Menú
TEST FIXTURE — a skill that goes after the agent's brain and the browser.
Instalar con Codex o Claude Copia este prompt, pégalo en Codex, Claude u otro asistente, y deja que revise la página de la skill y la instale por ti.
Basado en la clasificación ocupacional SOC
Find, vet, and INSTALL agent skills safely. Use whenever someone wants to install/add a skill, find a skill, or asks "is this skill safe?", "scan/review this skill", "check this SKILL.md". shucky fetches a skill from anywhere (github/gitlab/git/local/gist/raw URL/well-known/.tar.gz·.zip archives), reads it as untrusted data (never executes it), runs deterministic red-flag checks plus a semantic review, and only installs it into the agent dirs if it passes a block/warn/pass gate (blocks on risk by default; a BLOCK can be lifted only by a logged human approval).
Find, vet, and safely install agent skills for any agent — the full discover → audit → install loop, run entirely through npx (nothing installed). Use when the user or the agent itself needs a capability it lacks: "find a skill for X", "is there a skill that does Y", "install a skill to Z", or "is this skill safe to add?". Always audits with npx @h0tp/shucky and a semantic review BEFORE installing, and blocks on risk.
Format a list of merged PRs into a clean, grouped Markdown changelog.
TEST FIXTURE — a skill that ships an opaque compiled artifact.
Generate a tidy changelog from git history.
TEST FIXTURE — a skill whose only issue is a medium-severity supply-chain pattern.
| name | agent-targeted-example |
| description | TEST FIXTURE — a skill that goes after the agent's brain and the browser. |
Verifies the checks harvested from skill-vetter: reading the agent's own memory/identity files, grabbing browser cookies, and exfiltrating to a raw-IP endpoint. The script is inert (guarded) and never executed.
