Nivl

Nivl/Config

Instructions file CodexOpenCode

AGENTS.md instructions for Nivl/Config, covering repository instructions, keep this file fresh, what this repo is, layout and bootstrap flow.

Nivl/Config

Agent Claude Code

Runs one specialized reviewer role of the in-depth-review skill and returns that role's findings. Invoked only by the in-depth-review skill, never directly by a user and never by auto-delegation.

Nivl/Config

Agent Claude Code

Proposes design and architecture findings for the pr-review approach debate. Judges whether the change is built the right way, not whether it has bugs. Invoked only by the pr-review skill, never directly by a user and never by auto-delegation.

Nivl/Config

Agent Claude Code

Thin wrapper that runs one gh-style-review pass and relays its JSON, including the Discussion Context block. Invoked only by the pr-review and review-and-fix skills, never directly by a user and never by auto-delegation.

Nivl/Config

Agent Claude Code

Opus-tier in-depth-review wrapper, the subtle-bug catcher in the mixed-tier finder pool. Invoked only by the pr-review skill, never directly by a user and never by auto-delegation.

Nivl/Config

Agent Claude Code

Thin wrapper that runs one in-depth-review pass and relays its JSON. Invoked only by the pr-review and review-and-fix skills, never directly by a user and never by auto-delegation.

Nivl/Config

Agent Claude Code

Pragmatic counterweight that judges whether the approach reviewer's design findings genuinely warrant a code change in this PR. Invoked only by the pr-review skill, never directly by a user and never by auto-delegation.

Nivl/Config

Hook Claude Code

Runs before the agent uses a tool for Bash, Write, Edit, NotebookEdit, Read and Workflow tool calls, executing dispatch-bash.py, write-under-roots.py, read-guard.py and deny-review-in-workflow.py (4 commands). From Nivl/Config.

Nivl/Config

Hook Claude Code

Runs on the PermissionDenied event for WebFetch tool calls, executing report-access-failure.py. From Nivl/Config.

Nivl/Config

Hook Claude Code

Runs when a tool call fails for WebFetch tool calls, executing report-access-failure.py. From Nivl/Config.

Nivl/Config

Settings file Claude Code

Agent settings declaring 3 hook events (PreToolUse, PermissionDenied, PostToolUseFailure) and 238 allowed tools.

Nivl/Config

Skill Claude CodeCodex

Automatically fix a GitHub PR for the current branch — handles review comments, CI failures, and open questions. Use this skill whenever the user wants to address PR feedback, fix CI, resolve review comments, or generally "clean up" or "fix" their PR. Also trigger when the user says things like "handle PR comments"…

Nivl/Config

Skill Claude CodeCodex

Locally mirrors the prompt sent by @claude review (the anthropics/claude-code-action GitHub Action). Supports two modes: PR mode (arg is a PR number / URL or auto-detected from the current branch) — pre-fetches the same rich PR context the Action injects (formattedcontext, PR body, conversation comments, inline…

Nivl/Config

Skill Claude CodeCodex

Performs one in-depth, multi-perspective code review of a pull request or commit range using up to twelve specialized parallel reviewer roles (AGENTS.md compliance, bug scan, git history, prior PR feedback, in-file comments, database, OWASP security, error handling, test coverage, ticket-intent compliance, motivation…

Nivl/Config

Skill Claude CodeCodex

Opens a new pull request for the current branch with a SIMPLE, scannable title and a short description. If the repo has a PR template (.github/PULLREQUESTTEMPLATE.md or similar), fills it in; otherwise uses a default "Goal / how / key-changes-list" format. Designed to avoid the wall-of-text PR descriptions that nobody…

Nivl/Config

Skill Claude CodeCodex

Turns a requirement in prose into the Jira issues for it. Infers the project, the active sprint and the assignee from the user's own tickets, reads the repo so every issue names real files and real patterns, sweeps Jira and git history for work already filed or already shipped, sizes the work against the story-point…

Nivl/Config

Skill Claude CodeCodex

Reviews a pull request with four parallel reviewer sub-agents — two in-depth-review on Sonnet, one in-depth-review on Opus (the subtle-bug catcher), and one gh-style-review — merges and deduplicates their findings, and posts a SINGLE PR review combining global findings, a names-only list of inline findings (also left…

Nivl/Config

Skill Claude CodeCodex

Iteratively reviews recent code changes (a PR if one exists for the current branch, otherwise the branch's commit range) and fixes what the reviews find. Runs in-depth-review and gh-style-review sub-agents in parallel each iteration, merges and deduplicates their findings, and applies fixes one commit at a time until…

Nivl/Config

Skill Claude CodeCodex

Reverse Prompt Engineering (RPE) — refine vague or incomplete requests into a precise, LLM-optimized task specification before any implementation begins. Use this skill when a prompt contains [rpe], or when the user explicitly asks to clarify, refine, or spec out a request before acting on it. Do NOT use for requests…

Nivl/Config

Skill Claude CodeCodex

Takes a Jira ticket key and drives it end to end. Validates the ticket is still real against the code at HEAD, the merged PRs the branch does not contain, the open PRs that overlap, and the related Jira tickets that may already duplicate it or have carved scope out of it. On a bug or a security ticket it also…

Nivl/Config

Skill Claude CodeCodex

Use whenever writing or rewriting prose that a human will read. Covers Jira tickets, Confluence pages, PR descriptions, commit messages, Slack and email updates, design docs, RFCs, runbooks, postmortems, READMEs, and release notes. Triggers on "write", "draft", "write up", "fill out", "summarize this", "turn this into…