Iago Leal
Community@psiagoleal · Brazil
Focused in engineering softwares developing.
Agent Skills by Iago Leal
Showing 16 vetted skills indexed across 1 GitHub repositories.
secrets-guard
Blocks shell commands that could expose credentials, tokens, or environment secrets.
gates-de-conclusao
Defines verifiable completion gates and explicit abort conditions for long-running agent tasks.
paralelizacao-em-grafo
Decides when to split work into parallel agent slices and orchestrates fan-out and fan-in execution.
delegacao-openai-compat
Delegates bulk LLM work to OpenAI-compatible endpoints via shell scripts with egress checks and key-safe configuration.
meeting-minutes
Generate meeting minutes and summaries from recordings or transcripts with on-premise processing.
adr-writer
Creates and maintains Architecture Decision Records in a standardized Status/Context/Decision/Consequences format.
atribuicao-de-falha
Diagnoses AI agent failures and assigns the correct control layer for each root cause.
delegacao-a-subagentes
Guides delegation of sensitive-data tasks to local subagents while keeping confidential data off remote models.
micro-ticket-planner
Breaks user stories and broad demands into self-contained micro-tickets for agent execution.
handoff-updater
Maintains a concise CURRENT-STATE.md handoff document synchronized with each commit.
novo-projeto
Installs an AI agent governance framework into new or existing repositories with profile selection and non-destructive updates.
caveman
Compresses agent responses into telegraphic style at three intensity levels while preserving technical substance.
limites-de-uso
Reads AI account quota windows from a local SQLite database and plans work around resets.
pr-review-guard
Applies a four-gate checklist to review pull requests before merge.
docling-local
Converts PDFs to Markdown via a local docling service, preserving LaTeX equations and extracted figures.
mockup-lab
Prototype UI variants in HTML/CSS, render screenshots via headless browser, and export approved designs to Penpot as SVG.
Frequently Asked Questions About Iago Leal
FAQPage SchemaWhat tasks can I accomplish with Iago Leal's skill framework?▼
You can enforce secrets non-exposure, define verifiable completion gates, parallelize work across isolated worktrees, delegate volume tasks to OpenAI-compatible endpoints like LiteLLM or Ollama, generate meeting minutes from recordings, write ADRs, review PRs against OWASP risks, convert PDFs to Markdown, and prototype UI mockups exportable to Penpot.
Who is the target audience for these skills?▼
Software engineers and engineering leads running agent-assisted development who need governance controls: failure attribution diagnostics, subagent delegation for confidential data, usage-quota planning, compressed communication modes, and handoff documentation synced per commit across shifts and developers.
How do I install and run this skill framework?▼
Use the novo-projeto skill to install the agent governance framework (profile plus skill library) into a new or existing repository. It selects a profile via decision tree, preserves prior content like CLAUDE.md or AGENTS.md, and leaves the project ready for non-destructive --update runs.
What are the prerequisites and dependencies?▼
Dependencies vary per skill: local Whisper for audio transcription, a local docling service for PDF conversion, an OpenAI-compatible endpoint (LiteLLM, vLLM, Ollama, LM Studio, OpenRouter) for delegation, headless browser rendering for mockups, and a self-hosted Penpot instance for design export.
How does the framework handle confidential data and cost control?▼
The delegacao-a-subagentes skill routes sensitive data (PII, credentials, proprietary code) only to local models, never remote ones. The limites-de-uso and caveman skills track 5-hour and 7-day quota windows and compress responses to reduce per-turn token cost and history drag.