Other Entry Points

Not everything starts at the inbox. Each block below is one goal: the prompt to type and what really happens. Prompts marked shipped are the default_prompt of the skill (agents/openai.yaml).

See where things stand

Type (shipped):

Use $fireskill-progress para mostrar o progresso geral de todas as especificações. Orquestre transições automáticas com anúncio, execução e retomada na mesma conversa.

Or run it yourself:

bun run fireskill progress

Each spec is printed with its state and task and item counts. --json gives machine output, --watch prints again when a spec changes. The skill also offers node fireskills/skills/fireskill-progress/scripts/progress.mjs . (with --slug <slug> for one spec) and reads the context monitor and build_documentation.mjs --check to show documentation pending. It changes no gate and no checkbox. A token figure is always labeled estimated unless you provide measured usage.

Talk about a spec

Type (shipped):

Use $fireskill-interviewer para conversar sobre a spec no estado atual e registrar apenas respostas confirmadas.

Works on specs in draft, defined, planned, in-progress and review, and focuses the conversation by folder (for example, in defined it looks for ambiguity that affects plan, data, security or tests). It records confirmed answers in the right section and adjusts Effort. It does not approve gates, run tasks or move the spec. It does not interview to change a completed spec.

Decide what the product must store

Type (shipped):

Use $fireskill-data-discovery para conversar em linguagem simples sobre o que o sistema precisa guardar e registrar apenas respostas confirmadas em DATABASE.md.

Asks in everyday language ("O que você precisa lembrar sobre cada pedido, pessoa ou atendimento?", "Quem pode ver ou mudar essa informação?"), proposes a simple way to record each item and asks you to confirm, and writes only confirmed answers to .fireskills/DATABASE.md. It does not choose technology. The backlog calls it automatically when the idea needs stored data.

Start from an MVP document

Type (shipped):

Use $fireskill-mvp-milestone-interviewer para importar o MVP como M01 e gerar backlogs e specs Draft somente para requisitos de desenvolvimento, sem criar Inboxes nem copiar o contexto de negócio.

Requires an MVP.md at the repository root (there is none in Nuxfire out of the box). The skill runs importar_mvp.mjs --root <raiz> once: it creates specs/milestones/M01.md and, for each item that is a software deliverable, a backlog item and a Draft spec with Definition Gate: Pending. Vision, audience, principles and metrics stay only in MVP.md. It then interviews each backlog item, asking only what the MVP did not answer.

After the MVP is accepted:

Use $fireskill-roadmap-milestone-interviewer para entrevistar a evolução pós-MVP e propor milestones sem alterar o núcleo aceito sem confirmação.
Use $fireskill-milestone-governor para analisar milestones, manter o índice derivado de specs e backlog e sugerir ajustes confirmáveis.

You can also refresh the index yourself: bun run fireskill milestones sync updates specs.md and the progress of each milestone.

Keep the project context current

Type (shipped, one per file):

Use $fireskill-aux-stack para monitorar o stack e atualizar .fireskills/STACK.md sem perder dados existentes.
Use $fireskill-aux-rules para revisar mudanças e registrar regras em .fireskills/RULES.md preservando o conteúdo atual.
Use $fireskill-aux-database para monitorar persistência e atualizar .fireskills/DATABASE.md após mudanças de dados.

The flow calls them when the monitor asks. See Project Context.

Record and reuse lessons learned

Type (shipped):

Use $fireskill-lessons para registrar uma lição aprendida com sinal, resumo, regra preventiva e origem em .fireskills/LESSONS.md, ou para consultar as lições confirmadas antes de planejar uma fatia.

The commands it uses (you can run them):

bun run fireskill lesson add --signal gate_fail --summary "Import relativo no schema" --rule "Importar schemas de layers pelo pacote @nuxfire/*" --source "0042-billing"
bun run fireskill lesson list          # confirmed lessons
bun run fireskill lesson list --all    # plus candidates and quarantined
bun run fireskill lesson status
bun run fireskill lesson prune         # removes expired candidates

--signal is one of ac_gap, surviving_mutant, spec_precision_gap, spec_deviation, gate_fail; --summary, --rule, --source are all required. A lesson starts as candidate and is promoted to confirmed automatically when it recurs in two or more distinct sources.

Attack a spec, decide something risky, review a change

The prompts below are the shipped ones (agents/openai.yaml of each skill).