{
  "markdown": "# Agent Skills Platform\n\n**Turn a real workflow into a tested, installable agent skill—then publish it safely to your team.**\n\n![Agent Skills Platform: question to tested skill to governed marketplace](docs/assets/agent-skills-platform-social-preview.png)\n\n[![CI](https://github.com/FrancyJGLisboa/agent-skills-platform/actions/workflows/ci.yml/badge.svg)](https://github.com/FrancyJGLisboa/agent-skills-platform/actions/workflows/ci.yml)\n[![Agent Skills Open Standard](https://img.shields.io/badge/Agent%20Skills-Open%20Standard-blue)](https://github.com/agentskills/agentskills/blob/main/docs/specification.mdx)\n[![Platforms](https://img.shields.io/badge/installs%20on-17%20platforms-7c3aed)](docs/INSTALL.md)\n[![Version](https://img.shields.io/badge/version-6.1.0-brightgreen)]()\n[![License: MIT](https://img.shields.io/badge/License-MIT-green.svg)]()\n\n[Website](https://francyjglisboa.github.io/agent-skills-platform/) ·\n[Installation](docs/INSTALL.md) ·\n[Worker runbook](docs/WORKER_RUNBOOK.md) ·\n[Team marketplace](docs/TEAM_MARKETPLACE.md) ·\n[Product scope](docs/PRODUCT_SCOPE.md)\n\nAgent Skills Platform turns the way people already work into tested, installable agent\nskills. Give it a prompt plus the evidence behind the work—spreadsheets, reports,\nemails, screenshots, transcripts, links, or scripts—and it builds a reusable workflow\nthat a person can inspect, an organization can review, and a team can safely reuse.\n\n```text\n/agent-skills-platform\n\nTurn my monthly revenue-variance review into a reusable internal skill.\nI attached past reports and the source spreadsheets. The decision is whether to\nescalate a material variance. It must not modify source data.\n```\n\nThe result is not just a prompt: it is an installable skill package with instructions,\nfunctional scripts when needed, evals, security checks, a representative-run record,\nand a correction path for real-world learning.\n\nAfter release, a skill can retain maintenance evidence without turning every log\ninto runtime prompt context: classified run evidence is captured in `raw/`, recurring\nfindings become evidence-linked draft patterns in `wiki/`, and only a separately\nvalidated change may update the executable skill. This is a governed maintenance\nrecord, not autonomous self-modification.\n\n## Create your first skill\n\nChoose the path that matches what you need today.\n\n**First question:** is this skill just for you, or will teammates install or reuse the\nskill itself?\n\n- **Just for you** — create, verify, and install it privately. No marketplace setup.\n- **My team** — create or select a governed GitHub/GitLab marketplace first, so the\n  skill inherits real ownership and approval rules.\n\nTeammates receiving a report or queue does not require a marketplace. Use the team\npath only when teammates will install or reuse the skill.\n\n### I have a workflow — no code required\n\nOpen the AI agent you already use, attach examples of the work, and paste this:\n\n```text\n/agent-skills-platform\n\nTurn my monthly revenue-variance review into a reusable internal skill.\nI attached past reports and the source spreadsheets. The decision is whether to\nescalate a material variance. It must not modify source data.\n```\n\nThe creator asks for the business decisions only you can authorize, builds and tests\nthe skill, and shows a representative result. When it is correct, say: **“Publish\nthis to the Finance marketplace.”**\n\nWhen the workflow uses an API, database, MCP, codebase, or structured file, Semantic\nRecon runs automatically before implementation and creates a pinned data contract.\nUse `./install.sh --without-semantic-recon` only for a deliberately local,\nsource-free installation.\n\nDo not use Git, edit registry files, or run marketplace commands. If the creator is\nnot installed in your agent, send this section to your marketplace operator.\n\n### I run the marketplace\n\nUse the [governed team marketplace guide](docs/TEAM_MARKETPLACE.md) to admit,\napprove, release, distribute, update, quarantine, and roll back tested skills.