Analyze collected Campaigns post metrics for one social account. Use for weekly performance reports, retrospective rankings, or evidence-based content experiments. Do not use to publish content, schedule reports, or claim causation.
1 skill
Use at the very start of a large plan, before any drafting — put every primary source on disk, confirm scope, resolve unknown terms, audit the code with path:line, screenshot the current product, and probe live state (branch protection, CI time, release gates, tools, credentials). Also use for every revision (v2, v3), which re-reads the sources and not the previous version. Skip for a plan whose only source is one short message you can quote whole.
Plugin plan-ground-truth from marketplace devfellowship in devfellowship/skills. Commands read from the repo's manifest — 1 month ago.
plan-ground-truthUse at the very start of a large plan, before any drafting — put every primary source on disk, confirm scope, resolve unknown terms, audit the code with path:line, screenshot the current product, and probe live state (branch protection, CI time, release gates, tools, credentials). Also use for every
Analyze collected Campaigns post metrics for one social account. Use for weekly performance reports, retrospective rankings, or evidence-based content experiments. Do not use to publish content, schedule reports, or claim causation.
1 skill
Use BEFORE writing a plan for large software work — a new app, a redesign, a mobile version, a cross-service pipeline or integration, anything that will become 3+ PRs or be executed by other agents. Routes through the pack in a fixed order (ground truth → two-planner draft → acceptance gates → PR slicing → readiness review → execution handoff), names the file each phase produces and the hard gates a plan must pass before it is published. Skip for a one-file bug fix, a question, or a small edit to a plan that already exists.
1 skill
Use sempre que for abrir uma task de trabalho pela ferramenta `create_task` do MCP — inclusive quando a task nasce de um plano, de uma conversa ou de uma PR. Garante que a task nasça com etapa, com o "por quê" e ligada ao plano certo. Não use para editar task existente.
1 skill
Use when writing the acceptance criteria, rollout and risk sections of a large plan, and as the checklist a reviewer attacks a finished draft with. Turns every prose promise into a gate that can fail on its own — checks on the running build, a viewport × input × content matrix, a default-path and flag-exit rule, shared-state isolation, the human who must see each output, measured limits, test hygiene and real-device proof. Pipelines, webhooks and integrations use non-ui-gates.md. Skip only for a pure refactor with no user-visible output, no flag, no limit and no shared state.
1 skill
Use to draft a large plan with two independent planner agents — shared brief, blind drafts from different angles and models, an optional cross-read, a merge that picks instead of averaging, an adversarial review that tries to knock the plan down, a written dialog where the author may argue back, and a consistency pass with a draft verdict. Use after plan-ground-truth. Skip for editing an existing plan; for a small plan use the single-planner prompt in prompts.md.
1 skill
Use when a large plan is about to be executed by agents or people, and when resuming execution after a crash or handoff — executor briefs by section reference, who merges, verification tools by path, resource budget, requester corrections mid-flight, and the definition of done (seen running on the path real users get). Skip for executing a one-PR plan yourself.
1 skill
This page updates itself from your repo. Add the badge to your README, or get it Featured to pin it above the organic results with a followed link to your site.