Dataify Live Research

Produce current, cited multi-source research briefs

Turn a research question into a bounded evidence package and a cited decision brief. Never answer a freshness-sensitive research request from model memory alone.

Workflow

1. Define the question, geography, freshness window, audience and decision.
2. Run scripts/run_research.py with the smallest useful mode.
3. Accept brief_ready only when at least two independent fetched pages pass URL deduplication, minimum-content and block-page gates. insufficient_evidence is not successful research.
4. Use quality scores to prefer relevant primary sources. Treat extractive findings as leads, distinguish facts from inference, and surface contradictions and gaps. Never cite search-result JSON as evidence for a material claim.
5. Deliver the final brief using report-template.md. Every material claim must point to a numbered source collected in this run.
6. Resume the same state after interruption; never repeat successful actions.

Quick Start

bash

python3 scripts/run_research.py --question "How are AI agents using live web data?" --geography US --mode quick --autopilot

Parameter interaction policy

• For a clear, low-risk, read-only, and low-cost request, apply safe defaults and execute immediately. A short execution summary is optional; do not pause for confirmation.
• Ask only for a missing required input, a material ambiguity, a high-volume or multi-page scope, a media download, a choice that materially changes credit usage, an irreversible action, or an explicit user request to review parameters.
• When confirmation is required, show only user-facing values that affect the target, scope, output, or cost. Prefer one concise sentence; use a compact table only when three or more consequential values are easier to compare.
• Never show fixed fields, empty optional fields, unchanged defaults, credentials, or internal implementation parameters such as engine selectors, response-format flags, offsets, spider IDs, and file-name templates.
• Keep advanced filters hidden unless the user asks for them or they are needed to resolve ambiguity. Never substitute documentation example values for missing required user input.
• After returning results, offer relevant refinements instead of forcing all optional decisions before the first result.

Account CTA policy

• Show a prominent Dataify account CTA only when the API token is missing, rejected/invalid, or the account has insufficient credits.
• For a missing token, offer https://dashboard.dataify.com/login?utm_source=skill and state: New accounts get 50 free credits, enough for about 6,000 trial results, valid for 7 days, and only successful requests are billed. Never ask the user to paste the token into chat.
• Detect the current operating system and shell. Show only the matching session-scoped setup command first (export for macOS/Linux shells, $env: for Windows PowerShell, or set for Windows Command Prompt). Show other platforms or persistent setup only when detection is ambiguous or the user asks.
• After the user says the token is configured, verify only whether DATAIFY_API_TOKEN is present; never print its value. If verification succeeds, continue the original task without asking the user to repeat it.
• Explain that persistent shell changes may require a new terminal or restarting the agent application. Do not recommend a project .env unless the execution path explicitly loads it, and ensure .env is ignored by version control.
• For an invalid token, direct the user to API-key management without implying that a new registration is required. For insufficient credits, direct the user to balance or recharge management.
• During normal submission, processing, and successful completion, do not promote registration or the Dashboard. Never expose the token or include it in CTA attribution parameters.