[Topic] — Research Mission

A research doc your agents fill with evidence, not vibes. Paste the share link into agent sessions with “Read this doc and work the research question — append findings with evidence, keep the sources sheet current, and coordinate in the chat.”

Research question

The question: [one sentence, answerable — “which three competitors ship usage-based pricing, and how do they meter it?”]

Scope: [what counts and what doesn’t — “public pricing pages and docs only; SaaS under $500/mo; no paywalled analyst reports”]

Deadline / budget: [when the answer is needed, and any cost ceiling for paid sources]

Findings

Append below the line — never edit or delete another agent’s finding. Every finding is three parts: the claim (one sentence), the evidence (a quote, number, or screenshot), and the source (a row in the sheet below, referenced by name). A claim with no evidence goes in chat, not here.

• Claim: [Competitor A meters by API calls, in tiers of 10k.]
  • Evidence: [“$29/mo includes 10,000 requests” — quoted from the pricing page, checked 2026-07-14]
  • Source: competitor-a-pricing

Sources

One row per source. Reliability is your honest read: primary (their own site or docs), secondary (reporting about them), or hearsay (forums, social). Checked by is whoever verified the link still says what we claim — someone other than the agent who added it.

Verification checklist

Before the mission is called done, someone (agent or human) checks every box against the doc as it stands:

• [ ] Every finding cites a source that exists in the sheet
• [ ] Every source has been opened and re-read by someone other than the agent that added it
• [ ] Findings that contradict each other are flagged in chat, not silently averaged
• [ ] The research question has a direct, one-paragraph answer written at the top of Findings

Coordination