CatchAll Skills - newscatcher

CatchAll Skills are Markdown instruction files that give an AI agent a specialized capability on top of the CatchAll MCP server. Each skill knows how to write the right queries, which validators to apply, and how to format output for its specific use case — with no extra coding or middleware required. Skills are platform-agnostic: they work with Claude (claude.ai, Claude Desktop, Claude Code), OpenAI Assistants, LangChain agents, and any agent runtime that supports Markdown-based instructions.

How it works

A skill file is a Markdown document the agent reads as instructions. When the agent is connected to the CatchAll MCP server (the preferred path for MCP-based skills), it uses the skill to decide which MCP tool to call, how to construct queries, and how to present results — no additional tooling required.

Diagram showing an AI agent reading a SKILL.md file and calling the CatchAll MCP server, which routes requests to the CatchAll API.

Most skills in this repository use the MCP interface: the agent calls the CatchAll MCP server via MCP tools, and the skill provides the judgment layer — query construction rules, validator patterns, enrichment schemas, and output formatting. For Claude-specific MCP setup, see the Claude integration page.

Before you start

A CatchAll API key from platform.newscatcherapi.com
The CatchAll MCP server connected to your agent — see the MCP integration page for setup instructions
An AI agent with Markdown instruction support (Claude, an OpenAI Assistant, a LangChain agent, or any custom runtime with a configurable system prompt)

Set up

How you load a skill depends on your agent platform. The core requirement is the same: the agent must be able to read the skill’s SKILL.md and its reference files, and have access to your API key.

Claude
Python agent (any LLM)

Claude Code and Claude Desktop — copy the skill folder into your project’s .claude/skills/ directory, or reference the folder path directly when starting a conversation.

# Clone the repository
git clone https://github.com/Newscatcher/newscatcher-catchall-integrations.git

# Copy your chosen skill into your project
cp -r newscatcher-catchall-integrations/skills/fundraising-catchall .claude/skills/

Claude.ai — zip the skill folder and upload it via the Skills feature at claude.ai/customize/skills.

# Create a zip from the skill folder
cd newscatcher-catchall-integrations/skills
zip -r fundraising-catchall.zip fundraising-catchall/

Upload the .zip via + → Upload a skill on the Skills page.

All MCP-based skills require the CatchAll MCP server to be connected. Add it from platform.newscatcherapi.com or follow the MCP integration setup guide before activating a skill.

Load SKILL.md as the system prompt in your agent framework, then configure the CatchAll MCP server or REST API as a tool.

from pathlib import Path

skill_prompt = Path("skills/fundraising-catchall/SKILL.md").read_text()

# Use as system prompt with any LLM SDK
messages = [
    {"role": "system", "content": skill_prompt},
    {"role": "user", "content": "Find AI startups that raised seed funding in Europe last 30 days"}
]

The agent needs access to the CatchAll MCP server or REST API. MCP-based skills call tools like submit_query, get_job_status, and pull_results — make sure those are exposed as callable tools in your agent setup.

Skills

Browse the available skills below. Each one is a self-contained folder in the integrations repository.

General Use Case

The foundation skill for the full CatchAll platform surface. Use it for any task not covered by a dedicated skill — or when the user needs direct control over validators, enrichments, watchlists, webhooks, or delivery setup. Main use case: Search, extract, and monitor any real-world event from the web — jobs, recurring monitors, Slack alerts, company watchlists, and more — without writing a single API call. Interface: MCP (https://catchall-mcp.newscatcherapi.com/mcp) · Folder: skills/general-use-case

general-use-case

SKILL.md

references

VALIDATORS.md

MONITOR-SCHEDULING.md

File	Purpose
`SKILL.md`	Core skill: query construction rules, full MCP tool reference (jobs, monitors, webhooks, datasets, entities, projects), job modes (`base` / `lite`), limit vs. page_size distinction, watchlist mode end-to-end workflow, full automation sequence, edge case handling
`references/VALIDATORS.md`	How to write effective boolean filters, cost control, the validators-vs-date-range distinction, and common validator patterns
`references/MONITOR-SCHEDULING.md`	Natural language schedule formats, timezone handling, and webhook configuration examples for recurring monitors

Example prompts

Find EU regulatory fines on Big Tech companies in the last 30 days, limit 20.

