Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
What This Skill Does
Traces the inheritance chain of AI agent skills to detect cumulative capability drift. A skill audited as safe in generation 1 may have gained network access or weakened validation by generation 5 through incremental, individually plausible changes. Produces a drift score and mutation classification to flag lineages where the original audit no longer covers current behavior.
Analyzing only the current version misses how capabilities accumulated across forks, which is exactly the gap that makes inherited skills risky.
When to use it
- Auditing a forked skill before deploying it to a production agent
- Checking if a marketplace skill's 'verified' badge still reflects current behavior
- Identifying which generation in a lineage introduced network access
- Reviewing inherited skills in a multi-agent pipeline for capability creep
- Prioritizing which skills in a large library need re-review
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: Check drift for the `data-sanitizer` skill currently at generation 5
- 1Trace the skill's inheritance chain back to the original audited version in generation 1
- 2Compute a structured diff for each generation, noting new capabilities, changed permissions, and removed checks
- 3Classify each mutation as cosmetic, functional, capability-expanding, or safety-reducing
- 4Aggregate diffs into a single capability drift score from 0 to 100
- 5Output a full drift report with lineage tree, per-generation changes, and a re-audit verdict
Drift report with a score of 78/100 and a RE-AUDIT RECOMMENDED verdict, identifying that network access and remote schema fetching were added after the original audit and one input validation check was removed