How do I install the Searching And Extracting Web With Parallel CLI skill?

Run npx skillmds@latest add msewell/searching-and-extracting-web-with-parallel-cli in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Searching And Extracting Web With Parallel CLI safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Searching And Extracting Web With Parallel CLI?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Searching And Extracting Web With Parallel CLI free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Searching And Extracting Web With Parallel CLI?

msewell (@msewell) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Searching And Extracting Web With Parallel CLI by msewell

Name: Searching And Extracting Web With Parallel CLI
Author: msewell

Searching And Extracting Web With Parallel CLI

Runs Parallel CLI web search and URL extraction in non-interactive, text-first mode for AI agents. Chooses between `parallel-cli search` and `parallel-cli extract`, applies fast defaults, and adds source/date filters when needed. Use when the user asks to discover current web sources, gather citations from specific domains, or extract targeted information from known URLs.

msewell c11d0c5 2.5 KB Updated Sep 17, 2026

File contents

Searching and Extracting Web Content with Parallel CLI

Prerequisites

parallel-cli must be installed and available on PATH.
Verify authentication once per task when needed:
- parallel-cli auth

Non-goals

General file/URL-to-markdown conversion tasks.
OCR, image description, or audio transcription.
Unbounded browsing without a concrete search or extraction objective.
Interactive browser automation workflows (click/fill/login/stateful navigation).
Single-page interaction loops where the agent must act on DOM elements by node ID.
parallel-cli setup and account provisioning guidance.

Default command profiles

Search (default)

parallel-cli search "<objective>" \
  --mode fast \
  --max-results 5 \
  --excerpt-max-chars-per-result 800 \
  --excerpt-max-chars-total 6000

Search with source/recency constraints

parallel-cli search "<objective>" \
  --mode fast \
  --max-results 5 \
  --include-domains <domain1,domain2> \
  --after-date YYYY-MM-DD \
  --excerpt-max-chars-per-result 800 \
  --excerpt-max-chars-total 6000

Extract

Use only selected URLs (best few first), and keep excerpts small.

parallel-cli extract <url1> <url2> <url3> \
  --objective "<what to pull out>" \
  --excerpt-max-chars-per-result 700 \
  --excerpt-max-chars-total 2100

Notes:

extract --no-excerpts alone is invalid (it errors unless full content is requested).

Refinement strategy

If results are weak or noisy, refine automatically in this order:

Narrow query terms (more specific entities/version/date).
Add --include-domains for trusted sources.
Add --after-date for recency-sensitive tasks.

Failure handling

Exit code 3 (auth): report auth issue and required setup.
Exit code 4 (API failure): retry once with simpler args (fewer filters/URLs).
Exit code 5 (timeout): retry with reduced scope (fewer results, fewer URLs, tighter objective).

Output format to user

Return only high-signal results:

Key findings (concise bullets)
Source list (title + URL, optional date)
Notable uncertainty/gaps

---name: searching-and-extracting-web-with-parallel-clidescription: Runs Parallel CLI web search and URL extraction in non-interactive, text-first mode for AI agents. Chooses between `parallel-cli search` and `parallel-cli extract`, applies fast defaults, and adds source/date filters when needed. Use when the user asks to discover current web sources, gather citations from specific domains, or extract targeted information from known URLs.---# Searching and Extracting Web Content with Parallel CLI## Prerequisites- `parallel-cli` must be installed and available on `PATH`.- Verify authentication once per task when needed: - `parallel-cli auth`## Non-goals- General file/URL-to-markdown conversion tasks.- OCR, image description, or audio transcription.- Unbounded browsing without a concrete search or extraction objective.- Interactive browser automation workflows (click/fill/login/stateful navigation).- Single-page interaction loops where the agent must act on DOM elements by node ID.- `parallel-cli` setup and account provisioning guidance.## Default command profiles### Search (default)```bashparallel-cli search "<objective>" \ --mode fast \ --max-results 5 \ --excerpt-max-chars-per-result 800 \ --excerpt-max-chars-total 6000```### Search with source/recency constraints```bashparallel-cli search "<objective>" \ --mode fast \ --max-results 5 \ --include-domains <domain1,domain2> \ --after-date YYYY-MM-DD \ --excerpt-max-chars-per-result 800 \ --excerpt-max-chars-total 6000```### ExtractUse only selected URLs (best few first), and keep excerpts small.```bashparallel-cli extract <url1> <url2> <url3> \ --objective "<what to pull out>" \ --excerpt-max-chars-per-result 700 \ --excerpt-max-chars-total 2100```Notes:- `extract --no-excerpts` alone is invalid (it errors unless full content is requested).## Refinement strategyIf results are weak or noisy, refine automatically in this order:1. Narrow query terms (more specific entities/version/date).2. Add `--include-domains` for trusted sources.3. Add `--after-date` for recency-sensitive tasks.## Failure handling- Exit code `3` (auth): report auth issue and required setup.- Exit code `4` (API failure): retry once with simpler args (fewer filters/URLs).- Exit code `5` (timeout): retry with reduced scope (fewer results, fewer URLs, tighter objective).## Output format to userReturn only high-signal results:- Key findings (concise bullets)- Source list (title + URL, optional date)- Notable uncertainty/gaps

Frequently asked questions