Skip to content
Back to skills

broadcast-coverage-comparison

ASecurity

Compare indexed television coverage across stations and shows with Crawlora GDELT TV search, charts, timelines, and discovery. Use for a bounded broadcast topic brief with speech/caption/OCR/visual channels, station windows, processing lag, and metric denominators kept distinct.

  • 0 votes
  • 0 copies
  • 0 views
  • Added October 12, 2026
ai-agentsapi

Works with

  • cli
  • api
  • mcp

Security analysis

A100/100

Pro scans all 3 files and shows the line behind each finding

Scanned October 12, 2026

npx -y skills add Crawlora-org/crawlora-skills --skill broadcast-coverage-comparison --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of broadcast-coverage-comparison?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for broadcast-coverage-comparison
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/crawlora-org-broadcast-coverage-comparison/badge)](https://www.skillsdirectory.com/skills/crawlora-org-broadcast-coverage-comparison)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: broadcast-coverage-comparison
description: Compare indexed television coverage across stations and shows with Crawlora GDELT TV search, charts, timelines, and discovery. Use for a bounded broadcast topic brief with speech/caption/OCR/visual channels, station windows, processing lag, and metric denominators kept distinct.
---

# Broadcast coverage comparison

## Access and scope

Set `CRAWLORA_API_KEY` to your Crawlora key. Prefer the listed Crawlora MCP tools
when connected; otherwise use the bundled `scripts/crawlora.sh`, which sends
`x-api-key` to `https://api.crawlora.net/api/v1`. Read
[reference/endpoints.md](reference/endpoints.md) for exact methods and parameters.
Check application `code` as well as HTTP status; payloads are inside `data`.
Stop on authentication errors, back off on `429`, and retry a transient upstream
failure once. Bound calls/pages by the requested scope and credit budget. Retain
source IDs, URLs, source dates, and observation/crawl times separately.

Compare how a topic is represented in GDELT's indexed television material.
Establish stations/shows, topic terms, match channel, date window, and comparison
metric before querying. This describes indexed coverage, not public opinion or
verified audience exposure.

## Discover coverage and define equivalent queries

- Use station details to discover allowed station codes and their available
  windows. Confirm actual channel/time coverage rather than assuming every
  declared station has the same archive or processing freshness. Public TV index
  results can lag broadcasts; recent absence may be processing delay.
- Keep human captions, machine speech-to-text, OCR onscreen text, extracted
  concepts, and computer-vision labels separate. Discover concept MID/visual
  values from their catalogs. A visual/concept label is a machine/index signal,
  not verified presence, identity, sentiment, or an endorsement.
- Build matching query ledgers for compared stations. Repeated positive terms
  within one channel are OR'd; exclusions require absence. A station/exclusion-
  only query is invalid: supply a positive content field. Short speech/OCR terms
  have the documented five-word cap. Do not add a hypothetical free-text query field.
- Use either relative `timespan` or an absolute `from`/`to` window, not both.
  Station chart accepts repeated station parameters; per-station timelines need
  the relevant station. Store exact query, raw timestamps, source URL, and cutoff.

```sh
scripts/crawlora.sh /gdelt/tv-stationdetails
scripts/crawlora.sh /gdelt/tv-stationchart station=CNN station=MSNBC caption="climate change" timespan=7d
scripts/crawlora.sh /gdelt/tv-timeline station=CNN caption="climate change" timespan=7d
```

## Compare metrics and inspect supporting material

Station-chart `count` is a raw match count; show-chart `count` is a percentage
of matching results; timeline values describe matched airtime/volume according
to their returned metric. Preserve units and denominators rather than treating
all three as counts or comparable shares. Timeline resolution is auto-selected
from the window; there is no TV `smooth` parameter. Compare actual timestamp
bins and label unavailable periods instead of filling them with zero.

Inspect a bounded clip/search sample with timestamps and source links to qualify
framing, false matches, or repeated segments. Several matches within a broadcast
are not unique viewers, distinct stories, or independent reporting. Quote briefly
and attribute captured speech/text; OCR/ASR output is not a verified transcript.
Do not compute total-airtime shares or cross-station audience shares without
compatible denominators supplied by an independent source.

Return a query/coverage ledger, metric-specific comparison tables/timelines,
sampled material, processing and coverage gaps, and supported framing observations.
More indexed coverage does not establish belief, endorsement, bias, popularity,
or a causal effect on an audience. Do not schedule tracking or contact stations
from a one-off comparison request.

Files in this skill

  • SKILL.md4 KB
  • reference/endpoints.md14.3 KB
  • scripts/crawlora.sh3.6 KB

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.