New: 98,000+ skills indexed from GitHub repos with 100+ stars · every skill safety-checkedBrowse top skillsTop skills
Esc to closeBrowse all skills
Submit a skillBrowse skills
Submit a skillBrowse skills

Agent skill

FlowCraft Config Authoring

by GizClaw in GizClaw/flowcraft

Writes and validates a complete FlowCraft agent deployment: deploy.yaml, runtime settings, resource sub-documents and graph JSON, checked against a structural validator.

MITAuto-check passedAI & LLM Engineering

Install FlowCraft Config Authoring

skills CLI
$ npx skills add GizClaw/flowcraft --skill flowcraft-config -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install GizClaw/flowcraft flowcraft-config --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/GizClaw/flowcraft.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/flowcraft-config .claude/skills/flowcraft-config && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
flowcraft-config
GitHub stars
417
Token cost
~2.3k tokens
SKILL.md length
995 words
Files
27 (incl. scripts, references, assets)
Skills in repo
1
Repo updated
First seen
Licence
MIT

At a glance

Writes and validates a complete FlowCraft agent deployment: deploy.yaml, runtime settings, resource sub-documents and graph JSON, checked against a structural validator.

  • Works in 8 steps: Scope the deployment. Collect the… → Draft the deployment document.… → Write sub-documents. Read → …
  • Writing a new FlowCraft deployment configuration
  • SKILL.md covers Workflow, Cross-file invariants, Templates and Compatibility and versioning, plus 1 more section
  • Fixing a FlowCraft config build failure such as a missing route policy

What it does

The agent first scopes the deployment by listing the agents, capabilities and runtime needs (sessions, resume, event bus, checkpoint stores) required, then drafts the deployment document in the order resources, agents, runtime, deciding between whole-resource and item-level dependency references. Sub-documents for workspace, sandbox and other resources follow the owning module's schema, with workspace written before sandbox since custom sandbox runners may depend on it, and memory implementation details left to the app's own registered modules.

Graph JSON wiring prefers routing the inference.Router into the graph engine over pinning a model directly, requires the nested id form for model references, and needs script nodes to carry runtime and source fields with edges wired back to the inference node after tool nodes. A bundled shell script runs an L2 structural validator that strictly decodes the deployment document, the runtime subtree and the graph definition, rejecting unknown fields and checking shapes, but it does not build resources, resolve settings files, decode node configs or make provider calls. A minimal runnable deployment template is included as a starting point.

When your agent uses it

  • Writing a new FlowCraft deployment configuration
  • Fixing a FlowCraft config build failure such as a missing route policy
  • Wiring graph JSON nodes for an agent's capabilities

Example prompts

  • “Write a minimal FlowCraft deploy.yaml for an agent with chat and tool capabilities.”
  • “My graph build fails with "not registered". Fix the FlowCraft config.”
  • “Validate this FlowCraft deployment file and fix any structural errors.”

Requirements

  • Go, since the config validator pins the FlowCraft core module in its go.mod

Workflow steps

8 steps, taken from the first numbered list in SKILL.md.

  1. Scope the deployment. Collect the agents, capabilities (chat,
  2. Draft the deployment document. deploy.yaml is the convention
  3. Write sub-documents. Read
  4. Write graph JSON. Read references/graph.md.
  5. Assemble the Craft application. When the deployment runs inside a
  6. Validate structurally with L2. Run
  7. Fix errors. Each failure is prefixed [parse], [graph],
  8. Hand off. Report the structural validation result and what L2

What it can do on your machine

Read from SKILL.md and the folder at commit 88944e3. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Ships 1 file in scripts/, which the agent can run.

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

FlowCraft Config Authoring loads about 2.3k tokens when it runs, and up to ~24k if it reads all its reference files. Until then it costs about 173 tokens; SKILL.md has 995 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~173
When it runs · the whole SKILL.md, loaded when a task matches
~2.3k
With references · SKILL.md plus every file in references/, read only if the agent opens them
~24k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); the scripts in this folder are not scanned.

SKILL.md

The full file from GizClaw/flowcraft at commit 88944e3, republished under its MIT licence (© GizClaw). 995 words, ~2,286 tokens.

