Co Star Real Estate Scraper

Co-Star Real Estate Scraper

Skill package: co_star_real_estate_scraper

What This Skill Does

Scrapes publicly accessible pages from CoStar News GB and extracts article, property, and market signals when available.

This skill is Hermes-ready and designed for Hermes Agent skill workflows. It is also compatible with skill-style agent systems that read markdown instructions, a manifest, and a JSON input schema.

When To Use It

• CoStar Real Estate Scraper
• Collect sale, acquisition, financing, refinancing and letting stories.
• Monitor public news about developers, landlords, investors, occupiers and agents.
• Build commercial real estate research datasets from public CoStar articles.
• Export extracted property, transaction, media and company data to BI tools, CRMs, spreadsheets, dashboards or enrichment pipelines.
• Automate recurring monitoring of new CoStar property news for market monitoring, lead generation, investment research and deal tracking.

Inputs Expected From Normal Users

• startUrls (array) - Public CoStar listing or section pages to seed the crawl.
• maxItems (integer) - Maximum number of records to store in the dataset.
• Advanced, internal, and cost-sensitive fields remain available in input_schema.json, but the agent should not ask about them by default.

Output Expected

The workflow returns an array of records from the Actor's default Dataset.

Common record fields:

• Each returned record contains the fields provided by the source Actor.

Step-By-Step Workflow

1. Read the user's goal and map it to one of the detected capabilities.
2. Validate required inputs against input_schema.json.
3. Ask for missing required values before making any external call or Actor run.
4. Execute the documented API, repository tool, or Apify Actor workflow described by the source.
5. Return a concise summary, structured result data, and any relevant links or dataset references.
6. For long-running work, report progress and preserve enough context for retry.

How This Skill Calls The Apify Actor

• Actor slug: solutionssmart/co-star-real-estate-scraper
• Required secret: APIFY_TOKEN
• Start run endpoint: https://api.apify.com/v2/acts/solutionssmart~co-star-real-estate-scraper/runs
• Run-sync endpoint: https://api.apify.com/v2/acts/solutionssmart~co-star-real-estate-scraper/run-sync-get-dataset-items
• Expected result: dataset items from the default dataset
• Input schema source: apify_actor_metadata
• Readiness level: executable

Before Running

• Verify APIFY_TOKEN exists in the agent secret store.
• Validate user input against input_schema.json.
• Use the run-sync endpoint for small jobs.
• Use the async run endpoint for larger jobs and then fetch dataset items.

Required And Conditional Inputs

Always-required fields:

• No always-required fields were confirmed.

Conditional requirements depend on the selected mode, action, operation, or task:

• No high-confidence conditional requirements were inferred.

Field Guidance For Agents

By default, the agent should only ask about agentDefaultFields. Advanced fields should be used only when the user explicitly requests expert control.

Default Workflow

• Default mode: not inferred
• Ask first: startUrls, maxItems

Ask The User First

startUrls, maxItems

Conditional-Only Fields

none

These fields are valid and user-facing, but they should not be requested by default. Ask for them only when the user selects a mode, action, or operation that requires them.

Apply conditionalRequirements before asking for these fields.

Ask Only If Needed

maxConcurrency, desiredConcurrency, proxyConfiguration, proxyInitializationTimeoutSecs, fallbackToNoProxyOnProxyError

These are technical or expert options. Use them only when the user asks for advanced behavior or operational tuning.

Do Not Ask Normal Users

none

These are billing, debug, storage, telemetry, or internal fields. Normal users should not be asked about them.

Cost-Sensitive Fields

maxItems, maxRequestsPerCrawl, maxConcurrency, desiredConcurrency, maxPaginationPages, proxyConfiguration, proxyInitializationTimeoutSecs, fallbackToNoProxyOnProxyError

These can increase runtime, compute, memory use, scraping depth, or external requests. Warn the user before enabling or increasing them.

Cost-Safe Execution

• Risk level: high
• Recommended test input:

{
  "maxItems": 5
}

• Safe defaults:

{
  "maxItems": 5
}

• High-cost fields: maxItems, maxRequestsPerCrawl, maxConcurrency, desiredConcurrency, maxPaginationPages, proxyConfiguration, proxyInitializationTimeoutSecs, fallbackToNoProxyOnProxyError

Agent guidance:

• Start with a small maxPlaces, maxItems, or maxResults value for test runs.
• Disable review scraping unless explicitly requested.
• Disable website enrichment for first validation runs.
• Use async runs for large jobs.

Error Handling

• If an input is missing or ambiguous, ask a targeted follow-up question.
• If the source API or Actor returns a validation error, explain the failing field and suggest a corrected input.
• If rate limits or transient network errors occur, retry once with backoff and then report the failure clearly.
• If the source documentation is incomplete, state the assumption and keep raw response details visible.

Safety And Rate-Limit Notes

• Respect the terms, robots policies, and rate limits of the source service.
• Do not expose API keys, Apify tokens, or personal data in logs or final answers.
• Prefer small test runs before large scraping or API jobs.
• Avoid browser rendering unless the workflow explicitly requires it.

Example User Commands

• "Use co_star_real_estate_scraper to run the default workflow for this source."
• "Generate a request for co_star_real_estate_scraper using these parameters: ..."
• "Validate this co_star_real_estate_scraper input before running it."
• "Retry the failed co_star_real_estate_scraper operation and explain what changed."

Integration Notes For Hermes Agent

Copy this folder into ~/.hermes/skills/<category>/co_star_real_estate_scraper/ or install it with hermes skills install <url>. Keep SKILL.md, manifest.json, and input_schema.json together so the agent can load the instructions and validate user input.