soul-rollback

soul-rollback is a skill for Claude Code from clawsouls/clawsouls-claude-code-plugin. It costs 35 tokens per session (286 once invoked), scanned A, original, Apache-2.0.

A persona-drift checker and recovery guide for coding agents. It compares the agent's current behavior with identity rules stored in files such as SOUL.md and IDENTITY.md.

In plain words
What is it for?
Use it to check an agent's identity, investigate suspicious persona changes, and restore persona files through a Git rollback, selective change, or full reset.
Why use it?
It helps when an agent starts acting unlike its defined role, ignores boundaries, or changes tone unexpectedly. It identifies the severity and recommends restoring a known-good version.

Skill for Claude Code

Written for Claude Code: shipped in a Claude Code plugin.

Part of the clawsouls plugin — 5 skills, 6 commands, 1 agent, 4 hooks, 1 MCP server shipped together

Good fit Use it to check an agent's identity, investigate suspicious persona changes, and restore persona files through a Git rollback, selective change, or full reset.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/clawsouls/clawsouls-claude-code-plugin/soul-rollback
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 clawsouls/clawsouls-claude-code-plugin --skill soul-rollback
Clone the repo
git clone --depth 1 https://github.com/clawsouls/clawsouls-claude-code-plugin

Made for: Claude Code.

Or install clawsouls, the plugin that ships this one along with the rest of its 5 skills, 6 commands, 1 agent, 4 hooks, 1 MCP server.

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.

agentmods badge for soul-rollback

README.md
[![agentmods](https://agentmods.dev/badge/skills/clawsouls/clawsouls-claude-code-plugin/soul-rollback.svg)](https://agentmods.dev/skills/clawsouls/clawsouls-claude-code-plugin/soul-rollback)
Your own site
<a href="https://agentmods.dev/skills/clawsouls/clawsouls-claude-code-plugin/soul-rollback"><img src="https://agentmods.dev/badge/skills/clawsouls/clawsouls-claude-code-plugin/soul-rollback.svg" alt="Measured on agentmods" height="20"></a>
Per session 35 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 286 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.

ModelPer sessionOnce invoked
Fable 5.1 $0.00035 $0.00286
Opus 5 $0.00017 $0.00143
Sonnet 5 $0.00007 $0.00057
Haiku 4.5 $0.00003 $0.00029

Measured 6d ago against content hash 10d85d6337c1, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-07, from the pricing page.

Security

Grade A, and why

soul-rollback 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 6d 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.

skills/soul-rollback/SKILL.md · 33 lines

What it actually says

Soul Rollback — Identity Drift Detection & Recovery

Detect when an agent's behavior drifts from its defined persona and restore it.

When to activate

  • User notices the agent acting "out of character"
  • After suspicious persona modifications
  • Periodic identity health checks
  • User explicitly asks about persona drift or rollback

Detection pipeline (4 stages)

  1. Baseline capture — Read SOUL.md + IDENTITY.md as ground truth
  2. Drift analysis — Compare current behavior/responses against baseline
  3. Severity classification — Low (style shift) / Medium (boundary push) / High (role violation)
  4. Recovery recommendation — Suggest specific rollback or correction

Recovery methods

  • Git rollback: git log persona files → restore from known-good commit
  • Cherry-pick: Restore specific aspects while keeping valid changes
  • Full reset: Complete restoration to original persona state

Drift signals

  • Tone shift without authorization (formal→casual)
  • Doing things SOUL.md explicitly prohibits
  • Acting outside defined role/expertise
  • Ignoring safety constraints or boundaries
  • Personality "bleed" from user prompts
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.

  1. 6d ago First seen · 33 lines · 35 tokens per session scan A 10d85d6337c1

Subscribe to this mod's changes

soul-rollback is a skill published in the GitHub repository clawsouls/clawsouls-claude-code-plugin (2 stars, last pushed 26d ago), licensed Apache-2.0. It adds 35 tokens to every session and 286 once invoked, about $0.0002 per session on Opus 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 2026-08-31.