Skip to content

Claude Skills by Riztazz

github.com/Riztazz

4 skillsA× 40 installs1 views

Io Guard DevAHow io-guard's own code is laid out, written, tested and verified. Read before writing or reviewing code in this repository, adding a check or an MCP tool, writing a test, running replay, or reporting a result. Covers the package layout and its import rules, the Python rules the plugin keeps, the six test levels, replay, live checks on Windows, and where to read telemetry. The design itself is docs/design/architecture.md.ai-agentspythongoVotes: 0GitHub stars: 8
DoneAEvery refusal from the guard names a code and a tool. When a refusal arrives, the model needs one short reference to read, not a page of rules. MCP tools sit behind tool search by default, so a refusal that says "use io.edit" sends the model to a tool it has not loaded yet.ai-agentspythonshellVotes: 0GitHub stars: 8
Io GuardAWhat an io-guard code means and the call that fixes it, and which io tool does which file or shell job. Read it when a tool result names an io-guard code such as SHELL_WRITE, ANCHOR_AMBIGUOUS or FILE_LOCKED, when a note says io-guard fixed a call, or before several edits to one file, a splice between markers, a log append, a long run, or formatting changed lines.ai-agentspythonrustVotes: 0GitHub stars: 8
SettingsAOpens io-guard's settings page, where each check turns on or off, each rewrite mode and list changes, and a project overrides any of them. Use it when the user asks to see or change io-guard's settings, or picks it from the plugin menu.ai-agentsVotes: 0GitHub stars: 8

Claude Skills by Riztazz — 4 skills, security-graded | Skills Directory