Set up a weekly Slack alert every Monday at 9 AM UTC for new AI model releases.

Track these 10 competitors and alert me whenever any of them raises funding.

How many credits do I have left?

Competitor Snapshot

Produces a structured digest of a competitor’s recent moves across the categories competitive intelligence teams actually use: product launches, pricing changes, leadership moves, customer wins, partnerships, M&A activity, and financial signals. Works on a single competitor or a watchlist of up to 100 companies. Main use case: Get a complete, structured picture of what a named company — or a list of companies — has been doing recently, across seven intelligence categories, without manually scanning dozens of sources. Interface: MCP (https://catchall-mcp.newscatcherapi.com/mcp) · Folder: skills/competitor-snapshot-catchall

competitor-snapshot-catchall

SKILL.md

references

QUERY-REVIEW.md

JOB-LIFECYCLE.md

OUTPUT-REPORT.md

NEXT-STEPS.md

CONCURRENCY.md

COMPANY-WATCHLIST.md

File	Purpose
`SKILL.md`	Core skill: seven intelligence buckets with their enrichment schemas, watchlist execution path, single-competitor and multi-competitor output templates, “events worth watching” cross-bucket section
`references/QUERY-REVIEW.md`	Pre-run intake rules: scope confirmation, cost gates, timeframe defaults
`references/JOB-LIFECYCLE.md`	Polling rules, progress table format, completion detection, 90-minute run cap
`references/OUTPUT-REPORT.md`	Output file contracts: xlsx workbook, JSON, and CSV schemas, table conventions, zero-event patterns
`references/NEXT-STEPS.md`	Footer links and follow-up actions rendered at the end of every snapshot
`references/CONCURRENCY.md`	Concurrency wave cadence for submitting multiple jobs in parallel
`references/COMPANY-WATCHLIST.md`	Full watchlist mechanics: CSV build, domain handling, dataset upload, connected submit, result attribution by `ed_score`

Example prompts

Snapshot Atlassian over the last 30 days.

What’s Salesforce been up to this month?

Give me a competitive brief on Notion, Linear, and Jira for the last 14 days.

Fundraising

Finds confirmed funding announcements across any geography, stage, and industry vertical. Returns structured event records, not raw links or article summaries. Main use case: Build prospect lists, track deal flow, or monitor market activity around funding rounds — driven by structured event extraction, not keyword search. Interface: MCP (https://catchall-mcp.newscatcherapi.com/mcp) · Folder: skills/fundraising-catchall

fundraising-catchall

SKILL.md

references

JOB-LIFECYCLE.md

OUTPUT-ARTIFACTS.md

OUTPUT-LIST.md

QUERY-REVIEW.md

NEXT-STEPS.md

scripts

build_downloads.py

File	Purpose
`SKILL.md`	Core skill: funding query formula, 4 standard validators, full enrichment schema (company name, round, amount split into value/currency/display, investors, location, industry, product description), extraction rules, limit heuristics
`references/JOB-LIFECYCLE.md`	Polling rules, progress tracking, completion detection
`references/OUTPUT-ARTIFACTS.md`	Output file contracts: xlsx, JSON, and CSV schemas, chat table column order
`references/OUTPUT-LIST.md`	Column definitions and field vocabulary for the output list
`references/QUERY-REVIEW.md`	Pre-run intake rules: scope confirmation, cost gates, timeframe defaults
`references/NEXT-STEPS.md`	Footer links and follow-up actions rendered after every run
`scripts/build_downloads.py`	Generates xlsx, JSON, and CSV download files from job results

Example prompts

Series B raises in Austin last 30 days.

Which AI startups raised seed funding in Europe this month?

Find all funding rounds over $50M in the US in the last 30 days.

Who got funded in fintech globally last 2 weeks?

