Agent Spec logo

Agent Spec

Organization
vstorm-co
agent-spec

Change the agent spec — add or remove a field, rename one, tighten a rule, bump SPEC_VERSION, change publish-time validation, or touch YAML export/import. Use whenever editing app/agents/spec.py or anything that reads a stored spec. Every published agent and every client git repository holds a copy of this format, so a change that only works forwards breaks agents nobody touched.

Overview

Publishervstorm-co
Repositoryagenticos
Skill nameagent-spec
Stars
137
Forks
27
Bundled files
Instructions only
LicenseApache-2.0
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

    Published by vstorm-co on GitHub. Read the source before you install it.

Installation

Install the Agent Spec AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/vstorm-co/agenticos/tree/main/.claude/skills/agent-spec
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/vstorm-co/agenticos.git /tmp/agenticos
mkdir -p .claude/skills
cp -r /tmp/agenticos/.claude/skills/agent-spec .claude/skills/agent-spec
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Agent Spec in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Agent Spec on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Agent Spec is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

The agent spec — the format you cannot break

app/agents/spec.py. Reference page: docs/reference/spec.md (generated from the docstrings, so the reasoning belongs in the docstring). Concepts: docs/concepts.md.

The most load-bearing type in the platform: the Builder edits it, the database versions it, the factory instantiates it, and clients export it as YAML into their own git repositories.

The constraint

Old specs must keep loading. A published agent's spec is stored as it was written. A change that only parses new documents is a 500 on every run of something nobody touched.

model_config = ConfigDict(extra="forbid") means an unknown key is a hard error, so this is not theoretical. SPEC_VERSION is currently 13.

What each change costs

ChangeCost
Add an optional field with a defaultFree. Old specs take the default
Add a required fieldNot allowed. Give it a default, or a mode="before" validator that supplies one
Rename a fieldA mode="before" validator that moves the old key. CapabilityBindingSpec._fold_the_knowledge_capabilitys_own_rename is the worked example
Remove a fieldDrop it in a mode="before" validator and log a warning — never let extra="forbid" raise on a stored spec. See _MODEL_SETTINGS_WITHDRAWN
Narrow a rule (shorter max, tighter pattern, new enum)The dangerous one — see below
Move behaviour to a capabilityFold the old value into a binding, don't just drop it. _with_thinking_binding is the pattern

Bump SPEC_VERSION whenever a stored document needs migrating, and write the migration in the same change.

Narrowing a rule is a data migration

A narrower rule does not only reject new input — it makes existing rows unreadable, and a Pydantic model that refuses to validate one field of one row takes down the whole listing endpoint with a 500.

IngestionConfig and its OCR language codes are the worked example outside the spec: ocr_language went from "anything 2–16 characters" to Tesseract's ^[a-z]{3}(\+[a-z]{3})*$, and every row written before it held "en". The data migration shipped in the same revision.

If you narrow anything in a spec field, the same applies: find every stored value first, and migrate it in the same change.

Two invariants a migration must preserve

Idempotence. An explicitly-set value wins over a migrated one, so re-reading a spec the validator already migrated changes nothing. Both existing validators are written this way; keep it.

Refusal at publish, never at run time. An unknown capability id, an ungranted scope, a tool_approval keyed on a tool that does not exist, an unrenderable tool name, a secret_id of the wrong kind or from another organization, a personal MCP connection — all refused while somebody is looking at a form. Validation added to the runner instead is a broken agent in production rather than a red field in the Builder.

app/services/agent_registry.py owns publish validation. _tool_override_problems is the shape to copy: collect every problem and report them together.

What the spec deliberately excludes

Anything about where it runs (surfaces, channels, exposures) and who may use it (owner, sharing). Those are deployment and access facts; keeping them out is what lets the same spec be exported, reviewed and reused across organizations.

It also carries no secret values — only secret_id references. A spec goes into a client's git repository.

YAML

to_yaml keeps spec order rather than sorting, so a diff reflects what changed rather than where it happens to sort, and UUIDs become strings so the file round-trips through any reader. from_yaml raises ValueError on a non-mapping document, because a list or a bare string reaches Pydantic as an unhelpful type error.

Round-trip is part of the contract: spec == AgentSpec.from_yaml(spec.to_yaml()).

Test

tests/test_agent_spec_and_factory.py. Every migration needs a test that loads a verbatim old document — not one built from the current model, which cannot fail. Plus the round-trip, the duplicate-capability refusal, and each publish refusal.

app/agents/** is held to 100%. See the backend-tests skill.

Frequently asked questions

What does the Agent Spec AI skill do?

Change the agent spec — add or remove a field, rename one, tighten a rule, bump SPEC_VERSION, change publish-time validation, or touch YAML export/import. Use whenever editing app/agents/spec.py or anything that reads a stored spec. Every published agent and every client git repository holds a copy of this format, so a change that only works forwards breaks agents nobody touched.

Why use Agent Spec on TypingMind?

Because you install it once and use it with any model. Agent Spec is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Agent Spec in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/vstorm-co/agenticos/tree/main/.claude/skills/agent-spec. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Agent Spec?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Agent Spec?

As many as you like. As long as a model supports skills, you can use Agent Spec with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Agent Spec AI skill free?

Yes. It is published on GitHub by vstorm-co under the Apache-2.0 license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