Download SKILL.mdSave it as .claude/skills/flowcraft-config/SKILL.md (or your agent's skills folder). This skill also uses 26 other files; get the full folder from GitHub.
name
flowcraft-config
description
Author, validate, and troubleshoot complete FlowCraft configuration: deployment documents (deploy.yaml with the runtime section, inference/workspace/sandbox/tool sub-documents, core/memory contracts, and graph JSON node wiring), craft.yaml application definitions, and plugin.json manifests. Use when writing or reviewing FlowCraft configs, assembling an agent deployment on top of a Craft, adding runtime/session settings, building graph definitions, defining plugins and their permissions, resolving config build failures ("not registered", dead configuration, route policy missing, graph node errors), or copying a minimal runnable FlowCraft deployment or craft template.

FlowCraft Config Authoring

Write complete FlowCraft deployment configuration: deploy.yaml, the runtime section, resource sub-documents, and graph JSON. Validate with the L2 structural validator and fix errors against the reference cards.

Workflow

  1. Scope the deployment. Collect the agents, capabilities (chat, tools, memory, scripts, sandboxing), and runtime needs (sessions, resume, event bus, checkpoint stores).

  2. Draft the deployment document. deploy.yaml is the convention but any filename works; pass whatever path you choose to the validator. Read references/deploy.md first. Order the areas: resources → agents → runtime. Decide whole-resource vs item dep refs (infer vs ws/project).

  3. Write sub-documents. Read references/resources.md for the owning module's schema. Workspace before sandbox (custom sandbox runners may depend on workspace resources). Memory implementation modules are app-registered — omit implementation examples; core/memory contracts and hooks are fine to use.

  4. Write graph JSON. Read references/graph.md. Prefer routing: wire the inference.Router into the graph engine and omit model in inference nodes; pin model only when no router is wired. Model refs must use the nested id form; script nodes need runtime and source; wire edges back to the inference node after tool nodes.

  5. Assemble the Craft application. When the deployment runs inside a Craft, wrap it in craft.yaml: an inline deploy (or base_layers), plus ui, plugins (roots, tool_registry, node_targets) and host_tools where they matter. Read references/craft.md, then copy assets/minimal-craft as a starting point. A definition is checked by the host with craft.ParseDefinition; the standalone validator does not cover craft.yaml (see "Compatibility and versioning").

  6. Validate structurally with L2. Run skills/flowcraft-config/scripts/validate-config.sh <deployment-file> (or the installed copy's script). The validator pins the FlowCraft core module in its go.mod and works standalone from any directory. It is structural only: it strictly decodes the document through core/deploy.Parse (unknown fields are rejected, resource/agent entries are shape-checked), strictly decodes and validates the runtime subtree through core/runtime.DecodeConfig when present, and structurally validates graph definitions through core/graph.GraphDefinition.Validate. It does not build resources: no factory registry, no settings file/embed resolution, no node config decoding, no credentials, and no provider calls. Custom and app-registered kinds pass as long as they fit the resource envelope; their settings semantics are the host build's job.

    Sub-documents can also be validated on their own: validate-config.sh --type inference inference.yaml (types: deploy, inference, workspace, sandbox, tool, graph, agent). Standalone types are parsed as JSON/YAML with strict YAML conversion (duplicate keys rejected, single document enforced); graph additionally runs GraphDefinition.Validate (unique node ids, entry presence, edge endpoints). No sub-document schema is decoded in standalone mode — inference, workspace, sandbox, tool, and agent are syntax checks only.

  7. Fix errors. Each failure is prefixed [parse], [graph], or [build] ([parse] input/syntax errors, [graph] graph definition errors, [build] deploy document or runtime subtree errors); validation stops at the first error. Consult references/pitfalls.md for the error → cause map. The validator prints no warnings: custom and app-registered kinds pass structurally, and settings semantics it cannot check are the host build's responsibility.

  8. Hand off. Report the structural validation result and what L2 could not verify: settings schemas, factory resolution, file/embed references, and graph node configs are validated only when the host builds the deployment with its own registry. Note the registration code the host still needs. If the deployment relies on runtime agent registration, call out that it needs the live registry API (Runtime.RegisterAgent / UnregisterAgent; see references/runtime.md) and any dynamic_catalog default/WithToolAssembly requirement.

Show full SKILL.md (447 more words)Show less

Cross-file invariants

These invariants describe core semantics. The validator enforces the structural subset (see step 6); the rest are enforced when the host builds the deployment with its own factory registry.

