Global Rank · of 601 Skills
agent-decommissioning AI Agent Skill
View Source: b-open-io/prompts
CriticalInstallation
npx skills add b-open-io/prompts --skill agent-decommissioning 9
Installs
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.
Confirm retirement vs. update — Is this agent being permanently retired, or does it need an update/rename? If unsure, ask before proceeding.
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" -lCheck 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.mdCheck skills-lock.json:
grep "<agent-name>" ~/code/prompts/skills-lock.json 2>/dev/nullDetermine 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:
Identify the bot:
clawnet bot listLocate the bot by name or associated agent identity.
Stop the sandbox:
clawnet bot stop <bot-name>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.
Remove from ClawNet bot registry:
clawnet bot delete <bot-name>Confirm deletion with
clawnet bot list— the bot should no longer appear.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).
Remove the agent file:
rm ~/code/prompts/agents/<agent-name>.mdRemove the agent's avatar (if one exists):
ls ~/code/prompts/agents/avatars/ rm ~/code/prompts/agents/avatars/<agent-name>.*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.jsonEdit to remove the agent reference, then bump the plugin version (patch bump for removal).
Check skills-lock.json for any references to the agent and remove them.
Search for residual references across the plugin repo:
grep -r "<agent-name>" ~/code/prompts --include="*.md" --include="*.json" -lFix or remove any remaining references.
Phase 4: Update Roster and Routing
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.Update bOpen.ai roster — If there is a roster file or UI that lists team members, remove the retired agent's entry.
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'
## Skillssections - Dispatch instructions like "use
<agent-name>for X" - README or documentation files
- Other agents'
Phase 5: Publish
Stage all changes:
cd ~/code/prompts git status git add -AReview the diff — Confirm only the expected files are changed. No unintended modifications.
Commit:
git commit -m "Remove <agent-name> agent (decommissioned)"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.
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 inclaude agent list.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.
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.
Confirm no orphaned sandbox — Johnny verifies
clawnet bot listno 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/clawnetand~/code/clawnet-bot
Installs
Security Audit
View Source
b-open-io/prompts
More from this source
Power your AI Agents with
the best open-source models.
Drop-in OpenAI-compatible API. No data leaves Europe.
Explore Inference APIGLM
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
How to use this skill
Install agent-decommissioning by running npx skills add b-open-io/prompts --skill agent-decommissioning in your project directory. Run the install command above in your project directory. The skill file will be downloaded from GitHub and placed in your project.
No configuration needed. Your AI agent (Claude Code, Cursor, Windsurf, etc.) automatically detects installed skills and uses them as context when generating code.
The skill enhances your agent's understanding of agent-decommissioning, helping it follow established patterns, avoid common mistakes, and produce production-ready output.
What you get
Skills are plain-text instruction files — not executable code. They encode expert knowledge about frameworks, languages, or tools that your AI agent reads to improve its output. This means zero runtime overhead, no dependency conflicts, and full transparency: you can read and review every instruction before installing.
Compatibility
This skill works with any AI coding agent that supports the skills.sh format, including Claude Code (Anthropic), Cursor, Windsurf, Cline, Aider, and other tools that read project-level context files. Skills are framework-agnostic at the transport level — the content inside determines which language or framework it applies to.
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.