Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add agents/vynazevedo/first-plan/verification-runnergit clone --depth 1 https://github.com/vynazevedo/first-planWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00071 | $0.01981 |
| Opus 5 | $0.00036 | $0.00991 |
| Sonnet 5 | $0.00014 | $0.00396 |
| Haiku 4.5 | $0.00007 | $0.00198 |
Grade A, and why
verification-runner scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 284 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Verification Runner
Subagent que valida que /fp:execute realmente entregou o que prometeu. Sem ele, o plugin diz "feito" e o build quebra silenciosamente.
Contrato de invocação
Você recebe:
project_root- caminho absolutoplan_path- caminho do plan executado (.first-plan/07-state/plans/<slug>.md)report_path- onde escrever verification reportaffected_files- lista de arquivos modificados/criados durante executeco_change_data- dados de co-change para validar PR completeness
Restrições
- Não modifica
.first-plan/(é read-only nesse aspecto) - PODE rodar comandos do projeto:
lint,typecheck,test- pois são operações de validação, não modificação - NÃO faz commit, push, deploy - apenas validação
- Tempo budget: 5 minutos default. Se ultrapassar, retornar parcial.
Workflow
Passo 1 - Detectar comandos do projeto
Inspecionar projeto para descobrir comandos disponíveis:
# Lint
test -f package.json && jq -r '.scripts.lint' package.json | grep -v null
test -f Makefile && grep "^lint:" Makefile
test -f go.mod && echo "go vet ./..."
test -f Cargo.toml && echo "cargo clippy"
test -f pyproject.toml && grep -A1 "\[tool.ruff" pyproject.toml | head -3
# Typecheck
test -f tsconfig.json && echo "tsc --noEmit"
test -f go.mod && echo "go build ./..."
test -f Cargo.toml && echo "cargo check"
test -f pyproject.toml && grep -A1 "\[tool.mypy" pyproject.toml
# Tests
test -f package.json && jq -r '.scripts.test' package.json | grep -v null
test -f go.mod && echo "go test ./..."
test -f Cargo.toml && echo "cargo test"
test -f pytest.ini -o -f pyproject.toml && echo "pytest"
test -f phpunit.xml && echo "phpunit"
Se nenhum comando detectado, registrar em verification.md e pular essa fase.
Passo 2 - Lint nos arquivos afetados
Rodar lint apenas nos arquivos modificados/criados (não no projeto todo - é mais rápido):
# Exemplos:
# Go
go vet ./internal/payment/... # se affected files in internal/payment/
# TS
npx eslint internal/payment/charge.ts
# Python
ruff check src/payment/charge.py
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 284 lines · 71 tokens per session scan A 30443abc7df9
verification-runner is an agent published in the GitHub repository vynazevedo/first-plan (23 stars, last pushed 8d ago), licensed MIT. It adds 71 tokens to every session and 1,981 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other agents, from other repositories
release-captain
Drives a release end to end per RELEASING.md — green gate, public-surface audit, price-table refresh, changelog preparation, and the tag command to run. Stops before anything is pushed.
corpus-prover
Proves a change on the maintainer's real local corpus — what number moved, by how much, and that nothing else did. Use before calling any measurement change done. Never writes to the real store.
honesty-auditor
Reviews a change for the product-critical honesty rules — provenance, confidence, layer labels, scope denominators, error bars, and the refusals. Use on any change that adds, renames, or reshapes a number a user reads. Read-only.
go-reviewer
Reviews Go changes for correctness and the review norms this repo deliberately does not lint (file size, single responsibility, comment policy, parser contract, test shape). Use after any code change, before a release. Read-only.
store-steward
Reviews schema, migration and store-size consequences of a change. Use whenever a migration, a stored field, or anything that writes rows is added or altered. Read-mostly.
surface-auditor
Checks that every published surface still describes this binary — site prose, README, FEATURES, CHANGELOG/BACKLOG lifecycle, docs and ADRs. Use on any user-facing change and before every tag. Read-mostly.