CAIN-42 MCPGateAGENTIC AI 75 · #17
Agent skills as a portable ecosystem standard
Anthropic announced broader platform adoption and foundation stewardship of Agent Skills.
Last reviewed 2026-11-11
Agentic AI 75 / #17
2026 status: Ecosystem standardisation · Category: Agent interoperability
The 2026 development
Anthropic announced broader platform adoption and foundation stewardship of Agent Skills.
The governance gap
A portable skill is executable content that crosses vendor boundaries. The standard makes reuse easy; it also makes a malicious or over-privileged skill portable.
Where CAIN-42 fits: Moat 2 — Security-Context Continuity
Core question: Does authority survive every boundary? Identity, scope and delegation are carried across tools, agents, memories and systems.
Treat a skill as untrusted code with a manifest: declared tools, data scope, network needs and author. Sandbox it, scan the manifest, and grant only the declared capabilities at run time.
How the wiring works
- Parse the skill manifest and compare declared capabilities with requested ones.
- Static-scan the skill for dangerous calls and hidden network access.
- Run it in a sandbox with no inherited secrets and default-deny egress.
- Record the skill digest and the exact capability grant for each invocation.
What is real today
LIVE A public route in this deployment exercises this mechanism today, within its stated scope.
The hosted OpenClaw demo audits and sandboxes fixed sample skills. It does not certify the Agent Skills standard or any vendor implementation.
| Component in this repository | What the module says it is |
|---|---|
cain/isolation/skill_sandbox.py | CAIN Sovereign Isolation — Cryptographic Skill Sandboxing Engine. |
cain/runtime/skills.py | CAIN Sovereign — Agent Skill Trust Fabric (Prompt 7, Phase 7). |
keystone/openclaw.py | OpenClaw KEYSTONE: OpenClaw / ClawHub skills audited, Claude hardens, cryptographic sandbox proves isolation, signed. |
keystone/skill_static.py | Static analysis of agent skill / plugin code: what the code can DO, not which words it contains. |
Descriptions are the first line of each module's own docstring, read without importing it. A module path is a pointer into this repository, not a hosted endpoint. Its existence is checked by the test suite; its status for this use is the label above.
Check it yourself
- Run the live skill-sandbox demo
/openclaw-keystone - Fire a live decision through the CAIN-42 pipeline
/fabric/try— the same enforcement path every product sits behind.
Commercial package
| Offer | Portable Skill Review |
|---|---|
| Buyer | Agent-platform vendors and marketplace operators |
| Price (proposal) | $2,500–$5,000 per review |
| Bundle | Identity & Authority — Every agent, every delegation, every boundary. |
| Operator stream | Stream 8 of the 10-stream operator playbook |
CISO, IAM, platform security, agent-tool vendors is the bundle buyer. The entry offer for this bundle is Agent authority review; MCP/A2A boundary review; the range is $2,500–$7,500 per review; integration from $10,000. A price on this page is a proposal, not a validated rate, and no customer is claimed.
Buy this as a product or service
CAIN-A75-017 — Portable Skill Review DELIVERABLE
A bounded engine run on the buyer's sample: findings, a written report, and a SHA-256 artifact manifest. The manifest is unsigned: it is not a signature and not an attestation.
Kind: Assessment · Price (proposal): $2,500–$5,000 per review · Lead time: 5 business days.
Open the product page · all 75 products
What this page does not claim
What this page is not. CAIN-42 is not affiliated with, endorsed by, integrated into, or shipped with the organisation named in the 2026 item. The item is public context; the page describes how CAIN-42 would govern that class of capability. No page here claims a customer, a partnership, a certification, a patent, or a production integration. Prices are proposals. Built by a three-person team.