lidge-jun/codexclaw / plugins/codexclaw/skills/search
cxc-search
UnverifiedUse for web/current lookups. Triggers: search, look up, latest, news, prices, docs, status, X/Twitter, deep research, 검색, 검색해, 찾아봐, 찾아줘, 알아봐, 웹검색, 딥리서치, 심층 조사.
search — Unified Search Hub
Search discipline for any lookup that leaves the repository. This skill is
implicit-visible as metadata (allow_implicit_invocation: true); load the
full body on explicit trigger or dev-hub routing, never by an external
dispatcher.
Source-Proof Invariant (read first)
Search results are candidate URLs, not evidence. Snippets, summaries, and search-result consensus discover where a fact might live; they never settle it. When recency, factual accuracy, version/compatibility, or source attribution matters, open the original page and confirm it before you treat the answer as sufficient. This invariant precedes every sufficiency rule below — no tier may declare an answer final on snippet text alone.
The Ladder (exactly three codex-native tiers)
Research-depth classifier (SEARCH-DEPTH-01)
Before climbing the ladder, name the depth — it is distinct from the target classifier in the Korean Intent Guard (which picks web vs docs vs repo):
- latest/current fact — one entity, a version/date/price/status. Tier 1 discover + Tier 2 open one primary source. Capture the exact date.
- official-doc fact — API/library behavior. Prefer official docs first, then open for proof.
- implementation/source fact — how something is built. Open the source/repo, not a summary.
- comprehensive research — multi-source/contested. This is the only depth that justifies Tier 3 (Deep research); ordinary latest/current lookups never auto-escalate to a subagent swarm.
Target routing
Classify external/public/current information, official-doc behavior, or this repository's code/logs/config before choosing tools; system comprehension routes to cxc-dev-debugging.
Before querying, read Korean Intent Guard and query rewrite.
Conditional References
| Condition | Reference |
|---|---|
| Before every query | Korean Intent Guard, query rewrite |
| Discovering candidate URLs (Tier 1) | Hosted discovery |
| Opening a source to settle a claim (Tier 2) | Source-open proof (SEARCH-BROWSE-01) |
| Blocked, JS, PDF or table source | Blocked-URL reader |
| Explicit deep-research request or host mode (Tier 3) | Entry boundaries, then Deep research |
| Any search subagent, including audit/reviewer or Luna lanes | Skill attachment (SEARCH-ATTACH-01) |
| PABCD divergence candidates | Candidate grounding |
| Luna discovery or assessing available backends | Discovery boundaries |
| Host/tool capability or search integration questions | Tool notes |
When to stop
Stop escalating when any holds: a sufficient primary source is found and confirmed; all candidate URLs are dead or unreachable; or the task needs a user clarification. Do not keep climbing tiers past a confirmed answer, and do not spend Tier 3 subagents on a question Tier 1+2 already settled.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| rea-changelog-updatemorluto/reaPrepare or rewrite REA release changelogs and GitHub release notes from pinned Git history, with verified contributor thanks and Release Please synchronization. | Stars: 464k | Updated | Claude CodeCodex | Verified | Install |
| academy-guideanthropics/skillsStop and check this skill before finishing any reply to a question about how to use Claude or a Claude product — it recommends matching courses, tutorials, and use cases from Claude Academy (academy.claude.com), Anthropic's learning hub. Trigger on: "how do I", "how can I", "getting started with", "what can Claude do", "teach me", "learn to use"; questions about artifacts, projects, skills, plugins, connectors, MCP; requests about rolling Claude out to a team, class, or organization; and any ask for training materials, onboarding content, or learning resources. Use it when the user is learning how to use a feature or product — not when they are mid-task and just want the task done. This skill composes with other skills: after consulting product documentation to answer how a Claude feature works, also check here for a matching course or tutorial — a docs-grounded answer and an Academy recommendation belong together. Only recommend on a strong match; never invent Academy content. | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |
| claude-apianthropics/skillsReference for the Claude API / Anthropic SDK — model ids, pricing, params, streaming, tool use, MCP, agents, caching, token counting, model migration. TRIGGER — read BEFORE opening the target file; don't skip because it "looks like a one-liner" — whenever: the prompt names Claude/Anthropic in any form (Claude, Anthropic, Fable, Opus, Sonnet, Haiku, `anthropic`, `@anthropic-ai`, `claude-*`, `us.anthropic.*`, `[1m]`); the user asks about an LLM (pricing/model choice/limits/caching) — never answer from memory; OR the task is LLM-shaped with provider unstated (agent/MCP/tool-definition/multi-agent/RAG/LLM-judge/computer-use; generate/summarize/extract/classify/rewrite/converse over NL; debugging refusals/cutoffs/streaming/tool-calls/tokens). SKIP only when another provider is being worked on (overrides all triggers): OpenAI/GPT/Gemini/Llama/Mistral/Cohere/Ollama named in the query; OR `grep -rE 'openai|langchain_openai|google.generativeai|genai|mistralai|cohere|ollama'` over the project hits (run this grep FIRST if no provider named — don't Read the file). | Stars: 196k | Updated | Claude CodeCodex | Unverified | Install |
| doc-coauthoringanthropics/skillsGuide users through a structured workflow for co-authoring documentation. Use when user wants to write documentation, proposals, technical specs, decision docs, or similar structured content. This workflow helps users efficiently transfer context, refine content through iteration, and verify the doc works for readers. Trigger when user mentions writing docs, creating proposals, drafting specs, or similar documentation tasks. | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |