Skip to content

Docs

The reference.

How Gamify It connects, what each tool does, the tracking spec it measures against, and what it does with your data.

Endpoint

One URL, every assistant.

MCP URL:https://mcp.gamifyit.co/mcp
Transport
Streamable HTTP at one URL for every assistant.
Sign-in
OAuth, once, when you connect, with a consent screen every time an assistant asks for access.
Subscription
Checked on every call, never at connect. Without an active subscription every tool returns one activation message as a normal tool result.
Results
Laid out by the server, small (about 8k tokens or less) and returned in under 30 seconds.
Writes
Only the save tools write. Every write is non-destructive and idempotent with an operation id. There is no delete tool: deletion and exports live in your account.

Tools

20 tools, each one independent.

Your assistant picks the tools; you never call them by name. Each returns its result already laid out, plus compact structured data and citations with canonical URLs.

Knowledge

Search and read the knowledge base: techniques, cases, metrics and the playbook for any Stage Map cell.

  • searchRead-only

    Search techniques, frameworks, cases, metrics and playbooks.

  • fetchRead-only

    Read one knowledge entry: paraphrase, sources with pages, grade and caveats.

  • get_playbookRead-only

    The playbook for one Stage Map cell: in season, not now, lead techniques, KPIs, cases.

Locate and diagnose

Ask the questions worth the most, plan what to inspect, place the product on the Stage Map and score only the lenses the binding stage needs.

  • next_questionsRead-only

    The 3-5 questions that would raise confidence the most, and what each answer unlocks.

  • get_inspection_planRead-only

    What your assistant should inspect, for its mode and stack: repo checklist, screenshots, exports or interview.

  • locate_stageRead-only

    Stage Fix, Leak Map and Constraint Rules: the Stage Card, with a signed context token.

  • score_hookRead-only

    Scores the Hook links against the evidence, capped by what the evidence can support.

  • score_drivesRead-only

    The motivation profile of one key action in one phase, with caps and the cheapest missing drive.

  • map_decision_pointsRead-only

    Persuasion coverage at sign-up, permission, upgrade, checkout, renewal and referral.

  • score_monetizationRead-only

    Monetization Readiness: now, depth only, or not yet, with the levers in season.

Decide and design

Rank the next moves with every factor shown, size the experiment to your traffic and plan the tracking it needs.

  • rank_opportunitiesRead-only

    The Opportunity Board: a Next Move, two alternates and the Not-now list, with Size, Odds, Fit and Cost.

  • design_experimentRead-only

    An Experiment Card: KPIs, guardrails, sample size, weeks to read, or the low-traffic design.

  • plan_instrumentationRead-only

    The tracking-plan diff against Event Spec v1, and the minimal experiment kit when there is no flag tool.

Measure

Math on aggregate tables: cohorts, funnels, habit metrics and experiment readouts with confidence intervals.

  • compute_metricsRead-only

    Rates, curves and readouts from aggregate tables, including SRM and guardrail checks.

Status and memory

The weekly status, and the saves that build your history. Saving is always your choice.

  • get_statusRead-only

    This week’s task, what changed and readouts due, or any saved section of your history.

  • save_sessionSaves

    The first save, in one confirmation: passport, Stage Card, board and experiment.

  • save_productSaves

    Create or update the Product Passport.

  • save_evidenceSaves

    Store evidence and aggregate metric snapshots; update open questions and tasks.

  • save_assessmentSaves

    Store a Stage Card or a lens snapshot, with its diff.

  • save_experimentSaves

    Create or update an experiment, record arm counts, a decision or a learning.

Slash commands

Shortcuts in Claude Code.

The Claude Code plugin adds these commands and five skills. A bare connector has every tool but not the commands.

  • /gamifyit:startFirst session: where your product is stuck and the next move
  • /gamifyit:statusThis week’s task, what changed, readouts due and stale evidence
  • /gamifyit:scanA read-only repo scan; no code leaves your machine
  • /gamifyit:nextDiagnose the binding stage and pick the Next Move
  • /gamifyit:experimentSize the move as an Experiment Card, then build it on your go-ahead
  • /gamifyit:readoutRead an experiment at its horizon and record the decision
  • /gamifyit:reviewThe monthly or quarterly review: re-stage, re-score, drift checks

Resources

Things your assistant can read.

Handy while implementing a move, for example with an @ mention in Claude Code.

  • gamifyit://kb/{id}One knowledge-base entry
  • gamifyit://playbook/{maturity}/{stage}The playbook for a Stage Map cell
  • gamifyit://event-spec/v1Event Spec v1, for implementing tracking
  • gamifyit://product/{id}/passportA saved Product Passport
  • gamifyit://product/{id}/tracking-planThe saved tracking plan
  • gamifyit://product/{id}/learning-logWhat you tried and learned

Event Spec v1

The tracking plan Gamify It measures against.

Events are named object_action in snake_case, with stable English names. You don’t need all of them: the instrumentation plan asks only for what your next decision needs.

Every event carries:

  • user_id
  • account_id
  • plan
  • platform
  • country
  • acquisition_channel
Event Spec v1 events by family
FamilyEvents and key propertiesUnlocks
Identity
  • user_signed_up {trigger_type, trigger_form, channel, campaign, plan, platform}
  • account_created
  • user_invited {channel}
  • invite_accepted {channel}
Cohorts by trigger type and channel; account-level activation.
Setup
  • onboarding_step_completed {step_id}
  • permission_granted {type}
  • integration_connected
The step funnel, time to value and permission timing.
Value
  • value_event {name, quantity}
The North Star, the activation threshold, habitual rate and L28.
Session
  • session_started {trigger_type, trigger_form, channel, trigger_id}
Unprompted share (the internal-trigger proxy) and the trigger-type mix.
Messaging
  • message_sent {channel, trigger_type, campaign_id, trigger_id}
  • message_delivered {channel, trigger_type, campaign_id, trigger_id}
  • message_opened {channel, trigger_type, campaign_id, trigger_id}
  • message_clicked {channel, trigger_type, campaign_id, trigger_id}
  • channel_disabled {channel}
  • unsubscribed {channel}
Trigger performance, holdouts and the rented-engagement monitor.
Reward
  • reward_shown {type, reward_id}
Return rate after a reward, and novelty decay.
Investment
  • investment_made {form, loads_trigger, trigger_id}
Stored value per user, loaded-trigger yield and hook cycle time.
Social
  • share_created {surface}
  • referral_link_opened
  • referral_converted
  • kudos_given
K-factor, invite-to-activation and viral cycle time.
Commerce
  • paywall_viewed
  • price_shown {amount, step}
  • trial_started
  • subscription_started {reason}
  • subscription_renewed {reason}
  • subscription_upgraded {reason}
  • subscription_downgraded {reason}
  • subscription_paused {reason}
  • subscription_cancelled {reason}
  • refund_issued
  • chargeback_received
  • order_completed {value, discount_code, payment_method}
Conversion by tenure, revenue quality and discount dependence.
Feedback
  • survey_answered {instrument, score}
PMF score and habit strength surveys.
Experiment
  • experiment_assigned {experiment_id, variant, unit, venue}
  • experiment_exposed {experiment_id, variant, unit, venue}
Sample-ratio checks and readouts.

The one property most products miss

Privacy and security

Aggregates in, advice out.

A summary for builders and the admins who approve connectors.

  • Gamify It receives summaries and aggregates only: never source code, raw event rows or personal data about your users. Payloads that look like code or row-level data are rejected.
  • Your assistant reads your code, pages and data on your side, under your agreement with that assistant.
  • Diagnostic tools store nothing. A signed, tamper-proof context token carries state between calls until you choose to save.
  • Database steps prefer your own vendor connectors or a read-only role; production credentials never need to leave your machine.
  • Every query is bound to your account inside the database (row-level security that fails closed).
  • Logs and error reports are scrubbed of customer content. Exports and deletion are in your account.

Subprocessors

  • VercelWebsite and MCP endpoint hosting
  • SupabaseDatabase hosting (accounts, products, saved history)
  • StripePayments, invoices and sales tax
  • ResendTransactional email (sign-in, receipts, the Monday brief)
  • SentryError monitoring, with customer content scrubbed
  • MetaAd measurement on the website (the Pixel), only where enabled