self-tidy

Grooms an agent's own memory namespace, peer awareness memos, and agent-card for daily drift.

2|Updated Apr 5, 2026
One-click install
npx skills add https://github.com/witwave-ai/witwave --skill self-tidy-witwave-ai
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: self-tidy
Source: https://github.com/witwave-ai/witwave/tree/main/.agents/self/iris/.claude/skills/self-tidy
Command: npx skills add https://github.com/witwave-ai/witwave --skill self-tidy-witwave-ai

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Autonomous agents accumulate stale memory files, dead index pointers, oversized log ledgers, and agent-cards that drift from their actual deployed skills. This Skill performs bounded daily self-maintenance so each agent's own state stays coherent without human intervention or cross-agent conflicts. ## Core Features & Use Cases - Memory consolidation: Reconciles the MEMORY.md index against files on disk, fixing orphan files, dead pointers, and description drift, and rotates append-only ledgers that exceed ~500 KB to avoid JSON buffer crashes. - Cross-agent awareness: Reads peer memory indexes read-only and refreshes per-peer reference memos summarizing what each teammate is working on. - Agent-card drift check: Verifies the canonical .witwave/agent-card.md matches current skills, trigger phrases, and avatar URLs, applying minor fixes inline and flagging substantive drift for human review. - Use Case: On a staggered daily cron, each of five agents (iris, kira, nova, evan, zora) runs the same byte-identical skill against only its own namespace, commits at most 50 lines of changes, delegates the push to iris, and logs the outcome. ## Quick Start Say "self-tidy" or "do your daily self-maintenance" to have the agent groom its own memory namespace and agent-card.

Frequently Asked Questions about self-tidy

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run daily self-maintenance for an autonomous agent?▼

Trigger the skill by saying "self-tidy" or let the daily cron fire once per agent per 24 hours. The agent reads its name from AGENT_NAME, grooms only its own memory namespace and agent-card, and commits at most 50 lines of changes.

How does self-tidy handle oversized agent memory log files?▼

Append-only ledgers exceeding roughly 500 KB are rotated: the file is copied to a timestamped archive and the live file is rewritten with its frontmatter plus the most recent 1500 lines. This prevents JSON buffer overflow errors when the backend reads the file.

Can self-tidy edit other agents' files or source code?▼

No. The skill is strictly scoped to the running agent's own memory namespace and agent-card. Peer namespaces are read-only, source code is never touched, and cross-cutting changes belong to the separate team-tidy skill.

What is the difference between self-tidy and team-tidy?▼

self-tidy is byte-identical across all agents and scoped to each agent's own files, while team-tidy runs only on zora and handles cross-cutting consistency like CLAUDE.md and SKILL.md edits. self-tidy logs cross-agent drift for team-tidy to act on.

What happens if a self-tidy commit breaks CI?▼

The agent first attempts a fix-forward follow-up commit pushed via iris. If CI remains red, the self-tidy commit is batch-reverted and the failure is logged in self_tidy_log.md with a reverted status and reason.