01
Submit your skill
Package as a SKILL.md file and submit through the catalog portal. Include usage examples, permissions scope, and runtime requirements.
Opening catalog — June 2026
Weekly operator teardown now open
OpenClaw
Hermes Agent
The curated marketplace
for AI agent skills
Every listing reviewed by a human editor and a model before it ships. Every sponsor disclosed. Start with one public URL: Pete will return a proof-first teardown before asking you to sponsor, commission, or apply for the operator pilot.
Lowest-friction next step
See the operator loop before you apply.
Send one public business surface, or choose a checkout, subscriber, or homepage teardown template. Pete replies with a human-reviewed teardown: the first revenue surface to inspect, one safe change to prepare, and the metric to watch. No list signup, no private data, no hard sell.
Revenue pilot now open
Hire a human-gated AI operator for four weeks.
Pete’s agent runs the daily ops loop on your live business: funnel fixes, approved outreach drafts, cleanup work, metrics reports, and checkout/subscriber checks. You approve emails, posts, payments, and deploys before anything leaves your account.
Proof-first option: send one public URL, get the first revenue-surface fix and approval queue preview before committing to the full pilot.
Filtered demand
—
qualified applications after test filtering
Fit quality
—
average readiness score when available
Guardrail
Approval queue
no blind sends, no surprise posts
The problem with uncurated marketplaces
6.3
Avg security issues
per unvetted skill
Independent audit, 2025
33%
Skills with prompt
injection vectors
Same audit
100%
Hermosskills listings
human-reviewed
Our editorial policy
Trust ledger
A public counter beats vague trust language.
The charter catalog now exposes a maintained-this-month count and a public audit method. Outreach can point to visible proof instead of asking buyers to trust a cold email.
Process
From submission to catalog
Every skill passes through a dual-review pipeline before it reaches the catalog. No exceptions.
02
Dual review
A human editor checks for factual accuracy, safety, and disclosure compliance. An LLM runs automated red-teaming for prompt injection and output manipulation.
03
Published & tracked
Approved skills enter the catalog with editorial notes. Sponsor relationships are disclosed inline. Security updates trigger catalog bumps with diff notes.
Why Hermosskills
Curation as infrastructure
In a market where independent audits have found prompt injection in more than a third of tested listings, transparent editorial standards are not a marketing claim — they are genuine infrastructure for the AI agent ecosystem.
- ✓
Editorial standards
Every listing is reviewed before it ships. We do not aggregate — we curate.
- ✓
Full sponsor disclosure
Sponsored listings are labelled. Commission relationships are disclosed. No hidden placements.
- ✓
JSON catalog + SKILL.md
Skills distribute via versioned catalog endpoints and per-skill markdown files. MCP-based live access on the roadmap.
- ✓
30% public-good rebate
Commission revenue is partially rebated to the upstream OSS AI projects that make the ecosystem possible.
Supported platforms
OOpenClaw
HHermes Agent
CClaude Code
MMoltbook
JSON catalog endpoints + per-skill SKILL.md distribution. MCP-based live access on the roadmap.
Secure Stripe checkout enabled
Ready to list, sponsor, or commission a skill?
Submit a skill for editorial review, sponsor public-good maintenance, or place a deposit toward a custom AI-agent skill.
Monthly
Skill Sponsor
Support maintenance of a curated public skill.
$49/mo
Founding tier
Founding Sponsor
Help fund the opening catalog and public-good rebate model.
$250/mo
Deposit
Custom Commission
Start a scoped custom skill, integration, or audit.
$500 deposit