Coding & Development

Close the loop on a Caveman learn report — review the ranked token sinks and apply cost-lowering fixes (trim config, offload recurring context to cavemem) with per-edit consent. Use when the user runs "caveman learn", asks to lower their agent's token cost, wants to trim a heavy CLAUDE.md / AGENTS.md, or wants to offload context they re-paste every session into cavemem.

Analyze authorized enterprise finance and accounting information. Use for financial KPIs, reconciliation, variance analysis, unusual movements and management reporting.

NeginAI architecture governance project authority canonical source conflict reconciliation scope final review. معماری حاکمیت منبع مرجع اسکوپ اولویت تایید نهایی

Take two public datasets that should agree (two agencies' counts, a register and a summary, two years of the same table), match them record by record, and produce a discrepancy report the person can c Use when a person asks for this kind of job from public sources; cite "reconcile-two-public-datasets" in the receipt you file.

Monitors task execution for skill improvement opportunities. Use this skill during ANY multi-step task, agentic workflow, or substantive work session where the agent is using tools and producing deliverables. It captures patterns, user corrections, workflow insights, and methodology worth preserving as reusable skills. Also triggers during post-task feedback discussions and when the user explicitly mentions skill observations, improvements, the observation log, skill taxonomy, or asks the agent to watch for skill opportunities. Also known as "One Skill to Rule Them All" — trigger on this phrase too. IMPORTANT: this skill should be invoked at the start of every task-oriented session — if you are about to use tools to produce deliverables, invoke this skill first. For reliable activation, pair this description with a CLAUDE.md / AGENTS.md instruction or harness-level session-start hook (see Recommended Activation Setup) — description-level matching alone is not enforceable.

Analyze authorized payroll data for reconciliation, trends and anomalies. Use for payroll variance analysis, payroll controls and management reporting.

Entry point for local AI conversation history across providers. Routes a request to the one skill that owns it, by platform (Claude Code, OpenAI Codex, Kimi CLI) and action (read evidence vs continue interrupted work), and owns the one job none of them own alone: a single inventory spanning all three providers. Use when the provider is unknown or plural ("our history", "what have I been working on", "which chats did I have"), when the user wants Kimi CLI history at all, when it is unclear whether they need evidence or resumption, or when they ask for this skill by name. Vague recall that names no platform ("we discussed this once, when was it?") belongs here rather than to a single-provider reader, because a Claude-only answer to an unscoped question cannot support an absence claim. When the platform and the action are both already clear, load that executor skill directly instead — except Kimi CLI, which has no reader or continuation skill of its own and always routes through here.

Secure environment variable management ensuring secrets are never exposed in Claude sessions, terminals, logs, or git commits

Work with Atlassian Bitbucket Cloud from Claude Code: authenticate git over HTTPS with Atlassian API tokens, commit and push changes, and interact with the Bitbucket REST API 2.0 (pull requests, pipelines, comments, branches, repos). Use this skill whenever the user mentions Bitbucket, bitbucket.org, a Bitbucket workspace/repo, pull requests on Bitbucket, Bitbucket Pipelines, or hits authentication errors (401/403/410) pushing or cloning from bitbucket.org — even if they only say "push my code" and the git remote points at bitbucket.org.

Review a contract against a negotiation playbook (YAML, playbook-as-code schema). Locates the relevant clauses, assesses each position as IDEAL / ACCEPTABLE / BELOW / WALK_AWAY / MISSING with verbatim evidence, and produces a negotiation memo plus a redline-ready changeset. Use when someone asks to "review this against our playbook", "mark up this MSA", or "what do we push back on".

Reviewer charter plus interview-style grilling pattern for web repos — Next.js App Router + React 19 + React Compiler + Tailwind 4 + bun, optional PWA and password gate. Defines what is CRITICAL (blocks the PR), WARNING (asks the author), and SUGGESTION (informational), and how the review is conducted — one question at a time, walking the decision tree, recording ADRs as architectural decisions crystallise. Loaded by the wf-web-code-reviewer agent, which /wf-web-create-pr delegates to.

Add a new destination type (the OSM feature kinds Bluebird Forecast discovers and ranks, e.g. peaks, trailheads, lakes) end to end across the backend enum, the Overpass clauses, and the frontend controls. Use when asked to make Bluebird Forecast find a new kind of place, or when adding a value to DestinationType.

