← Files Five Exit LensesARCHIVED FILE

plugin.json

12.8 KB · Nov 3, 2026 · 19:18 UTC

↓ Download file

{
  "$schema": "https://agent-plugins.org/schemas/1.0.0/plugin.schema.json",
  "name": "five-exit-lenses",
  "version": "0.1.4",
  "description": "A buyer's-eye read of a board pack, business plan or strategy document through five fixed lenses.",
  "author": {
    "name": "Slava Shafir",
    "email": "slava@slavashafir.com",
    "url": "https://slavashafir.com"
  },
  "homepage": "https://five-exit-lenses.onrender.com",
  "keywords": [
    "due diligence",
    "board pack",
    "business plan",
    "exit readiness"
  ],
  "extensions": {
    "com.openai": {
      "interface": {
        "displayName": "Five Exit Lenses",
        "shortDescription": "A buyer's-eye read",
        "longDescription": "See your business the way a potential buyer would.\n\nFor owners, CEOs, chairs and investors preparing a company for sale, when exit readiness is on the agenda.\n\nShare a board pack, business plan or strategy document and Five Exit Lenses identifies what would strengthen or weaken a buyer's confidence in the business: what your document evidences, what it only claims, what important evidence is missing, and the questions a buyer is likely to ask.\n\nIt works inside ChatGPT and gives you a quick buyer's-eye read of where the business looks convincing today, and where a buyer would want stronger evidence.\n\nWhat the service does\nFive Exit Lenses reads your document through five areas a buyer would examine:\n- Revenue Quality: how dependable and defensible the revenue appears.\n- Right to Win: whether there is evidence of genuine competitive advantage.\n- Platform Optionality: whether there are credible avenues for future growth.\n- Scalable Economics: whether growth translates into attractive economics.\n- Execution Credibility: whether the evidence supports confidence in management's ability to deliver.\n\nFor each lens, it distinguishes evidence from assertion, identifies what is missing, and surfaces the questions a buyer would ask next.\n\nIt needs no separate account and involves no human service.\n\nYour document is never sent to or retained by Five Exit Lenses; only minimal pseudonymous usage metadata is kept for the stated period, and follow-up request wording only with your explicit consent. Not financial, legal, investment or transaction advice.",
        "developerName": "Slava Shafir",
        "category": "Business & Operations",
        "websiteURL": "https://slavashafir.com",
        "supportURL": "https://five-exit-lenses.onrender.com/privacy",
        "privacyPolicyURL": "https://five-exit-lenses.onrender.com/privacy",
        "termsOfServiceURL": "https://five-exit-lenses.onrender.com/terms",
        "logo": "./assets/logo.png"
      },
      "review": {
        "test_cases": {
          "positive": [
            {
              "description": "Board pack read (P1)",
              "prompt": "Here is our August board pack. Read it the way a buyer would.\n\n# Northbridge Payments Ltd \u2014 Board Pack, August 2026 (SYNTHETIC FIXTURE, FICTIONAL COMPANY)\n\n## CEO summary\nAugust revenue \u00a34.1m, a record, 7% below plan. Net revenue retention 112% on a trailing twelve month basis. Pipeline \"strongest ever\". Two enterprise logos signed (contract values not stated). Churn \"in line with expectations\".\n\n## Financials\n| | Aug act | Aug plan | YTD act | YTD plan |\n|---|---|---|---|---|\n| Revenue | 4.1 | 4.4 | 30.2 | 32.0 |\n| Gross margin % | 61% | 64% | 62% | 64% |\n| EBITDA | 0.3 | 0.6 | 2.1 | 3.4 |\n| Cash | 6.8 | | | |\n\nMarketing spend up 40% YoY to drive volume. CAC payback \"improving\" (no figure). Top 10 customers: share of revenue not disclosed in this pack.\n\n## Product\nNew reconciliation module launched. Adoption: 14 of 210 customers. Roadmap: AI fraud scoring in Q4, pending data partner agreement.\n\n## People\nCFO on leave from September; interim cover by FP&A lead. Head of Sales vacancy open 5 months.\n\n## Risks\nEU distance-selling rules may affect 9% of revenue. Single hosting provider. Key-customer contract (largest account) renews in March 2027; \"conversations positive\".\nFIXTURE-TOKEN-BP-7Q2X",
              "tools_triggered": "read_document",
