// Default for web lookup/research/latest/current/URL/package/docs questions. Uses kagi→codex→claude-code→openai-cli→perplexity, escalating only if needed. Prefer this over provider skills unless user explicitly names kagi, perplexity, codex/claude-code/openai-cli, or ai-search.
| name | web-search |
| description | Default for web lookup/research/latest/current/URL/package/docs questions. Uses kagi→codex→claude-code→openai-cli→perplexity, escalating only if needed. Prefer this over provider skills unless user explicitly names kagi, perplexity, codex/claude-code/openai-cli, or ai-search. |
| allowed-tools | ["Bash","Read"] |
Single entry point for web research. Runs providers in a strict cost/speed order and lets you escalate when the previous tier's answer is not good enough.
| Tier | Provider | Why first / when |
|---|---|---|
| 1 | kagi quick | Fastest grounded answer, source confidence %, free w/ subscription |
| 2 | codex | Deeper synthesis, clean canonical URLs, code-aware; uses gpt-5.5 with low reasoning |
| 3 | claude-code | Second analytical opinion when codex hedges or gets it wrong; uses Haiku 4.5 with low thinking |
| 4 | openai-cli | OpenAI API web-search fallback; uses gpt-5.3-chat-latest |
| 5 | perplexity | Most citations, last resort for hard or very recent topics |
Always start at tier 1. Only escalate to the next tier if the current answer is insufficient (vague, hedged, missing key facts, contradicts known truth, no real URLs, or explicitly says it could not find X).
Do not call multiple tiers in parallel — the whole point is to save tokens and time when tier 1 already answers the question.
Single script, one provider per call:
~/.agents/skills/web-search/search.sh --provider <kagi|codex|claude-code|openai-cli|perplexity> "<query>" [--purpose "<why>"]
Examples:
# Tier 1 — start here every time
~/.agents/skills/web-search/search.sh --provider kagi \
"latest stable Firefox version macOS user agent" \
--purpose "update hardcoded UA in nvim plugin"
# Tier 2 — escalate if tier 1 was vague or missed the version
~/.agents/skills/web-search/search.sh --provider codex \
"latest stable Firefox version macOS user agent" \
--purpose "update hardcoded UA in nvim plugin"
# Tier 3+ — only when needed
~/.agents/skills/web-search/search.sh --provider claude-code "..."
~/.agents/skills/web-search/search.sh --provider openai-cli "..."
~/.agents/skills/web-search/search.sh --provider perplexity "..."
After each tier's output, ask yourself:
If tier-1 answer satisfies all four → stop. Don't run tier 2+.
KAGI_API_KEY env (already set) and ~/.kagi.toml with
session token. Wrapper does cd $HOME automatically. Output includes
source confidence percentages and follow-up questions.ai-search skill's search.mjs, using its current model defaults and
reasoning settings.openai binary on PATH and
OPENAI_API_KEY in the environment.PERPLEXITY_API_KEY env. Wraps perplexity-search
skill's --ask mode (sonar model).kagi search or
perplexity --search directly.perplexity --deep directly.defuddle or summarize skill, not search.kagi — full kagi-cli toolbox (translate, summarize, news, batch, etc.)perplexity-search — search/research/reason/deep modesai-search — direct provider control for codex/claude-code/openai-cliReference for the features supported by presenterm, a terminal-based markdown slideshow tool. Use this skill whenever the user works with presenterm: building or editing a presenterm deck, asking which presenterm features/commands/themes exist, how slides/columns/code blocks/diagrams/images/themes/exports work, what a comment command does, or mentions "presenterm", "terminal slides", or "markdown presentation" rendered in the terminal. This skill documents what presenterm supports — it does not prescribe slide content, wording, or presentation design.
Convert an article, newsletter, or document into a short Kokoro-narrated audio read-up and upload it as a private episode to the user's "📥 Listen Later" Spotify show. ONLY trigger on explicit phrases like "listen later", "read-up", "save as audio for the commute", "add to my listen-later feed". Do NOT trigger on generic summarize, TTS, save-to-spotify, podcast, or cover-art requests — those route to the `save-to-spotify` skill.
Low-level provider web search via codex, claude-code, or openai-cli. Use only when user names ai-search/codex/claude/openai-cli search, compares providers, or wants native search; otherwise use web-search.
Manage and deploy projects on exe.dev VMs. Use this whenever the user mentions exe.dev, exe.xyz hosts, `ssh exe.dev`, deploying Docker/GHCR apps to exe.dev, replacing or updating VMs, configuring exe.dev proxy/public access/custom domains, or setting up GitHub Actions/secrets/API tokens for exe.dev deployments.
Must read guide on creating/editing mermaid charts with valiation tools
Analyze coding-agent session transcripts to discover patterns that could become AGENTS.md rules, skills, or prompt templates. Mines your usage history for automation opportunities across local coding-agent tools.