What is safe-agent-change?

A controlled workflow for changing a live Synthflow phone agent. It records the current version, saves a draft, shows the difference, waits for approval, publishes, verifies, and can roll back if verification fails.

Use it to edit an agent's prompt, greeting, voice, speech recognition, language model, or actions, then publish and verify the approved change.

It reduces the risk of accidentally changing unrelated settings or breaking an agent that already works. Drafts and approval keep unreviewed edits from going live.

safe-agent-change — Skill for Claude Code by SynthFlowAI

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Any agent

npx skills add SynthFlowAI/AnthropicPlugin --skill safe-agent-change

Clone the repo

git clone --depth 1 https://github.com/SynthFlowAI/AnthropicPlugin

Made for: Claude Code.

Or install synthflow, the plugin that ships this one along with the rest of its 7 skills, 2 MCP servers.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

README.md

[![agentmods](https://agentmods.dev/badge/skills/synthflowai/anthropicplugin/safe-agent-change/github.svg)](https://agentmods.dev/skills/synthflowai/anthropicplugin/safe-agent-change)

Your own site

<a href="https://agentmods.dev/skills/synthflowai/anthropicplugin/safe-agent-change"><img src="https://agentmods.dev/badge/skills/synthflowai/anthropicplugin/safe-agent-change/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

Your own site · 80×15

<a href="https://agentmods.dev/skills/synthflowai/anthropicplugin/safe-agent-change"><img src="https://agentmods.dev/badge/skills/synthflowai/anthropicplugin/safe-agent-change.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 133 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.

When invoked 2,131 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.

Origin original No closer match found in the catalogue.

Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

Model	Per session	Once invoked
Fable 5.1	$0.00133	$0.02131
Opus 5.5	$0.00053	$0.00852
Sonnet 5.5	$0.00027	$0.00426
Haiku 4.5	$0.00013	$0.00213

Measured 5d ago against content hash 036d8114fa21, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

safe-agent-change scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 5d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

plugins/synthflow/skills/safe-agent-change/SKILL.md · 129 lines

How it starts

The opening of the file, as written. The whole thing — 129 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Safe Agent Change

Purpose

Change a live agent without breaking what already works. Every edit goes through the same loop:

baseline → draft → diff → approval → publish → verify (→ rollback if verify fails)

Agents that skip steps in this loop are where most production incidents start: attached actions silently lost on republish, a prompt from one agent written over another, a failed update retried on top of a draft that had in fact been saved, or a draft published that changed more than the user asked for.

This playbook runs on the Synthflow MCP tools, whether they come from the plugin's bundled server or the official Synthflow connector.

Tools this uses

get_current_workspace, list_subaccounts — confirm which workspace the agent lives in.
list_agents, search_agents_by_prompt — find the agent when the user hasn't given a model_id.
get_agent — current settings and prompt.
get_agent_actions — the actions attached to the agent (the baseline you must preserve).
list_agent_versions — published versions and any existing draft.
update_agent — saves changes as a draft. Nothing goes live.
get_agent_version_diff — what the draft changes compared to live.
publish_agent_draft — makes the draft live.
rollback_agent_version — restores a previously published version.
list_voices — required before any voice change.

Some tool descriptions mention show_version_diff. That UI tool doesn't exist over MCP — show the diff to the user yourself, in the chat.

Step 0 — Pin the target

Workspace. If the agent may live in a subaccount, call list_subaccounts and pass that subaccount's workspace_id on every call below. Never guess or reuse a workspace_id from elsewhere — tools reject any workspace that isn't the current one or one of its subaccounts.
Agent. Resolve the exact model_id. If the user named an agent, use list_agents (or search_agents_by_prompt) and confirm the match when more than one agent could fit. Never guess.
Environment. If the user keeps separate dev/test and production agents, confirm which one this change is for. Default to the test agent when one exists and the user hasn't said otherwise.

Read the full file on GitHub · 129 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

5d ago First seen · 129 lines · 133 tokens per session scan A 036d8114fa21

Subscribe to this mod's changes

safe-agent-change is a skill published in the GitHub repository SynthFlowAI/AnthropicPlugin (0 stars, last pushed 5d ago), licensed MIT. It adds 133 tokens to every session and 2,131 once invoked, about $0.0005 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-11-09.

DevOps and cloud

safe-agent-change next to 6 skills from other repositories
Mod	Type	Stars	Last push	Licence	Scan	Tokens	Works with
safe-agent-changeSynthFlowAI/AnthropicPlugin	Skill	0	5d ago	MIT	A	133	Claude Code
imagegenopenai/codex	Skill	135k	today	Apache-2.0	A	113	Codex
playwright-climicrosoft/playwright	Skill	100k	today	Apache-2.0	A	19	Claude Code
browser-automationopenclaw/openclaw	Skill	394k	today	MIT	A	36	Claude Code
local-ai-agentsmicrosoft/ai-agents-for-beginners	Skill	80k	8d ago	MIT	A	200	Claude Code, Codex
next-cache-components-adoptionvercel/next.js	Skill	144k	today	MIT	A	95	Claude Code
chat-pet-sprite-creationmicrosoft/vscode	Skill	195k	today	MIT	A	53	Claude Code, Codex