Spotlight · the agent link
Your agent does the work.
The coding agent you already run pairs once over MCP, pulls approved briefs straight from the plan, and opens the fix as a PR in your repo. GTM re-verifies — and the score moves. Your keys never leave your box.
MCP-nativeconservative-diffapproval at dispatchyour repo · your keys
claude-code · paired to gtm.inc
$ gtm apply #482
pulled brief · security · rate limit
editing api/routes.py +38 −2
opened PR #482 → your-repo
→ status: running → PR open
$ gtm reverify #482
security 58 → 78 · readiness +6
$
Multi-lens review
Copy, pricing, security and journeys — each scored with evidence you can trace.
Readiness score & gates
A deterministic roll-up with beta (≥65) and launch (≥75) gates. Every point traces to a run.
Living plan & Gantt
Gaps become tasks on a timeline with gates, critical path and re-baselined versions.
Direct agent tasking
Brief any gap to your paired agent over MCP. Status flows back to every page.
Doc collation
Your repo & upload docs in one versioned workspace — rendered, diffed, searchable, shareable.
Prioritised focus
The highest-impact gaps ranked by score gain — always know what to do next.
360° assessment
Every angle on one declared ontology — product, infra, legal, data-governance — scored comparably.
Real-time collaboration
Bring your team into the plan — live presence, shared cursors, threads on any finding.
Best-practice review
Each score set against curated launch best-practice and an anonymised cohort baseline.
Activity feed
Runs, agent tasks, gates and score changes — one honest journal of the whole launch.
Human architecture review
A senior engineer reviews your architecture in the same plan your agent works from.
Shared reports
Tokened, revocable share links render your scorecard logged-out — with a clear source.
Lens engine
Each lens is a scoped assessment with its own rubric on proven run machinery — quotes, holds, hard caps. Deterministic and evidence-linked.
Assessment engine
Every product maps onto one declared, versioned ontology. Scores compare across products; absences are values, never guesses.
GTM-link persona
gtm-scout · gtm-fixer · gtm-assessor — one identity carrying the fences: read-only, secrets never leave your box, nothing without approval.