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 skills/thaleslaray/plugins/pesquisanpx skills add thaleslaray/plugins --skill pesquisagit clone --depth 1 https://github.com/thaleslaray/pluginsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/thaleslaray/plugins/pesquisa)<a href="https://agentmods.dev/skills/thaleslaray/plugins/pesquisa"><img src="https://agentmods.dev/badge/skills/thaleslaray/plugins/pesquisa.svg" alt="Measured on agentmods" height="20"></a>What 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 | $0.00210 | $0.07071 |
| Opus 5 | $0.00105 | $0.03535 |
| Sonnet 5 | $0.00042 | $0.01414 |
| Haiku 4.5 | $0.00021 | $0.00707 |
Grade B, and why
pesquisa scanned grade B with 2 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 4d 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.
Reads agent configuration directoriesmediumAgent snooping
.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.
- Se não sabe onde está: rodar `grep -r "perplexity" ~/.claude/ --include="*.json" -l` Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
3. Bash: `curl -s "https://r.jina.ai/<url>"` — segundo fallback, grátis sem conta, markdown limpo, ~10s How it starts
The opening of the file, as written. The whole thing — 534 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Pesquisa em Funil v3
Evolução do /pesquisa (v2) com 3 técnicas roubadas de skills do marketplace:
| Técnica | Origem | O que muda |
|---|---|---|
| CRAAP scoring | claim-investigation | Score 0-100 por critério, não só tier |
| Phase Gate por fatos-chave | claim-investigation | ≥2 fontes independentes por fato central |
| Domain filters | Tavily pattern | Queries restritas a domínios confiáveis |
O Funil v3
Nível 1: Varredura (perplexity_search paralelo)
│ Apresentar → usuário aprova direção (HITL)
▼
Nível 2: Análise (perplexity_reason) + identificar fatos-chave + sub-tópicos
│ Apresentar + listar sub-tópicos + fatos-chave → usuário escolhe (HITL)
▼
Nível 3: Deep Dive PARALELO (N × perplexity_research + firecrawl)
│ Phase Gate v3: sub-perguntas respondidas? + fatos-chave em ≥2 fontes? + CRAAP?
│ Se não → gap filling focado
▼
Relatório final → salvo como ~/pesquisas/[tema]-[data].md
Regra inviolável: human-in-the-loop entre cada nível. Não pular sem aprovação.
★ v3.1 — defesa em profundidade pós-incidente 2026-04-29
Esta skill foi auditada após dogfood que revelou /pesquisa rodando "internamente" (modelo se autovigia), gerando relatórios com seções obrigatórias ausentes e CRAAP inflacionado. Mudanças:
- Phase Gate via subagent independente (Passo 5): não é mais
executar internamente, éAgent({subagent_type: Explore, ...})produzindoverification_report.json - CRAAP rubric calibrada com 3 exemplos por Tier (Passo 4) — sem isso, scores convergem pra 88-95
- Definição programática de "fonte independente": domínios distintos + não-duplicação de paper primário
- PostToolUse hook validador em
~/.claude/skills/pesquisa/scripts/validate-report.py— bloqueia Write em~/pesquisas/*.mdse relatório não tiver todas seções obrigatórias OU se >50% fatos têm <2 fontes
Bloco contrastive (good vs bad fluxo)
❌ Execução RUIM (incidente 2026-04-29):
1. Spawn 5 perplexity_search paralelos → ok
2. Apresentar Nível 1 + AskUserQuestion → ok
3. Receber resposta → ok
4. perplexity_reason no Nível 2 → ok
5. AskUserQuestion sub-tópicos → ok
6. perplexity_research × 4 paralelos → ok
7. Phase Gate: "executar internamente" → modelo PULA, marca 5/5 checks sem evidência
8. Escrever relatório → modelo PULA seção "Contradições Identificadas" silenciosamente
9. CRAAP scores: 95, 92, 90, 88 → distribuição irreal de auto-justificação
10. F4 com "1 substack + 2 LinkedIn + 1 video" → marcado como ALTO sem ≥2 Tier A/B
What ships with it
3 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 4d ago First seen · 534 lines · 210 tokens per session scan B 6c54d91596f8
pesquisa is a skill published in the GitHub repository thaleslaray/plugins (2 stars, last pushed 3mo ago), licensed MIT. It adds 210 tokens to every session and 7,071 once invoked, about $0.0011 per session on Opus 5. A static security scan graded it B with 2 findings (reads agent configuration directories, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
systematic-debugging
Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes.
brainstorming
You MUST use this before any creative work - creating features, building components, adding functionality, or modifying behavior. Explores user intent, requirements and design before implementation.
auto-perf-optimize
Run agent-driven VS Code performance or memory investigations. Use when asked to launch Code OSS, automate a VS Code scenario, run the Chat memory smoke runner, capture renderer heap snapshots, take workflow screenshots, compare run summaries, or drive a repeatable scenario before heap-snapshot analysis.
chat-perf
Run chat perf benchmarks and memory leak checks against the local dev build or any published VS Code version. Use when investigating chat rendering regressions, validating perf-sensitive changes to chat UI, or checking for memory leaks in the chat response pipeline.
chat-pet-sprite-creation
Use when creating or changing VS Code chat pet sprite art, sprite sheets, state animations, eye treatments, Stable/Insiders variants, or pet transitions under src/vs/workbench/contrib/chat/browser/widget/media/chatPet.
cpu-profile-analysis
Analyze V8/Chrome CPU profiles (.cpuprofile) and DevTools trace files (Trace-.json). Use when: profiling performance, investigating slow functions, comparing code paths, finding bottlenecks, analyzing timeToRequest, understanding call trees from sampling profiler data, analyzing layout/paint/rendering, investigating…