業界シグナルレーダー

---
name: akta-industry-signal-radar
description: Discover private companies and produce a bounded market-signal radar from reviewed Akta entities plus selected public-web evidence.
---

# Industry Signal Radar

## Loadout capabilities

Use <action-tag>api:akta_api:industry_search</action-tag> to identify companies inside the approved industry and geography scope. Use the News operation from <integration-tag>api:akta_api</integration-tag> to retrieve dated signals for only the selected company set and time window. Use the Exa Search Action from <integration-tag>composio:exa_tools</integration-tag> to test the reviewed set against a bounded public-news query, then use <action-tag>api:tinyfish_api:fetch_urls</action-tag> to extract only selected official or primary-source pages for material signals.

## Outcome

Produce a market radar that identifies a bounded set of relevant companies, clusters current signals, and states coverage and unknowns. Do not claim the result is an exhaustive market map or a continuous monitor.

## Required inputs

- Industry definition, included and excluded segments, and geography.
- Company filters and maximum number of companies to review.
- News date window and signal themes.
- Permitted public-source scope and maximum approved Loadout credit spend.

## Workflow

1. Restate the market boundary in operational terms: industry, geography, company filters, exclusions, result limit, news window, and signal themes.
2. Run Industry Search and review the returned company identities. Remove obvious out-of-scope entities and present uncertain inclusions separately.
3. Select the smallest company set that answers the question. Preflight the bounded News Signals request for that set and show the quote before execution.
4. When public corroboration is requested and the optional Actions are available, run a bounded Exa query for the same company set, themes, and date window. Select only relevant sources, then fetch the chosen official or primary-source URLs with Tinyfish. Otherwise continue with the Akta-only radar and state that public corroboration was not run.
5. Deduplicate coverage of the same event across Akta and public sources, preserve publication dates and source type, and group signals by the user’s themes rather than by unsupported sentiment.
6. Return the radar with explicit coverage limits, companies without signals, and actual credit use. Recommend a future rerun interval only when asked; do not imply that monitoring is already scheduled.

## Output contract

- `Scope`: industry definition, filters, exclusions, geography, and dates.
- `Company set`: reviewed companies and inclusion rationale.
- `Signal clusters`: dated events, affected companies, Akta and public evidence, and interpretation.
- `Quiet companies`: reviewed companies with no returned signal in the window.
- `Coverage limits`: uncertain matches, excluded segments, and provider gaps.
- `Cost`: quoted and actual credits plus Actions called.

## Failure and recovery

- If Industry Search is too broad, tighten one filter at a time and show how the scope changed.
- If too few companies match, report the exact boundary that constrained the result before proposing an expansion.
- If the news quote exceeds budget, reduce the company count or date window and request approval again.
- If no signals are returned, report a quiet window, not proof that nothing happened.
- If Akta and public sources disagree, preserve both and lower confidence instead of merging them.