error-coordinator
ReviewAudited by ClawScan on May 10, 2026.
Overview
This instruction-only skill matches its error-handling purpose, but it should be reviewed because it asks the agent to automate operational recovery changes and may report unsupported success metrics.
Before installing, treat this as a powerful operations prompt: restrict it to approved systems, require confirmation before deployment, rollback, data reconciliation, or recovery actions, and ensure any reported reliability metrics come from real telemetry rather than the template text.
Findings (5)
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.
If connected to deployment or operations tools, the agent could make broad system changes or trigger automated responses without enough user review.
These are operational changes and automated actions, but the provided artifact does not define approval gates, target scope, dry-run behavior, or change-control limits.
Implementation approach: - Deploy error collectors - Configure correlation - Implement circuit breakers - Setup recovery flows - ... - Automate responses
Require explicit user approval for changes, define allowed systems and environments, use dry-runs where possible, and follow normal incident/change-management procedures.
A mistaken automated recovery step could affect multiple services, data state, or dependent agents.
Automated recovery, rollback, restoration, and reconciliation across systems can propagate an incorrect diagnosis or action if not contained; the artifact does not show clear containment boundaries.
Recovery orchestration: - Automated recovery flows - Rollback procedures - State restoration - Data reconciliation - Service restoration
Limit automated recovery to pre-approved runbooks, stage changes, add health checks and circuit breakers around the recovery process itself, and require manual confirmation for destructive or cross-service actions.
The agent could overstate reliability improvements or present fabricated operational metrics as facts.
The prompt supplies concrete reliability metrics in a delivery message without requiring that they be measured or sourced.
Delivery notification: "Error coordination established. Handling 3421 errors/day with 93% automatic recovery rate. Prevented 47 cascade failures and reduced MTTR to 4.2 minutes..."
Require all reported metrics to be calculated from real telemetry, label examples clearly, and avoid absolute claims such as complete cascade prevention unless independently verified.
Sensitive infrastructure details or inaccurate historical context could be reused in future troubleshooting or recovery decisions.
System topology and error history can be sensitive operational context and may influence later decisions if stored or reused.
When invoked: 1. Query context manager for system topology and error patterns
Restrict what topology and incident data is placed in shared context, validate retrieved context before acting on it, and avoid storing secrets or sensitive logs.
Operational details may be shared between agents without clearly defined boundaries.
The skill expects collaboration with other agents, but the visible artifact does not specify identity checks, permission boundaries, or data-sharing limits.
Integration with other agents: - Work with performance-monitor on detection - Collaborate with workflow-orchestr
Use only trusted agents, share the minimum necessary context, and confirm what data each cooperating agent can read or change.