Agent Skills Platform

The governed platform for building, verifying, and distributing agent skills.

Website ·
Installation ·
Worker runbook ·
Team marketplace ·
Product scope

Agent Skills Platform turns the way people already work into tested, installable agent
skills. Give it a prompt plus the evidence behind the work—spreadsheets, reports,
emails, screenshots, transcripts, links, or scripts—and it builds a reusable workflow
that an organization can review, publish, and improve.

After release, a skill can retain maintenance evidence without turning every log
into runtime prompt context: classified run evidence is captured in raw/, recurring
findings become evidence-linked draft patterns in wiki/, and only a separately
validated change may update the executable skill. This is a governed maintenance
record, not autonomous self-modification.

Create your first skill

I have expertise to turn into a skill — no code required

Open the AI agent you already use, attach examples of the work, and paste this:

/agent-skills-platform

Turn my monthly revenue-variance review into a reusable internal skill.
I attached past reports and the source spreadsheets. The decision is whether to
escalate a material variance. It must not modify source data.

The creator asks for the business decisions only you can authorize, builds and tests
the skill, and shows a representative result. When it is correct, say: “Publish
this to the Finance marketplace.”

When the workflow uses an API, database, MCP, codebase, or structured file, Semantic
Recon runs automatically before implementation and creates a pinned data contract.
Use ./install.sh --without-semantic-recon only for a deliberately local,
source-free installation.

Do not use Git, edit registry files, or run marketplace commands. If the creator is
not installed in your agent, send this section to your marketplace operator.

I run the marketplace

Use the governed team marketplace guide to admit,
approve, release, distribute, update, quarantine, and roll back tested skills.

I am evaluating the platform

Read the product scope,
organizational acceptance protocol, and
technical implementation guide.

Why teams use it

• Preserve expert judgment. A skill captures the question, evidence, decision,
  and success measure behind recurring work.
• Trust what is shared. Skills carry validation, security checks, evals, and a
  representative run before they are published.
• Learn without runtime bloat. Maintenance keeps evidence, draft patterns, and
  rejected changes separate from the concise instructions an agent executes.
• Govern team use. The marketplace provides ownership, approvals, versioned
  releases, discovery, rollback, quarantine, and compatibility evidence.

How work moves through the organization

SME supplies examples and approves the result
        ↓
Creator builds and verifies a skill
        ↓
Marketplace operator governs and publishes it
        ↓
Colleagues install an approved version and use it

The SME owns business meaning. The marketplace operator owns distribution and policy.
See roles and handoffs.

Read more when needed

Contributing

What happens behind the scenes

An Agent Skill is a reusable workflow package that guides an agent from a
recognized situation to a verified outcome. It can use retrieved knowledge, MCP
tools, APIs, deterministic scripts, and agent judgment, but it is not itself a
RAG system, MCP server, or agent runtime.

RAG supplies knowledge. MCP supplies capabilities. The harness supplies
execution. A skill organizes them into a governed path toward a verified
outcome.

Reason where interpretation is necessary. Execute and verify with deterministic
controls where reproducibility matters. External models, APIs, and changing data
may vary rather than promising identical outputs.

Humans establish meaning. The factory does not expect you to know the correct
prompt or semantic contract; it asks one bounded question at a time. The flow is:

• Messy problem
• Agent inspects evidence
• Proposed / conflicting meanings
• Human authority decision
• Interview READY
• Build, prove, publish

Every skill is checked as one connected system. The skill graph links its
instructions, scripts, evaluations, and expected outputs. Two structural
requirements confirm that every expected result is tested and every predictable
multi-step workflow has one reliable entry point. Four checks—specification,
pipeline, security, and evaluation schema—run in parallel. Finally, a
representative run proves that the skill produces a useful result. The graph also
enforces every_expected_is_reachable and deterministic_multistep_has_orchestrator.

See CONTRIBUTING.md and CODE_OF_CONDUCT.md.
Contributions require the contributor assignment.

License

MIT. See LICENSE. Copyright © 2026 Francy J G Lisboa, also known as
Charuto. See ownership.