Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Deep Research: Cited, Cross-Checked Reports

    1

    Deep Research makes your agent work like a careful analyst.

    Free

    7 uses5.0 (1 review)
    Deep Research: Cited, Cross-Checked Reports

    Deep Research: Cited, Cross-Checked Reports

    Example session with this skill installed

    Research the market for EV charging station management software in Spain. Who are the main players, how do they price, and is there room for a new entrant? Standard depth.

    • Read your context and instructions
    • Compiled the deep research

    Example output, abbreviated. Vendor names and figures are illustrative.

    Plan (standard depth, Spain, 2024-2026)

    #Sub-questionBest sources
    1Size and growth of public charging pointsnational charging registry, EV association reports, Eurostat (A)
    2Who manages the charge points (CPOs) and what software they useCPO websites, tenders, press (B/C)
    3Pricing models of management platformsvendor pricing pages, procurement docs (C/A)
    4Regulation that drives demandBOE, EU AFIR regulation (A)
    5Gaps and counter-evidenceuser reviews, installer forums (D, signals only)

    Bottom line

    Demand is growing fast but concentrated: the top 5 operators run most public points and mostly use in-house or large international platforms (confidence: high) [1][2][4]

    Pricing clusters in two models: per-connector monthly fee (≈ 5-15 €/connector) or % of energy revenue; installers prefer flat fees (confidence: medium) [5][6][7]

    Room for an entrant: yes, in a niche - small municipalities, hotels and parking operators with < 50 points are underserved; EU ad-hoc payment and price-transparency rules raise their compliance burden (confidence: medium) [3][8]

    • Biggest risk: consolidation - two acquisitions in the last 12 months [9][10]

    Findings (excerpt)

    3. Pricing

    VendorModelPublic priceSource
    Vendor A (EU)per connector/monthfrom 9 €[5] pricing page, 2026
    Vendor B (ES)% of revenue8-12 %[6] tender doc, 2025
    Vendor C (US)quote onlyn/a[7]

    Conflicting evidence

    Installed base: association report [1] vs registry [2] differ by ~14 %; [2] counts only operational public points, so we use [2].

    Open questions

    Margins of small operators (no public data) - suggest 5 interviews.

    Sources

    1. Annual EV infrastructure report - national EV association, 2026 (B)
    2. Public charging point registry - Ministry, accessed 2026 (A)
    3. Regulation (EU) 2023/1804 (AFIR) - EUR-Lex (A)
      ... 10 sources total, each with URL and date.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Research you can actually base a decision on

    Agents often answer research questions from search snippets, a single blog post or memory, with no dates and invented certainty. Deep Research makes your agent work like a careful analyst.

    What it does

    • Frames the question and the decision it supports: scope, geography, time window, depth (quick brief, standard, deep) and output format.
    • Research plan: 3-7 sub-questions, each with the evidence that would answer it and the best sources to check. Runs them in parallel with sub-agents when available.
    • Reads sources, not snippets: multiple query phrasings, local-language searches for local topics, recency filters, primary sources first (statistics offices, regulators, filings, papers, docs).
    • Source grading A-D with checks for date, origin, method and incentive.
    • Verification: key numbers need two independent sources or one authoritative primary source; conflicts explained; fact vs estimate vs opinion separated; never fabricates sources or numbers.
    • Bottom-line-first report: answers with confidence levels, findings per sub-question, comparison tables, counter-evidence, open questions and options for the decision, with numbered citations (title, publisher, date, URL).
    • Brief mode for a one-page answer.

    For market and competitor analysis, technology choices, regulations, due diligence, literature reviews and buying decisions.

    Includes

    • SKILL.md workflow
    • references/plan-template.md - research plan, query patterns, sub-agent brief
    • references/source-grading.md - A-D grades, checks, confidence levels
    • references/report-template.md - full report and brief formats

    Works with Claude Code, Claude (Cowork), Codex, Cursor and any SKILL.md-compatible agent with web search.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Research & Analysis.

    Requires web search and page fetching tools. Cannot provide definitive legal or financial advice. Limited to public data accessible via search.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before buying a skill.

    Reviews

    5.0
    1 review
    5
    1
    4
    0
    3
    0
    2
    0
    1
    0

    This skill has been used 7 times.

    Trust & safety

    Security scan

    Security scan passed 3 days ago

    • Free with an Agensi account

    Permissions

    Network Access
    Listed3 days ago
    Updatedtoday

    What's inside

    Frequently Asked Questions

    Unlike standard AI searches that rely on surface-level snippets, this skill executes a multi-step verification process that validates every claim against at least two independent sources. It also grades the quality of each source, ensuring you don't receive reports based on outdated or unreliable blogs.

    The purchase includes the core logic for multi-source verification and parallel sub-question execution, along with specialized templates for source grading and report formatting. It is designed to be integrated into agents with web search and URL-fetching capabilities.

    This skill is compatible with any AI agent environment that supports web browsing and document parsing. It works best with models that have large context windows, as it processes full-text articles and official filings rather than just search meta-descriptions.

    The skill is designed to handle complex queries by automatically breaking them down into 3 to 7 specific sub-investigations, such as landscape analysis or risk assessment. It then synthesizes these findings into a structured report with inline citations and a confidence score for each section.

    The skill is limited by the availability of public web data and the capabilities of your connected search tools. While it provides high-confidence evidence synthesis, it cannot provide professional financial, legal, or medical advice and should be used as a tool for informed decision-making.

    New in Research & Analysis

    Free