ivuorinen

ivuorinen/skills

Plugin Claude Code

Hostile audit toolkit for Claude Code: adversarial review, security, tests, docs, architecture, reliability and agent-enforcement audits, with a bundled MCP server.

ivuorinen/skills

Plugin Claude Code

Nitpicker — hostile audit toolkit: /nitpicker dispatches specialist audit commands (adversarial review, security, tests, docs, types, architecture, perf, reliability, caching, concurrency, error handling, leaks, deps, licensing, CI, commits, migrations, observability, API contracts, a11y, i18n, privacy, confi.

ivuorinen/skills

Agent Claude Code

Verifies a release is ready to ship — checks skill and command validity, findings store, version sync, conventional commits, and CI status.

ivuorinen/skills

Agent Claude Code

Audits the nitpicker router and its command files for cross-command convention violations — naming, dispatch-table sync, findings-store usage, and behavioural consistency.

ivuorinen/skills

Hook Claude Code

Runs after a tool call finishes for Write, Edit, Write, Edit, Write, Edit, Write, Edit, Write, Edit, Write, Edit, Write, Edit and Bash tool calls, executing validate-skill-hook.py, validate-json-hook.py, check-version-sync-hook.py, ruff-hook.py, validate-audit-findings-hook.py, validate-rules-hook.py…

ivuorinen/skills

Hook Claude Code

Runs when the agent finishes a response, executing stop-reminder.py via uv with --quiet. From ivuorinen/skills.

ivuorinen/skills

Hook Claude Code

Runs before the agent uses a tool for Bash, Bash, Read, Glob, Bash, Bash and Bash tool calls, executing deny-agents-path-hook.py, deny-unsafe-git-hook.py, guard-ctx-ok-hook.py and ask-destructive-restore-hook.py with --quiet and --no-verify (6 commands). From ivuorinen/skills.

ivuorinen/skills

Settings file Claude Code

Agent settings declaring 3 hook events (PostToolUse, Stop, PreToolUse).

ivuorinen/skills

Skill Claude CodeCodex

Use for any question about a codebase, its architecture, file relationships, or project content — especially when graphify-out/ exists, where the question should be treated as a graphify query first. Turns any input (code, docs, papers, images, videos) into a persistent knowledge graph with god nodes, community…

ivuorinen/skills

Skill Claude CodeCodex

Scaffolds a new nitpicker audit command through the full RED → GREEN → REFACTOR → adversarial-review → validate cycle, wiring its command file, dispatch-table row, and docs together. Use when adding a new /nitpicker command to this repository, or when asked to create, scaffold, or add an audit command here.

ivuorinen/skills

Skill Claude CodeCodex

Checks that the current branch is ready for a release-please release PR — conventional commit types, version manifest sync, findings-store state, and CI status. Use when preparing or verifying a release, cutting a release PR, or confirming a branch is safe to merge for release.

ivuorinen/skills

Skill Claude CodeCodex

Runs TDD pressure scenarios against a skill before and after it is written, proving the skill actually changes agent behaviour instead of merely reading well. Use when verifying a new or edited skill has real effect, establishing a RED baseline before authoring, or confirming a GREEN result after.

ivuorinen/skills

Skill Claude CodeCodex

Routes audit requests to the right /nitpicker command. Use when the user wants to run one of the hostile audit commands in this repo, or asks what audit commands are available.

ivuorinen/skills

Skill Claude CodeCodex

Validates every SKILL.md and command file in this repository against the Agent Skills specification and the repo's own conventions, reporting structural errors before they reach CI. Use when verifying skills are well-formed — before a release, after adding or editing a skill, or when a skill-validation gate fails.

ivuorinen/skills

Instructions file GitHub Copilot

Copilot instructions for ivuorinen/skills, covering copilot cloud agent instructions, what this repository is, repository layout, the commands and file format — non-negotiable rules.

ivuorinen/skills

MCP server Claude CodeCodexCursor +2

MCP server "nitpicker" as configured in ivuorinen/skills. Runs ${CLAUDE_PROJECT_DIR:-.}/skills/nitpicker/scripts/mcp_server.py with python3. Needs 1 environment variable to run.

ivuorinen/skills

Instructions file CodexOpenCode

AGENTS.md instructions for ivuorinen/skills, covering agent instructions and working on this repo.

ivuorinen/skills

Instructions file

Claude Code instructions for ivuorinen/skills, covering claude.md, what this repo is, development commands, commands and agent skills spec compliance.

ivuorinen/skills

Skill Claude CodeCodex

Hostile audit toolkit: one entry point dispatching specialist commands — adversarial review, security, tests, docs, types, architecture, performance, reliability, caching, concurrency, error handling, resource leaks, dependencies, licensing, CI, commits, migrations, observability, API contracts, a11y, i18n, privacy…

ivuorinen/skills

Command

The default audit command (commands/audit.md) copies every task below into the agent's task list at run start — in Claude Code one TaskCreate/TodoWrite entry per task, the equivalent task tracker in other agents. This list is the audit's coverage contract: audit is "exhaustive" only when every task has been addressed.

ivuorinen/skills

Command

Read this file before executing any command file. Every rule here applies to every command unless the command file explicitly overrides it.

ivuorinen/skills

Command

Formats for the three markdown artifacts the teach command writes into docs/lessons/. Lessons, reference documents, and assets are HTML and have no fixed template — their rules live in teach.md.

ivuorinen/skills

Command

Hostile single-shot accessibility audit of the codebase's UI layer against WCAG 2.2 AA: assume the interface is unusable without a mouse and a screen until the code proves otherwise.

ivuorinen/skills

Command

Hostile audit of a project's hook coverage against its own evidence base: assume every recurring failure is unguarded until a hook is proven to fire on it, specify the missing hook in the host harness's correct shape, and on approval wire it and fire it on the evidence input to prove it binds.