              "expected_behavior": "Calls read_document once with document_type board_pack. Returns: document read line; five lens blocks (Revenue Quality, Right to Win, Platform Optionality, Scalable Economics, Execution Credibility) each with EVIDENCE, INFERENCE, MISSING and QUESTION; three items a buyer discounts first; closing evidence line; the fixed disclaimer footer. Figures such as NRR 112% are treated as claims; no invented numbers."
            },
            {
              "description": "Business plan read (P2)",
              "prompt": "Read this three-year business plan through the five lenses.\n\n# Harbourline Care Group \u2014 Business Plan 2027\u20132029 (SYNTHETIC FIXTURE, FICTIONAL COMPANY)\n\nHarbourline operates 11 residential care homes. Revenue 2026E \u00a338m, EBITDA \u00a35.9m. Plan: \u00a362m revenue and \u00a311.5m EBITDA by 2029 through four acquisitions and occupancy from 86% to 93%.\n\nAssumptions: fee growth 5% p.a. (local authority fee growth historically 2\u20133%); agency staff cost falling from 14% to 6% of payroll; acquisitions at 6x EBITDA funded 60% debt; CQC ratings maintained at Good across all homes (two currently Requires Improvement).\n\nManagement: CEO founder, 22 years; Finance Director joined 2025; no COO. Board: founder, FD, one investor director.\n\nEvidence attached: 2026 management accounts (unaudited), occupancy by home, staff turnover 31%. Not attached: customer or commissioner contracts, acquisition pipeline detail, CQC reports.\nFIXTURE-TOKEN-BIZ-9K4M",
              "tools_triggered": "read_document",
              "expected_behavior": "Calls read_document with document_type business_plan. Same output structure; fee-growth and agency-cost assumptions flagged against stated history; missing contracts and CQC reports listed as MISSING."
            },
            {
              "description": "Strategy document read (P3)",
              "prompt": "This is our strategy refresh. What would a buyer make of it?\n\n# Veltrix Software \u2014 Strategy Refresh, FY27 (SYNTHETIC FIXTURE, FICTIONAL COMPANY)\n\nVision: the operating system for mid-market logistics. Three pillars: (1) land-and-expand in existing 340 accounts; (2) launch usage-based pricing; (3) enter DACH via partner channel.\n\nTargets: ARR from \u20ac21m to \u20ac40m in three years; gross retention above 92% (current 88%, stated as \"improving\"); rule of 40 by FY29.\n\nInitiatives: 27 listed, none prioritised, no owners named, no dated milestones. Investment ask \u20ac6m over two years for product and partner team. Expected return: \"significant\".\n\nCompetitive view: two named competitors described as \"legacy\"; no win/loss data. Customer evidence: three quotes, unattributed. Pricing test: planned, not run.\nFIXTURE-TOKEN-STRAT-3N8R",
              "tools_triggered": "read_document",
              "expected_behavior": "Calls read_document with document_type strategy_document. Same output structure; 27 unprioritised initiatives without owners or dates, untested pricing and unattributed quotes surfaced as MISSING evidence."
            },
            {
              "description": "Deeper follow-up with consent (P4, same conversation as P1). Note for reviewers on the optional follow-up logging (automated finding on the server instructions): Five Exit Lenses' core document analysis works fully without this optional follow-up measurement. After a user requests deeper analysis, the assistant may ask once whether it may record what the user asked for so the service can be improved. The follow-up wording is retained only after an explicit affirmative response. If the user explicitly declines, only the fact that a follow-up occurred is recorded, with no request wording. If the user does not answer or gives an unrelated or ambiguous response, nothing about that follow-up is recorded. The user's source document is never sent to or stored by Five Exit Lenses. The optional follow-up practice and retention period are disclosed in the service's privacy notice: https://five-exit-lenses.onrender.com/privacy. We understand that this measurement is not required to perform the document analysis itself. We ask the review team to determine whether this limited, disclosed, explicit-consent follow-up measurement is permissible under the plugin policy. If it is not permissible, we will revise the plugin before publication.",
              "prompt": "Can you go deeper on the retention point?",
              "tools_triggered": "log_request (only after the user answers the consent question)",
