TrackCTL

Concept alignment — fresh start after desk-flip · 2026-05-21. This upgraded review page preserves the original concept decision, scope, risks, and next-step recommendation while adding the current alignment feedback and approval controls.

Alignment page: alignment/concept-exploration-trackctl.html Source skill: /concept-exploration Archived baseline: docs/history/archive/2026-06-01/114006/alignment/concept-exploration-trackctl.html Review state: pre-approval

Decision Summary

Concept verdict Focus TrackCTL on agent-configurable product analytics for PostHog.
Wedge AI coding agents can configure, deploy, and verify analytics without browser click-ops.
Immediate test Validate init, deploy, and check against a real PostHog Cloud project in week 1.

The original page was a compact concept checkpoint. This upgrade keeps the same strategic answer: start with a CLI-only TrackCTL that makes analytics setup legible to AI coding agents and defer the dashboard, SDK, hosting, auth, and enterprise surface area until the core zero-human-intervention test is proven.

Value Wedge

The only product analytics tool an AI coding agent can fully configure, deploy, and manage without ever opening a browser. An agent runs init, edits YAML, runs deploy, and analytics are live.

Claim Evidence From Original Page Inference Decision Impact
Agent-first configuration is the wedge. The page centers agents running init, editing YAML, and running deploy. The product should optimize for machine-operable setup before adding human UI surfaces. Approve or correct the concept anchor before downstream ICP and roadmap work rely on it.
Browser avoidance is part of the value proposition. The original value wedge explicitly says "without ever opening a browser." Click-ops elimination is not a nice-to-have; it is the differentiator to preserve. Keep CLI, YAML, and PostHog API work ahead of dashboard or auth expansion.

Primary Beneficiary

Agent-native developers — developers who routinely use AI coding agents (Claude Code, Cursor, Codex) and want their agent to handle analytics setup alongside code generation. Their pain is binary: their agent literally cannot set up analytics today.

Audience implication

The review page should not drift back to "solo devs broadly" as the primary beneficiary. The sharpened segment is developers who delegate implementation work to coding agents and need analytics setup to fit that workflow.

Coverage gap

This concept page names the segment but does not prove reachability, willingness to adopt, or buying behavior. Those questions remain for the ICP and customer-story artifacts, not this concept checkpoint.

Validation Gate

An AI agent (Claude Code or Codex) successfully runs init / deploy / check against a real PostHog Cloud project with zero human intervention.

Must happen in week 1.

Pass condition Agent completes the setup and verification loop without browser handoff.
Failure signal If human browser intervention is required, the primary wedge has not been proven.

Scope And Non-Goals

Ships now (CLI)
  • trackctl init
  • trackctl deploy
  • trackctl check
  • YAML config schema
  • Lockfile state management
  • PostHog API client
Deferred
  • Browser SDK
  • Managed hosting
  • Web dashboard
  • Enterprise governance
  • CI/CD integration
  • Auth layer

Infrastructure such as durable hosting, auth, a dashboard, and enterprise governance stays out of the first validation loop. It can be promoted only after the agent-native CLI journey proves the concept.

Riskiest Assumptions

Status Assumption Confidence Handling What Would Change It
HIGH A4: An AI agent can complete init/deploy/check with zero intervention. Highest-risk concept assumption and the week-1 validation target. A real PostHog run requiring human browser work would weaken the wedge.
MEDIUM A2: Agent-native developers are a meaningful and growing segment. Concept-level claim; ICP work should test reachable demand and pain intensity. Low reachable audience or weak urgency would force a narrower wedge.
MEDIUM A3: Click-ops analytics setup is painful enough to drive adoption. Strategic pain is plausible but needs customer/story evidence. If users tolerate existing click-ops, TrackCTL must reposition around reuse or governance.
RESOLVED A1: PostHog API covers enough for config-as-code (full CRUD confirmed). Original page treats API coverage as resolved. Unexpected API gaps during implementation would reopen this assumption.

What Changed From Old Brief

Dimension Old Brief Updated Concept Decision Impact
Scope 3-package monorepo (CLI + SDK + web) CLI-only, single package Reduce implementation surface until the CLI validation loop works.
Beneficiary Solo devs broadly Agent-native developers Target users whose agents currently cannot complete analytics setup.
Validation Phase 18 of 18 Week 1 Move the existential risk test to the front of the project.
Roadmap 18 phases planned upfront Build minimum to validate, then decide Keep downstream work conditional on validation evidence.

Recommended Next Step

Define the agent-native developer ICP

The original recommendation was /icp. In Codex command text, this maps to the ICP discovery skill, but the page preserves the original recommendation as page-specific context instead of turning it into a new execution instruction.

Preservation Checkpoint

Original Element Preserved Location Notes
Page title and concept date Header and document title Title remains "TrackCTL — Concept Alignment"; date remains 2026-05-21.
Value wedge and beneficiary Value Wedge and Primary Beneficiary sections Original wording is retained and supplemented with decision impact.
Validation gate and scope split Validation Gate and Scope And Non-Goals sections Week-1 gate and CLI/deferred lists remain intact.
Assumptions, old-brief changes, and next step Riskiest Assumptions, What Changed From Old Brief, and Recommended Next Step sections Risk labels and old-to-new diffs are preserved with added context for review.
Original generation note This checkpoint Generated by /concept-exploration; prior research archived at archive/research/.

Approval Gates

Value wedge decision

Should this page preserve the agent-native analytics wedge as the concept anchor?

Scope and non-goals

Is the CLI-first scope and deferred infrastructure list still the right concept boundary?

Validation gate

Is the week-1 zero-human-intervention PostHog validation gate strong enough?

Assumption confidence

Are the risk labels and confidence handling acceptable for this concept checkpoint?

Post-approval route

Should the preserved concept recommendation remain ICP discovery for agent-native developers?

Compile Review YAML

Use feedback YAML for concerns or clarification before final approval. Use final answers only when every required gate question is answered.

Five required gate questions remain unanswered.