security-hardening

Use when a task is otherwise complete and you need final CI-parity gating; run golangci-lint at the end, fix findings in-scope, rerun validation, and enforce…

Run the final CI-parity gate after implementation/review so local completion matches CI outcomes, then report done only after required checks are green.

Name and block the specific failure modes of AI-generated code and AI-generated reports that pass a shallow check

Implement one task card at a time from `docs/PLAN.md` while preserving scope, architecture, and validation

Review a completed task implementation for compliance with the plan, the constitution, architecture, tests,

Visual and interaction design discipline for the venture product template — companion to `frontend-development`

Frontend implementation discipline for the venture product template (Task 46+) — SvelteKit front, Go API back,

Testing discipline for a system whose entire value proposition is "evidence-based completion; no self-reported

A general, multi-pillar review pass usable on any diff in this repo — broader than `task-review` (which is

Enforce a single, consistent Go coding standard across `internal/*`, `cmd/*`, and tooling, so that any agent's

This repository's whole product is AI agents executing tasks with real authority (SCM writes, deploys, billing).

Keep every task's implementation inside its complexity budget — correct, minimal, and maintainable, not merely