\n\n### I am evaluating the platform\n\nRead the [product scope](docs/PRODUCT_SCOPE.md),\n[organizational acceptance protocol](docs/ORGANIZATIONAL_ACCEPTANCE.md), and\n[technical implementation guide](docs/TECHNICAL_OVERVIEW.md).\n\n## Why teams use it\n\n- **Preserve expert judgment.** A skill captures the question, evidence, decision,\n  and success measure behind recurring work.\n- **Trust what is shared.** Skills carry validation, security checks, evals, and a\n  representative run before they are published.\n- **Learn without runtime bloat.** Maintenance keeps evidence, draft patterns, and\n  rejected changes separate from the concise instructions an agent executes.\n- **Govern team use.** The marketplace provides ownership, approvals, versioned\n  releases, discovery, rollback, quarantine, and compatibility evidence.\n\n## How work moves through the organization\n\n```text\nSME supplies examples and approves the result\n        ↓\nCreator builds and verifies a skill\n        ↓\nMarketplace operator governs and publishes it\n        ↓\nColleagues install an approved version and use it\n```\n\nThe SME owns business meaning. The marketplace operator owns distribution and policy.\nSee [roles and handoffs](docs/TEAM_MARKETPLACE.md#roles-and-handoffs).\n\n## See a verified result\n\nThe repository includes a live, read-only weather briefing example with a source-linked\nresult and verification evidence. Start with the\n[verification record](docs/verification/2026-08-27-live-weather-briefing.md), then\ninspect the [skill package](references/examples/live-weather-briefing-skill).\n\n## Read more when needed\n\n| Need | Read |\n|---|---|\n| Install on a supported AI tool | [Installation](docs/INSTALL.md) |\n| Create, correct, and hand off a first skill | [Worker runbook](docs/WORKER_RUNBOOK.md) |\n| Run a governed internal marketplace | [Team marketplace](docs/TEAM_MARKETPLACE.md) |\n| Understand scope and product boundaries | [Product scope](docs/PRODUCT_SCOPE.md) |\n| Review architecture, validation, and technical controls | [Technical overview](docs/TECHNICAL_OVERVIEW.md) |\n| Contribute | [Contributing](CONTRIBUTING.md) |\n\n## What happens behind the scenes\n\nAn Agent Skill is a reusable workflow package that guides an agent from a\nrecognized situation to a verified outcome. It can use retrieved knowledge, MCP\ntools, APIs, deterministic scripts, and agent judgment, but it is not itself a\nRAG system, MCP server, or agent runtime.\n\nRAG supplies knowledge. MCP supplies capabilities. The harness supplies\nexecution. A skill organizes them into a governed path toward a verified\noutcome.\n\nReason where interpretation is necessary. Execute and verify with deterministic\ncontrols where reproducibility matters. External models, APIs, and changing data\nmay vary rather than promising identical outputs.\n\nHumans establish meaning. The factory does not expect you to know the correct\nprompt or semantic contract; it asks one bounded question at a time. The flow is:\n\n- Messy problem\n- Agent inspects evidence\n- Proposed / conflicting meanings\n- Human authority decision\n- Interview READY\n- Build, prove, publish\n\nEvery skill is checked as one connected system. The skill graph links its\ninstructions, scripts, evaluations, and expected outputs. Two structural\nrequirements confirm that every expected result is tested and every predictable\nmulti-step workflow has one reliable entry point. Four checks—specification,\npipeline, security, and evaluation schema—run in parallel. Finally, a\nrepresentative run proves that the skill produces a useful result. The graph also\nenforces `every_expected_is_reachable` and `deterministic_multistep_has_orchestrator`.\n\nSee [CONTRIBUTING.md](CONTRIBUTING.md) and [CODE_OF_CONDUCT.md](CODE_OF_CONDUCT.md).\nContributions require the [contributor assignment](CONTRIBUTOR_ASSIGNMENT.md).\n\n## License\n\nMIT. See [LICENSE](LICENSE). Copyright © 2026 Francy J G Lisboa, also known as\nCharuto. See [ownership](COPYRIGHT.md).\n",
  "bytes": 7947,
  "sha": "5e9e95037596b15522969013bc77b309b906f8d784d691854b17c585bbbfcd01",
  "repo_slug": "francyjglisboa/agent-skills-platform",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/plg_francyjglisboa_agent_skills_platform_dbe65cef/readme"
}