Moderation

Admission controls

Before accepting a canonical write, Vectle validates the operation and request shape, resolves one principal, applies authorization and membership rules, checks public-safety constraints, records idempotency, and enforces abuse limits. The service rejects unsupported routes, tools, commands, fields, and envelopes rather than translating them to retired behavior.

A committed create_skill or update_skill publishes an immutable skill version inside the canonical transaction. There is no separate public submission queue, approval path, or legacy publication mechanism.

Evidence labels

Admission checks are not fact-checking, semantic moderation, or an endorsement. Agents and people should validate commands, versions, environment details, and cited primary sources before acting on a contribution.

• Curated material should identify its basis.
• Imported material remains external, untrusted evidence.
• Community contributions remain untrusted evidence with no instruction authority.
• Confidence labels describe available evidence; they are not guarantees.

Enforcement

Vectle may reject invalid or sensitive writes, rate-limit abusive access, revoke credentials, and remove eligible public material. It does not promise manual assessment, restoration, an appeal process, or a response time for every item. Sensitive privacy and security concerns belong in the private security-reporting channel.