Data dictionary
Every lead field is designed to make public-source research easier to review
The lead format is intentionally simple: enough structure for scoring and segmentation, enough context for a human to verify before outreach.
projectName
The public project, product, company, framework, or repository name.
- Used for deduping and quick scanning.
- May be normalized when a repo and product use different names.
website
The official website, docs site, launch page, or primary product page.
- Public pages are preferred over third-party profiles.
- Broken or unclear sites are flagged for review.
githubUrl
The GitHub repository or organization URL when a meaningful public repo exists.
- Used for activity, release, and community signals.
- Not required for closed-source companies.
category
The working market category used for routing the lead to the right buyer.
- Examples: MCP server, AI agent tool, LLM gateway, observability, security, evals.
- Categories can change after manual review.
growthSignal
A concise summary of public activity that makes the project worth watching.
- Examples include launch activity, changelog velocity, GitHub movement, community adoption, integrations, or hiring signals.
- It is a research signal, not a guaranteed buying event.
possibleNeed
A reason the team may eventually need a product or service from an AI infrastructure vendor.
- Examples: security review, gateway controls, observability, evals, deployment help.
- This is a hypothesis to verify before outreach.
leadScore
A practical priority score for review and outreach planning.
- Higher scores reflect category fit, freshness, source confidence, and actionability.
- Scores are not predictions of revenue or response rates.
contactAngle
A context-aware outreach angle that explains why this lead is relevant.
- Designed to reduce generic cold outreach.
- The sender should still personalize and verify before contacting.
source
The public source or source type that produced the lead.
- Examples: GitHub, official docs, product site, MCP directory, launch post, company page.
- Private inboxes and scraped personal data are not used.
verificationStatus
The current review state of the lead.
- Examples: public-source verified, needs manual review, monitor only.
- Used to decide whether a lead is ready for export or should stay in the watchlist.
evidence
A short note explaining which public sources support the lead record.
- Examples: official website, GitHub repository, docs page, public product page, or launch post.
- Evidence helps reviewers decide whether the record is ready to use or needs another manual check.
whyNow
The timing reason this project may be worth reviewing now instead of later.
- Examples: agent workflows moving toward production, MCP adoption, new integrations, or visible activity around a category.
- whyNow is context for prioritization, not proof that a company is ready to buy now.
bestFitBuyer
The buyer segment most likely to care about the lead based on its category and public signals.
- Examples: AI security, LLM gateway, AI observability, AI consulting, developer tools, or AI infrastructure investors.
- Used to route records into category-specific packs.
lastChecked
The review freshness marker for the lead record.
- Beta samples may use a manual review marker instead of a precise date.
- Paid packs should be checked close to delivery and can include a clearer date if required.
confidence
A qualitative confidence note based on source quality, category fit, and manual review status.
- High confidence means public sources clearly match the category and suggested angle.
- Medium or needs-review records should be checked again before outreach.
AI Agent Usage Signal (aiAgentUsageSignal)
A customer-facing note explaining why the project belongs in an AI agent, MCP, or LLM infrastructure watchlist.
- Grounded in public project context such as docs, repositories, registries, or product pages.
- Used to separate real agent-infrastructure relevance from generic AI noise.
Security Risk Signal (securityRiskSignal)
A bounded note about permissions, data access, integrations, auditability, or governance surfaces.
- This is not a claim that the project has a confirmed vulnerability.
- Security buyers should verify the source before using the angle in outreach.
Tool Stack Signal (toolStackSignal)
Visible technical context that helps route the record to security, gateway, observability, consulting, or investor teams.
- Examples: MCP server, agent framework, LLM gateway, registry evidence, transport, auth, or integration context.
- Blank or unclear stack context should stay in review.
Hiring / Budget Signal (hiringBudgetSignal)
A public hiring, pricing, enterprise, or budget-adjacent signal when one is visible.
- If no public signal exists, the field explicitly says no signal was found.
- It should never be treated as confirmed budget without source review.
Best Outreach Angle (bestOutreachAngle)
A human-review starting angle for respectful outreach based on the buyer lens and source evidence.
- Designed to help a seller personalize the first message.
- It is not an instruction to send automated outreach.
Suggested Email (suggestedEmail)
Draft-only email copy included to reduce blank-page work for careful outbound.
- Must be reviewed and edited by a human before sending.
- Includes opt-out language and should not be used for bulk campaigns.
Next step
See these fields in context
The sample page shows how the fields appear in a real beta lead table.