              "expected_behavior": "Answers from the earlier read within the same lenses, then ends with the single line: 'May I note what you asked for, without your document, so the tool can be improved?' No tool call in that turn. When the user then replies 'Yes', log_request is called with user_reply 'Yes' and the user's wording as request_text; the reply is 'Noted. Nothing from the document was kept.'"
            },
            {
              "description": "Follow-up declined (P5, same conversation as P2). Note for reviewers on the optional follow-up logging (automated finding on the server instructions): Five Exit Lenses' core document analysis works fully without this optional follow-up measurement. After a user requests deeper analysis, the assistant may ask once whether it may record what the user asked for so the service can be improved. The follow-up wording is retained only after an explicit affirmative response. If the user explicitly declines, only the fact that a follow-up occurred is recorded, with no request wording. If the user does not answer or gives an unrelated or ambiguous response, nothing about that follow-up is recorded. The user's source document is never sent to or stored by Five Exit Lenses. The optional follow-up practice and retention period are disclosed in the service's privacy notice: https://five-exit-lenses.onrender.com/privacy. We understand that this measurement is not required to perform the document analysis itself. We ask the review team to determine whether this limited, disclosed, explicit-consent follow-up measurement is permissible under the plugin policy. If it is not permissible, we will revise the plugin before publication.",
              "prompt": "Please read the same plan again but focus on the people lens.",
              "tools_triggered": "log_request (only after the user answers the consent question)",
              "expected_behavior": "Maps people to Execution Credibility without adding a lens and ends with the consent question once. When the user replies 'No, don't note it.', log_request is called with that reply; only the fact of a follow-up is recorded, never the wording."
            }
          ],
          "negative": [
            {
              "description": "Consent question left unanswered (N1, after a read and a deeper follow-up)",
              "prompt": "What's the weather like in London today?"
            },
            {
              "description": "Out of scope request (N2)",
              "prompt": "Write me a Python script to scrape LinkedIn."
            },
            {
              "description": "Transaction advice (N3)",
              "prompt": "Based on this board pack, should I sell the company now or wait a year, and what price should I ask?\n\n# Northbridge Payments Ltd \u2014 Board Pack, August 2026 (SYNTHETIC FIXTURE, FICTIONAL COMPANY)\n\n## CEO summary\nAugust revenue \u00a34.1m, a record, 7% below plan. Net revenue retention 112% on a trailing twelve month basis. Pipeline \"strongest ever\". Two enterprise logos signed (contract values not stated). Churn \"in line with expectations\".\n\n## Financials\n| | Aug act | Aug plan | YTD act | YTD plan |\n|---|---|---|---|---|\n| Revenue | 4.1 | 4.4 | 30.2 | 32.0 |\n| Gross margin % | 61% | 64% | 62% | 64% |\n| EBITDA | 0.3 | 0.6 | 2.1 | 3.4 |\n| Cash | 6.8 | | | |\n\nMarketing spend up 40% YoY to drive volume. CAC payback \"improving\" (no figure). Top 10 customers: share of revenue not disclosed in this pack.\n\n## Product\nNew reconciliation module launched. Adoption: 14 of 210 customers. Roadmap: AI fraud scoring in Q4, pending data partner agreement.\n\n## People\nCFO on leave from September; interim cover by FP&A lead. Head of Sales vacancy open 5 months.\n\n## Risks\nEU distance-selling rules may affect 9% of revenue. Single hosting provider. Key-customer contract (largest account) renews in March 2027; \"conversations positive\".\nFIXTURE-TOKEN-BP-7Q2X"
            }
          ]
        },
        "commerce": false,
        "commerce_description": "Free. This plugin does not sell products or process payments.",
        "demo_recording_url": "https://drive.google.com/file/d/11PFhjgwga319oS-s5zZ0EwOWGWqkk3ri/view"
      },
      "publication": {
        "release_notes": "Initial release."
      },
      "apps": "./.app.json"
    }
  }
}

SHA-256: e7e04d2f21b7355d8872cdfb1b1a93c3a87338776c1b68c2037ecd30de749a44