  • Memory hooks bind a whole memory.Assembly resource; the settings schema is impl-owned. Concrete implementations are app-registered.
  • Sandbox runners are sandbox.Runner resources; first-party impls (local, bwrap, seatbelt) take no deps, custom impls declare their own.
  • runtime.event_bus is required when a runtime section exists, and event_bus/checkpoint_store/dynamic_catalog.tools values must name resources in the document (the host build resolves them by contract).
  • sessions.resume: true requires checkpoint_store.
  • Agent tools is an allow-list, not a catalog declaration; policy is harness state, not engine settings.
  • Graph model refs: model: {id: {provider, name}}.
  • Runtime-registered agents reuse the deployment assembly path; their names must not collide with deployed agents, and with a dynamic_catalog they need a tool mapping or a default.
  • craft.yaml wraps the deployment document: deploy and base_layers are mutually exclusive and one is required, craft.id and craft.version are required, and min_host_version is a compatibility floor checked against the host (craft.Version).
  • A plugin's gated sections are fail-closed: mcp / skills / hooks / nodes declared without mcp:provide / skills:provide / hooks:provide / nodes:provide are dropped, and an unknown permission is rejected rather than ignored.
  • plugins.tool_registry and plugins.node_targets are mounting points resolved against the composed document — the named resource or agent must exist — and a plugin can never register a resource kind.

Templates

Copy assets/minimal-deploy as a starting point: one agent, graph inference node, and a validated runtime section. Replace the model/provider references and workspace root, then extend per the workflow.

Copy assets/minimal-craft when the deployment runs inside a Craft: it wraps an inline deploy with ui, plugins and host_tools, and ships a sample plugin directory.

Compatibility and versioning

One skill version pins one FlowCraft version. The validator's go.mod requires exactly github.com/GizClaw/flowcraft/core v0.4.10; the schema cards in this skill document that release (no driver/* or backends/* modules are required, since the validator never constructs factories). When FlowCraft releases a new version, bump the pin and reconcile the cards in the same change. Custom kinds are structurally valid by design; their factories and settings schemas live in the host application.

The validator pins core only and does not validate craft.yaml: no craft/vX.Y.Z release exists, so a standalone validator cannot resolve the craft module. A craft definition is checked by the host with craft.ParseDefinition, and its factories by the host registry.

Reference index

  • deploy.md — deploy document schema, dep refs, first-party kinds, registration.
  • runtime.md — runtime section, sessions, dynamic registration, reload.
  • resources.md — inference/memory/workspace/ sandbox/tool/event/checkpoint/delegation/script-runtime sub-documents.
  • graph.md — graph JSON, node configs, engine build settings.
  • pitfalls.md — known drift points and the error → cause map.
  • craft.md — craft.yaml and plugin.json schemas, permissions and drop rules, host_tools bindings, hostmcp primitives.

© GizClaw, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 26 other files (scripts, references, assets) in skills/flowcraft-config of GizClaw/flowcraft.

  • SKILL.md
  • agents/openai.yaml
  • assets/minimal-craft/craft.yaml
  • assets/minimal-craft/graphs/assistant.json
  • assets/minimal-craft/inference.yaml
  • assets/minimal-craft/plugins/hello/plugin.json
  • assets/minimal-craft/plugins/hello/skills
  • assets/minimal-craft/workspace.yaml
  • assets/minimal-deploy/deploy.yaml
  • assets/minimal-deploy/graphs/assistant.json
  • assets/minimal-deploy/inference.yaml
  • assets/minimal-deploy/workspace.yaml
  • assets/minimal-deploy/workspace
  • … and 14 more

Open the folder on GitHubat commit 88944e3

Compare with similar skills

FlowCraft Config Authoring next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

FlowCraft Config Authoring compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
FlowCraft Config Authoring this skillGizClaw/flowcraft417—~2.3kAutomated safety check: PassMIT
Microsoft Foundrymicrosoft/GitHub-Copilot-for-Azure2551 repos~6.7kAutomated safety check: PassMIT
Build On Basebase/skills121—~804Automated safety check: PassMIT
Deploy AI Agentbolivian-peru/os-moda119—~1.1kAutomated safety check: PassApache-2.0
Langsmith Deploymentsoba-labs/langchain-agent-skills107—~1.7kAutomated safety check: PassMIT
Google Cloud Agent SDK Masterjeremylongshore/tons-of-skills-marketplace2.8k—~1.4kAutomated safety check: PassMIT

Similar skills

  • Microsoft Foundry

    microsoft/GitHub-Copilot-for-Azure

    Official

    Build, deploy, evaluate, optimize, fine-tune, and manage Microsoft Foundry agents, models, and resources end to end.

    255 GitHub starsUsed in 1 repo~6.7k tokens
    AI & LLM EngineeringAuto-check passed

  • Build On Base

    base/skills

    Complete Base development playbook. An agent skill from base/skills.