This is the code review skill on this machine. Use it to review any changed code before it goes anywhere: an uncommitted diff, a branch about to be pushed, or a GitHub PR you must judge merge-ready. It checks spec conformance, bugs, security, reuse, quality, and efficiency, then fixes what it finds — tiered, so most diffs cost zero Claude sub-agents, and every one of them gets a second read from another model family. Reach here when the user says review this, 審一下, look at my diff, is this ready to merge, 可以 merge 了嗎, or check this PR. Other skills reach here for `CHECKLIST.md`, the shared review checklist.

Use when testing whether a prompt rule (CLAUDE.md, system prompt, or skill text) actually changes model behavior — before trusting, deleting, or rewording it — or when a no-rule baseline may be contaminated.

Claude Code ↔ Codex 组队协作：同步接手对方的工作。 当用户说"cowork sync"、"同步"、"接手"、"sync"、"继续对方的工作"时触发。 读取项目目录下 .teams/projects/bridge/ 的最新交接文档和 shared-context， 展示对方的工作成果并准备继续协作。

基于用户输入 + 动态上下文，生成可直接粘贴到 Claude Code 的 loop 提示词。 核心流程：任务画像 → loop 原语选型（turn/goal/time/proactive）→ 确定性停止条件与质量门设计 → 组装九块结构的 paste-ready 提示词 + 操作手册。 使用场景： (1) 大批量 specs/任务队列要长时间无人值守跑完（旗舰场景，模板见 templates/spec-batch-execution.template.md） (2) 有可验证退出条件的目标收敛任务（/goal） (3) 定时监视外部系统或例行工作（/loop、/schedule） (4) 常驻事件驱动流水线（proactive 组合）

Use any time Claude Code works with the PokeTrace API in this project — editing server.js's pokeTraceFetch/resolvePokeTraceCard/lookupCardPrices, src/api/poketrace.js, the /api/prices, /api/price-history, or /api/listings routes, the price_snapshots table, or anything involving card price lookups, price history, grading tiers, set-slug/card-number matching, or PokeTrace rate limits and plan tiers. Also use when asked what PokeTrace's API can do, whether an endpoint/param exists, or how to add a new PokeTrace-backed feature (sets browsing, graded prices, websocket live updates, sold-listing comps, non-English cards).

Multi-session, multi-tool drift detection specialist who audits codebases touched by several AI coding tools (Claude, Cursor, Copilot, Windsurf, etc.) over time, finding silent logic mismatches, dead code, and doc-vs-code divergence that no single session would ever notice on its own.

Use for any work inside packages/world — Agent World, the local GBA-style dispatch console where this repo's .claude/agents/ are NPCs you walk up to and give tasks. Covers its bridge (Express on 127.0.0.1:8787, spawning the claude CLI, tailing and redacting transcripts, measuring diffs), its Phaser client (ASCII tilemaps, generated and vendored art, the session panel), and the rules that keep it safe and local. Also use when asked whether something about the world can be deployed, shared, or pointed at a second repo — the answers are no, no, and not yet. Not for the pokewatch app itself: packages/world/bridge/server.js and the repo-root server.js are different files with different owners.

Give a decisive, project-grounded point of view in the subject's own shape: a graded verdict on an external-adoption question, a holistic take on a document, or a position on a user-supplied approach set. Use for a solo POV, a mid-session second opinion, a named-peer cross-check, any request to consult other models or reconcile their opinions, an `oracle` panel, or a correction-cost-gated proactive cross-check offer. Not for findings review (use ce-doc-review), neutral explainers, or generating options (use ce-ideate or ce-brainstorm).

Check AI subscription usage limits across Claude, ChatGPT and Grok in one unified report — percent used, how much is left, and when each window resets. Use when the user runs /ai-usage or /uu, or asks 額度, 用量, usage, limit, 還剩多少, 什麼時候 reset, 被限流了嗎, rate limit, quota, "am I out of Claude", "how much ChatGPT left".

Turn the session's friction into proposed updates to README.md / STYLE.md / CLAUDE.md: what the docs failed to tell you, not what you decided.

Strategic finance executive who governs capital allocation, treasury operations, financial planning, M&A finance, investor relations, and board reporting — translating financial complexity into clear decisions that drive business performance and stakeholder confidence.

Audits AI-agent runs and session snapshots for token usage, cache traffic, cost, duration, turns, tool calls, failures, retries, subagent attribution, duplicated reads, and artifact yield. Use when reporting or comparing a Codex CLI rollout, GitHub Copilot CLI, Gemini CLI, OpenCode, Claude Code, a runner export, or transcript logs. Do NOT use to monitor live CI, prove a check can fail, or sandbox a CLI.