Audit and prune agent context files for token efficiency.
npx skills add https://github.com/TylerSimons1127/vibe --skill context-doctor-tylersimons1127
Please help me install this Agent Skill. Skill: context-doctor Source: https://github.com/TylerSimons1127/vibe/tree/main/skills/context-doctor Command: npx skills add https://github.com/TylerSimons1127/vibe --skill context-doctor-tylersimons1127
High-intent search queries and answers about installing and using this skill.
Run the context doctor to audit everything loaded each turn, including SOUL.md, standing memory, and skill descriptions. It flags descriptions over 300 characters, stale skills, and tool-schema bloat, then trims, merges, or quarantines offenders.
The context doctor identifies skills never invoked in recent sessions and overlapping skills covering the same ground. It can merge duplicates into the stronger skill or quarantine stale ones in ~/.hermes/quarantine/ for inspection.
No, it never hard-deletes skills. Problematic skills are quarantined to ~/.hermes/quarantine/ so you can inspect them before deciding on permanent removal, keeping the process reversible.
Skill descriptions should stay under 300 characters because every description rides along in the prompt each turn. Longer descriptions are flagged by the audit and trimmed to reduce token bloat.
Yes, the confidential disclosure check examines whether any skill might leak secrets or PII in its output or description. Flagged skills can be quarantined or trimmed to remove the disclosure risk.
agentscope-aiSynchronize Hermes Worker configs, skills, and shared files from MinIO.
PotLockValidate and guide AgentSkill creation, modification, and auditing workflows.
SamuelSahaReduce AI token consumption via shallow skill loading, context pruning, and output compression.
giovannimnzIndex GSD Hermes skills with names and descriptions via skill_view.