    121 GitHub stars~804 tokensUpdated yesterday
    AI & LLM EngineeringAuto-check passed

  • Deploy AI Agent

    bolivian-peru/os-moda

    Deploy and manage AI agent workloads with GPU checks, API key management, and health monitoring

    119 GitHub stars~1.1k tokensUpdated 3 mo ago
    AI & LLM EngineeringAuto-check passed

  • Langsmith Deployment

    soba-labs/langchain-agent-skills

    Deploy and operate production agent servers with LangSmith Deployment.

    107 GitHub stars~1.7k tokensUpdated 1 mo ago
    AI & LLM EngineeringAuto-check passed

  • Google Cloud Agent SDK Master

    jeremylongshore/tons-of-skills-marketplace

    Build, evaluate, deploy, publish, or modernize Google ADK agent applications with Google's maintained agents-cli workflow.

    2.8k GitHub stars~1.4k tokensUpdated today
    AI & LLM EngineeringAuto-check passed

  • Adk Engineer

    jeremylongshore/tons-of-skills-marketplace

    Execute software engineer specializing in creating production-ready ADK agents with best practices, code structure, testing, and deployment automation.

    2.8k GitHub stars~836 tokensUpdated today
    AI & LLM EngineeringAuto-check passed

Works with

Categories

Questions about FlowCraft Config Authoring

What does FlowCraft Config Authoring do?

Writes and validates a complete FlowCraft agent deployment: deploy.yaml, runtime settings, resource sub-documents and graph JSON, checked against a structural validator. The agent first scopes the deployment by listing the agents, capabilities and runtime needs (sessions, resume, event bus, checkpoint stores) required, then drafts the deployment document in the order resources, agents, runtime, deciding between whole-resource and item-level dependency references. Sub-documents for workspace, sandbox and other resources follow the owning module's schema, with workspace written before sandbox since custom sandbox runners may depend on it, and memory implementation details left to the app's own registered modules.

When should I use FlowCraft Config Authoring?

FlowCraft Config Authoring fits situations like: writing a new FlowCraft deployment configuration; fixing a FlowCraft config build failure such as a missing route policy; wiring graph JSON nodes for an agent's capabilities.

How do I install FlowCraft Config Authoring in Claude Code?

Run `npx skills add GizClaw/flowcraft --skill flowcraft-config -a claude-code`. Or copy the skill folder (skills/flowcraft-config in GizClaw/flowcraft) into .claude/skills/flowcraft-config in your project. Claude Code loads it when a task matches its description.

How do I install FlowCraft Config Authoring in Codex?

Run `npx skills add GizClaw/flowcraft --skill flowcraft-config -a codex`. Or copy the skill folder (skills/flowcraft-config in GizClaw/flowcraft) into .agents/skills/flowcraft-config in your project. Codex loads it when a task matches its description.

Can I use FlowCraft Config Authoring in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add GizClaw/flowcraft --skill flowcraft-config -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/flowcraft-config, .gemini/skills/flowcraft-config, .github/skills/flowcraft-config and .opencode/skills/flowcraft-config in your project.

What does FlowCraft Config Authoring need to run?

SKILL.md names no scripts, command-line tools or credentials: FlowCraft Config Authoring is instructions for the agent only. Our summary lists: Go, since the config validator pins the FlowCraft core module in its go.mod.

Does FlowCraft Config Authoring access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is FlowCraft Config Authoring safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. The check reads SKILL.md only: the scripts in the folder are not scanned, so read them before running anything.

What licence does FlowCraft Config Authoring use?

FlowCraft Config Authoring is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does FlowCraft Config Authoring use?

About 2.3k tokens (SKILL.md is roughly 9.1k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full. Its references folder adds about 21k tokens, read only when the agent opens those files.

What are the alternatives to FlowCraft Config Authoring?

Skills that share tags, products or a category with FlowCraft Config Authoring: Microsoft Foundry (microsoft/GitHub-Copilot-for-Azure, 255 stars), Build On Base (base/skills, 121 stars), Deploy AI Agent (bolivian-peru/os-moda, 119 stars) and Langsmith Deployment (soba-labs/langchain-agent-skills, 107 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains FlowCraft Config Authoring?

GizClaw (a GitHub organization) maintains it in GizClaw/flowcraft, which has 417 GitHub stars. The repository was last updated on October 10, 2026.

Source: GizClaw/flowcraft on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.