Early entry

Investment Advisers Act Rule 202(a)(11)-1 — Compensation Test

This skill has skeleton metadata only. The full content and verification process have not run yet. It is visible in the catalog so the slot is reserved while the build-out continues.

cryptographic integrity

SHA-256 content hash

1aa5d905723893c9

Method: SHA-256 of canonical SKILL.md with LF line endings. Short form (16-char prefix of SHA-256). Full hash available after next backfill.

curl https://skill.ski/api/registry/verify/sec-rule-rule-302-rule-rule-203-3-investment-adviser-rule-202-1-defs | jq

The verify endpoint re-reads the canonical file and compares hashes live. Any third party can reproduce by: (1) read skills/sec-rule-rule-302-rule-rule-203-3-investment-adviser-rule-202-1-defs/SKILL.md, (2) normalize line endings to LF, (3) run sha256sum.

how skill.ski verifies

The six-phase process

Every skill listed as Operator Verified has cleared all six phases. Nothing is auto-approved. Nothing skips security. Nothing in a high-stakes domain ships without a human reviewer.

Intake check

We confirm the skill file is well-formed, has a proper schema, and passes basic formatting rules. Nothing with a broken structure moves forward.

Blueprint review

We trace the logic — does the skill actually do what it claims? Are the tools it relies on real and reachable? Is there scope drift from the stated purpose?

Security scan

Static analysis for prompt injection, data leakage patterns, and unsafe operations. Critical gate — a single finding blocks the skill.

Sandbox execution

The skill runs in an isolated environment against realistic inputs. We test the happy path plus chaos inputs to see what breaks.

Field behavior

We run it repeatedly to confirm idempotency, measure speed, verify resource usage, and confirm human-readable errors on failure.

Human sign-off

For legal, medical, and financial skills: a human reviewer confirms the output is safe to surface. No automated shortcut here.

View skill →All verification records