as part of this workflow.


Phase 2: ClawNet Bot Teardown (Johnny's Domain)

If the agent has a live ClawNet bot deployment, delegate to Johnny (clawnet-bot:clawnet-mechanic) for full infrastructure teardown. Do not skip this phase if a bot exists — orphaned sandboxes waste resources.

Invoke: Skill(clawnet:clawnet-cli) before issuing any ClawNet commands.

Johnny's teardown steps:

  1. Identify the bot:

    clawnet bot list

    Locate the bot by name or associated agent identity.

  2. Stop the sandbox:

    clawnet bot stop <bot-name>
  3. Archive or remove the BAP identity — The bot has a Bitcoin Attestation Protocol identity. Archive it if the identity history matters; remove it if a clean break is preferred. Johnny knows the right call here based on the bot's history.

  4. Remove from ClawNet bot registry:

    clawnet bot delete <bot-name>

    Confirm deletion with clawnet bot list — the bot should no longer appear.

  5. Signal completion to Satchmo — Once the sandbox is stopped and bot removed, Johnny confirms so Satchmo can proceed with plugin removal.

Key repos for Johnny: ~/code/clawnet and ~/code/clawnet-bot.


Phase 3: Plugin Removal (Satchmo's Domain)

Proceed only after Phase 2 is complete (or confirmed that no bot exists).

Working directory: ~/code/prompts (bopen-tools plugin source).

  1. Remove the agent file:

    rm ~/code/prompts/agents/<agent-name>.md
  2. Remove the agent's avatar (if one exists):

    ls ~/code/prompts/agents/avatars/
    rm ~/code/prompts/agents/avatars/<agent-name>.*
  3. Update .claude-plugin/plugin.json — If the agent was explicitly listed in the plugin manifest, remove its entry. Read the file first:

    cat ~/code/prompts/.claude-plugin/plugin.json

    Edit to remove the agent reference, then bump the plugin version (patch bump for removal).

  4. Check skills-lock.json for any references to the agent and remove them.

  5. Search for residual references across the plugin repo:

    grep -r "<agent-name>" ~/code/prompts --include="*.md" --include="*.json" -l

    Fix or remove any remaining references.


Phase 4: Update Roster and Routing

  1. Update Martha (front-desk) — Remove the routing rule that dispatches to the retired agent. Read ~/code/prompts/agents/front-desk.md, find the routing entry, and delete it. Martha must not route to an agent that no longer exists.

  2. Update bOpen.ai roster — If there is a roster file or UI that lists team members, remove the retired agent's entry.

  3. Update any agents whose prompts reference the retired agent — From the Phase 1 grep results, edit each file that mentions the retiring agent and remove or update the reference. Common locations:

    • Other agents' ## Skills sections
    • Dispatch instructions like "use <agent-name> for X"
    • README or documentation files

Phase 5: Publish

  1. Stage all changes:

    cd ~/code/prompts
    git status
    git add -A
  2. Review the diff — Confirm only the expected files are changed. No unintended modifications.

  3. Commit:

    git commit -m "Remove <agent-name> agent (decommissioned)"
  4. Push to master — Pushing to master IS publishing. The marketplace picks up the latest commit automatically:

    git push origin master

Phase 6: Verify

After publishing, confirm the decommissioning is clean.

  1. Confirm the agent no longer appears in the plugin — After users run claude plugin update bopen-tools@b-open-io, the agent should not appear in claude agent list.

  2. Verify Martha no longer routes to the retired agent — Test with a query that would previously have been routed to the retired agent. Martha should either return "not available" or route elsewhere.

  3. Check for broken references — Search the updated repo one more time:

    grep -r "<agent-name>" ~/code/prompts --include="*.md" --include="*.json"

    This should return no results.

  4. Confirm no orphaned sandbox — Johnny verifies clawnet bot list no longer shows the removed bot.


Responsibility Summary

Task Owner
Confirm retirement intent Satchmo
Dependency audit Satchmo
ClawNet bot stop Johnny
BAP identity teardown Johnny
Bot registry removal Johnny
Agent .md file removal Satchmo
Avatar removal Satchmo
Plugin manifest update Satchmo
Martha routing update Satchmo
Roster update Satchmo
Cross-agent reference cleanup Satchmo
Publish (git push) Satchmo
Post-publish verification Both

Key Paths

  • Plugin repo: ~/code/prompts
  • Agent files: ~/code/prompts/agents/
  • Agent avatars: ~/code/prompts/agents/avatars/
  • Plugin manifest: ~/code/prompts/.claude-plugin/plugin.json
  • Front-desk (Martha): ~/code/prompts/agents/front-desk.md
  • ClawNet repos: ~/code/clawnet and ~/code/clawnet-bot

Installationen

Installationen 9
Globales Ranking #601 von 601

Sicherheitsprüfung

ath Safe
socket Critical
Warnungen: 1 Bewertung: 90
snyk Low
EU EU-Hosted Inference API

Power your AI Agents with the best open-source models.

Drop-in OpenAI-compatible API. No data leaves Europe.

Explore Inference API

GLM

GLM 5

$1.00 / $3.20

per M tokens

Kimi

Kimi K2.5

$0.60 / $2.80

per M tokens

MiniMax

MiniMax M2.5

$0.30 / $1.20

per M tokens

Qwen

Qwen3.5 122B

$0.40 / $3.00

per M tokens

So verwenden Sie diesen Skill

1

Install agent-decommissioning by running npx skills add b-open-io/prompts --skill agent-decommissioning in your project directory. Führen Sie den obigen Installationsbefehl in Ihrem Projektverzeichnis aus. Die Skill-Datei wird von GitHub heruntergeladen und in Ihrem Projekt platziert.

2

Keine Konfiguration erforderlich. Ihr KI-Agent (Claude Code, Cursor, Windsurf usw.) erkennt installierte Skills automatisch und nutzt sie als Kontext bei der Code-Generierung.

3

Der Skill verbessert das Verständnis Ihres Agenten für agent-decommissioning, und hilft ihm, etablierte Muster zu befolgen, häufige Fehler zu vermeiden und produktionsreifen Code zu erzeugen.

Was Sie erhalten

Skills sind Klartext-Anweisungsdateien — kein ausführbarer Code. Sie kodieren Expertenwissen über Frameworks, Sprachen oder Tools, das Ihr KI-Agent liest, um seine Ausgabe zu verbessern. Das bedeutet null Laufzeit-Overhead, keine Abhängigkeitskonflikte und volle Transparenz: Sie können jede Anweisung vor der Installation lesen und prüfen.

Kompatibilität

Dieser Skill funktioniert mit jedem KI-Coding-Agenten, der das skills.sh-Format unterstützt, einschließlich Claude Code (Anthropic), Cursor, Windsurf, Cline, Aider und anderen Tools, die projektbezogene Kontextdateien lesen. Skills sind auf Transportebene framework-agnostisch — der Inhalt bestimmt, für welche Sprache oder welches Framework er gilt.

Data sourced from the skills.sh registry and GitHub. Install counts and security audits are updated regularly.

EU Made in Europe

Chat with 100+ AI Models in one App.

Use Claude, ChatGPT, Gemini alongside with EU-Hosted Models like Deepseek, GLM-5, Kimi K2.5 and many more.

App herunterladen:

Kundensupport