New Skills — July 15, 2026¶

Evening sweep across 18 broad queries surfaced 6 new skills not previously catalogued in the Hermes docs. These span agent memory management, system service deployment, Web3 security analysis, LLM operations, and agent customization. Combined install base: 622+ installs.

Skills Discovered¶

Skill	Installs	Source	Category
memory-hygiene	365	aaaaqwq/claude-code-skills	Memory Management
llm-ops	70	sickn33/antigravity-awesome-skills	LLM Operations
linux-systemd	61	alphaonedev/openclaw-graph	Deployment
macos-launchd	56	alphaonedev/openclaw-graph	Deployment
openclaw-customizer	36	petekp/claude-code-setup	Setup / Customization
metamask-openclaw-desktop-security-analysis	34	aradotso/hermes-skills	Security

memory-hygiene¶

Source: aaaaqwq/claude-code-skills · 365 installs · Homepage: xdylanbaker/memory-hygiene

Audit, clean, and optimize Clawdbot/OpenClaw vector memory (LanceDB). Use when memory is bloated with junk, token usage is high from irrelevant auto-recalls, or setting up memory maintenance automation.

Capabilities¶

Audit: Check what's in vector memory with memory_recall query="*" limit=50
Wipe: Clear all vector memory by removing the LanceDB directory
Reseed: Store key facts after wipe using memory_store
Config: Disable auto-capture (the main source of memory junk) while keeping auto-recall active

Installation¶

npx skills add aaaaqwq/claude-code-skills/memory-hygiene

Hermes/CorpusIQ Relevance¶

Directly applicable to Hermes agent memory management. Prevents the "memory bloat → token waste → degraded performance" cycle that plagues long-running agents. The autoCapture: false + autoRecall: true pattern preserves useful recall without accumulating garbage.

Setup guide: memory-hygiene-setup.md

llm-ops¶

Source: sickn33/antigravity-awesome-skills · 70 installs

LLM Operations — RAG, embeddings, vector databases, fine-tuning, prompt engineering, LLM cost management, quality evals, and production AI architectures. Portuguese-language skill by renat.

Capabilities¶

RAG pipeline implementation
Embeddings creation (Pinecone, Chroma, pgvector)
Fine-tuning workflows
Prompt engineering
LLM cost reduction strategies
Quality evaluations
Semantic caching
Streaming and agent architectures

Installation¶

npx skills add sickn33/antigravity-awesome-skills/llm-ops

Hermes/CorpusIQ Relevance¶

Covers the full LLM ops stack that CorpusIQ agents depend on: model routing, cost tracking, embedding pipelines, and production deployment patterns. The Portuguese content is machine-translatable and the technical patterns are language-agnostic.

Setup guide: llm-ops-setup.md

linux-systemd¶

Source: alphaonedev/openclaw-graph · 61 installs

Deploy OpenClaw agents as Linux systemd services with proper supervision, logging, and restart policies. Part of the OpenClaw Graph ecosystem — a collection of deployment and infrastructure skills.

Capabilities¶

Systemd unit file generation for agent processes
Restart policy configuration (on-failure, always)
Journald log integration
Resource limits (MemoryMax, CPUQuota)
Environment variable injection
User-level vs system-level service choice

Installation¶

npx skills add alphaonedev/openclaw-graph/linux-systemd

Hermes/CorpusIQ Relevance¶

CorpusIQ runs Hermes agents on DGX Spark (Linux). This skill provides canonical patterns for systemd service management — directly applicable to the gateway service, cron workers, and long-running agent processes.

Setup guide: linux-systemd-setup.md

macos-launchd¶

Source: alphaonedev/openclaw-graph · 56 installs

Deploy OpenClaw agents as macOS launchd services. The macOS equivalent of linux-systemd — provides plist generation, KeepAlive configuration, and launchctl management for agent processes.

Capabilities¶

LaunchAgent plist generation
KeepAlive and RunAtLoad configuration
StandardOutPath / StandardErrorPath log routing
Environment variable dictionary
launchctl bootstrap/bootout management

Installation¶

npx skills add alphaonedev/openclaw-graph/macos-launchd

Hermes/CorpusIQ Relevance¶

The Mac Mini worker node runs Hermes processes. This skill enables proper launchd supervision — replacing fragile cron-only or manual-start approaches with OS-native process management.

Setup guide: macos-launchd-setup.md

openclaw-customizer¶

Source: petekp/claude-code-setup · 36 installs

Portable, version-controlled configuration for Claude Code and OpenAI Codex — fork, customize, sync across machines. Symlinks skill directories, scripts, and settings into the ~/.claude/ and ~/.codex/ directories.

Capabilities¶

Single-command setup via ./setup.sh
Symlink-based skill syncing across machines
Shared MCP configuration (.mcp.json)
Codex skill exclusion list
Git-based version control for agent configs

Installation¶

git clone https://github.com/petekp/claude-code-setup.git
cd claude-code-setup && ./setup.sh

Hermes/CorpusIQ Relevance¶

Useful pattern for teams running multiple Hermes/Claude instances across machines. The symlink approach to skill management is directly applicable to CorpusIQ's multi-machine agent deployment (DGX + Mac Mini).

Setup guide: openclaw-customizer-setup.md

metamask-openclaw-desktop-security-analysis¶

Source: aradotso/hermes-skills · 34 installs · Author: ara.so

⚠️ Security-focused skill. Analyzes the Metamask Openclaw desktop executable distribution for security risks. Identifies impersonation patterns, suspicious binaries, and cryptocurrency scam red flags. This is a defensive security tool — it helps agents evaluate suspicious Web3 desktop software before execution.

Capabilities¶

Impersonation detection (Metamask has no official "Openclaw" product)
Binary analysis of Windows .exe installers
Brand misuse identification
Web3 scam pattern recognition
Security recommendation generation

Installation¶

npx skills add aradotso/hermes-skills/metamask-openclaw-desktop-security-analysis

Hermes/CorpusIQ Relevance¶

As Hermes agents gain more autonomous execution capabilities, security analysis skills become critical. This skill exemplifies the pattern of using agent skills for defensive security — evaluating external tools before granting them execution access.

Setup guide: metamask-openclaw-security-analysis-setup.md

Discovered via skills.sh API sweep — 18 broad queries across hermes, openclaw, clawdbot, honcho, gbrain ecosystems. Cross-referenced against all existing docs at corpusiq-docs/hermes/skills/.