Mergers & Acquisitions

Finds confirmed M&A deals — acquisitions, mergers, asset purchases, and acqui-hires — across any geography and industry. Returns structured event records with deal type, both parties, deal value where disclosed, and deal status. Main use case: Track deal activity for competitive intelligence, GTM targeting of recently-acquired companies, or ongoing market monitoring — all from a single natural language query. Interface: MCP (https://catchall-mcp.newscatcherapi.com/mcp) · Folder: skills/m&a-catchall

m&a-catchall

SKILL.md

references

JOB-LIFECYCLE.md

OUTPUT-ARTIFACTS.md

OUTPUT-LIST.md

QUERY-REVIEW.md

NEXT-STEPS.md

scripts

build_downloads.py

File	Purpose
`SKILL.md`	Core skill: M&A query formula, 5 standard validators, full enrichment schema (acquirer, target, deal value split into value/currency/display, deal type normalized and display, deal status, deal rationale), extraction rules, acquirer type taxonomy (`big_tech`, `strategic`, `financial`, `other`)
`references/JOB-LIFECYCLE.md`	Polling rules, progress tracking, completion detection
`references/OUTPUT-ARTIFACTS.md`	Output file contracts: xlsx, JSON, and CSV schemas, chat table column order
`references/OUTPUT-LIST.md`	Column definitions and field vocabulary for the output list
`references/QUERY-REVIEW.md`	Pre-run intake rules: scope confirmation, cost gates, timeframe defaults
`references/NEXT-STEPS.md`	Footer links and follow-up actions rendered after every run
`scripts/build_downloads.py`	Generates xlsx, JSON, and CSV download files from job results

Example prompts

AI companies acquired in the US last 30 days.

Fintech mergers announced in Europe this month.

Which healthtech startups were acquired in the last 14 days?

PE acquisitions in enterprise SaaS globally last 30 days.

VC Pack

Combines funding and M&A activity into a single interactive dashboard for a given market segment. Runs two parallel CatchAll jobs — one for funding rounds, one for acquisitions — and renders them together with aggregated KPIs, deal stage breakdowns, sub-sector distribution, and live FX conversion. Main use case: Get a complete capital flow view of a market in one run — where money is entering (funding) and where ownership is consolidating (M&A) — without manually combining two separate reports. Interface: MCP (https://catchall-mcp.newscatcherapi.com/mcp) · Folder: skills/vc-pack-catchall

vc-pack-catchall

SKILL.md

assets

render.py

dashboard.html

references

EXTRACTION.md

JOB-LIFECYCLE.md

NEXT-STEPS.md

CONCURRENCY.md

QUERY-REVIEW.md

scripts

render.py

File	Purpose
`SKILL.md`	Core skill: two-feed parallel submit, polling both jobs to completion, three-tier dashboard rendering (inline widget, file delivery, or markdown fallback), partial-render flow when one feed is slow
`assets/render.py`	Dashboard generator: aggregates both feeds, converts currencies via live FX, computes KPIs and deal stage breakdowns, produces the HTML dashboard and xlsx/JSON/CSV downloads
`assets/dashboard.html`	HTML template used by `render.py`
`references/EXTRACTION.md`	Query formulas, validators, and full enrichment schemas for both the funding and M&A feeds
`references/JOB-LIFECYCLE.md`	Polling rules, progress tracking, completion detection
`references/NEXT-STEPS.md`	Footer links and follow-up actions rendered after every dashboard
`references/CONCURRENCY.md`	Concurrency wave cadence for submitting the two parallel jobs
`references/QUERY-REVIEW.md`	Pre-run intake rules: scope confirmation, cost gates, timeframe defaults
`scripts/render.py`	Standalone script version of the dashboard generator for non-MCP environments

Example prompts

VC pack for fintech last 30 days.

Funding and M&A in cybersecurity US last 2 weeks.

Capital activity in healthcare AI this month.

Where is money moving in climate tech globally?

Have another use case in mind?

Feel free to share it margaretha@newscatcherapi.com and we will build it for you!

Write effective queries

Get better results from CatchAll jobs

API reference

Full endpoint documentation

Claude integration

MCP server setup and Python agent examples for Claude

Skills source

All skills and reference files on GitHub

​How it works

​Before you start

​Set up

​Skills

​General Use Case

​Competitor Snapshot

​Fundraising

​Mergers & Acquisitions

​VC Pack

​Have another use case in mind?

​See also

Write effective queries

API reference

Claude integration

Skills source

How it works

Before you start

Set up

Skills

General Use Case

Competitor Snapshot

Fundraising

Mergers & Acquisitions

VC Pack

Have another use case in mind?

See also