Skill Optimizer
Analysis
This instruction-only skill coherently audits local skills and waits for user choices before changes, but users should review any proposed edits or deletions because they affect installed skills.
Findings (2)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Checks for instructions or behavior that redirect the agent, misuse tools, execute unexpected code, cascade across systems, exploit user trust, or continue outside the intended task.
Only after the user selects action items should you: - rewrite metadata - sync `agents/openai.yaml` - merge overlapping skills - delete duplicates or obsolete skills
This gives the agent authority to change or remove installed skill files, but the same instruction makes those changes conditional on user-selected action items.
Checks for exposed credentials, poisoned memory or context, unclear communication boundaries, or sensitive data that could leave the user's control.
Unless the user explicitly provides extra logs or transcript files, use: - the current conversation history - the current workspace's local skill directories - installed skill directories that are directly visible from the environment
The audit relies on conversation context and local skill files as inputs. This is expected for the purpose, but users should understand what local/contextual data is being reviewed.