git:20260908.b205970 to git:20261031.5d810c9

67 added, 9 removed. Audit A to A.

---
name: "Competitor Comparison"
- description: "Compare products against a specific buyer need using documented capabilities and explicit unknowns."
+ description: "Compare products or vendors against criteria chosen from the decision. Every table cell has a source, prices carry the date they were checked, and a fact that was not found is marked unknown. Use to choose a tool, to position a product, or to check a claim about a competitor."
---
# Competitor Comparison
+ ## When to use
+
+ - The user must choose between products, vendors or plans.
+ - The user wants to position their product against named alternatives.
+ - Not for an open research question. Use Evidence Brief for that.
+
## Inputs
- Get the buyer use case, alternatives, and decision criteria. Use dated product material or evidence provided by the user.
+ 1. **The decision.** Who decides, what they choose, and what is a must-have. If this is missing, ask once with
+ `ask_user`: "What decision does this comparison feed?"
+ 2. **The options.** 2 to 6 products. If the user names none, propose up to 4 from a search and ask them to confirm.
+ 3. **Tool path.** Call `openbot_browser.list_tabs`, then use a new tab. Read each vendor's product pages, pricing
+ page, documentation, changelog and status or limits pages. Search the `openbot` namespace with `tool_search`
+ for an installed plugin that holds the user's notes (Notion, Airtable).
+ 4. **Paste path.** Trial notes, sales calls, spreadsheets or screenshots that the user gives.
- ## Workflow
+ ## Steps
- 1. Define criteria before ranking alternatives. Include constraints such as setup effort, workflow fit, and operating limits where relevant.
- 2. Build a compact evidence matrix. Distinguish documented capability, a tested result, marketing claims, and unknowns.
- 3. Do not invent current prices, customers, features, or limitations. Where evidence is stale, label it and identify the check needed.
- 4. Explain tradeoffs for the stated use case. Avoid an overall winner if the criteria or evidence do not justify one.
+ 1. Write the decision in one sentence.
+ 2. Choose 5 to 8 criteria from the decision, not from the vendors' feature lists. For each, write what counts as
+ meeting it, and mark it must-have or nice-to-have. Show the criteria once and ask the user to confirm them.
+ 3. For each option and criterion, find a source: official documentation, the pricing page, the changelog, or a
+ test that you or the user ran. Record the source and the date you read it.
+ 4. Fill each cell with a short value and its source number. Write "unknown" when the documentation and pricing
+ page do not answer it. Write "not offered" only when the vendor says so.
+ 5. For pricing, record the plan name, the list price, the currency, the billing unit (per seat, per month, per
+ usage), and the date checked. "Contact sales" is a value. Do not estimate a price.
+ 6. Mark claims such as "fastest" or "most secure" as "vendor claim". Do not count them as Meets.
+ 7. Score each cell with the rubric. Then name the best fit for each scenario the user cares about.
+ 8. Write a short check plan for the unknowns that could change the result.
- ## Deliverable
+ ## Rubric
- Return a comparison table, the strongest fit for each relevant scenario, evidence dates, and a short validation plan for the most important unknowns.
+ - **Meets:** a documented or tested capability satisfies the criterion.
+ - **Partial:** it covers part, needs a higher plan, or needs a workaround. Say which.
+ - **Does not meet:** the vendor states it is absent, or a test failed.
+ - **Unknown:** no source found. Unknown never counts as Meets.
+ - An option that does not meet a must-have is out for that scenario.
+ - Name an overall winner only when the must-haves separate the options. Else give the best fit per scenario.
+
+ ## Output
+
+ ```
+ Decision: <one sentence>. Checked <date>.
+
+ | Criterion (must/nice) | Option A | Option B | Option C |
+ | SSO on the team plan (must) | Meets [1] | Partial: Enterprise only [4] | unknown |
+ | Price, 10 seats, monthly (must) | USD 120 per month, Team plan, checked 2026-10-31 [2] | Contact sales [5] | ... |
+
+ Best fit
+ - <scenario>: <option>, because <cells>.
+
+ Unknowns to check
+ - <option, criterion>: <how to check: ask sales, run a trial step, read a page>
+
+ Sources
+ [1] <title>. <vendor>. Accessed <date>. <site address>
+ ```
+
+ ## Approvals and stop conditions
+
+ - Do not sign up for trials, book demos, contact sales, or enter payment details without the user's yes.
+ - Do not publish the comparison unless the user says "publish".
+ - Stop and ask when a pricing page needs a login or a quote form.
+
+ ## Common failures
+
+ - Criteria picked after the research, so they favor one option.
+ - A feature written from memory instead of from a source.
+ - A price with no date, plan name or billing unit.
+ - An empty cell filled with a guess instead of "unknown".
+ - One winner when the answer depends on the scenario.