creatio-agent-kit
Creatio development for AI agents: recipes, package toolkits, a describe layer, a UI/backend test framework, MCP servers and skills for Claude Code, Codex CLI and GitHub Copilot CLI.
Safety scan
A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.
Badge for your README
Markdown
[](/listings/artemivzhenko--creatio-agent-kit)HTML
<a href="/listings/artemivzhenko--creatio-agent-kit"><img src="/api/listings/artemivzhenko--creatio-agent-kit/badge.svg" alt="findsafeskills safety rating" /></a>Install
Install instructions from the README
As written by the repo's author. findsafeskills has not verified these steps.
Requires Python 3.10+, .NET 8 SDK, at least one agent CLI (Claude Code, Codex CLI or GitHub Copilot CLI), and clio for the package round trip.
git clone <this repository> cd creatio-agent-kit py -3 agent_setup.py # python3 on macOS and Linux; --dry-run prints the commands, --check plans only py -3 agent_setup.py --agents codex # one agent instead of every one found: claude, codex, copilot
The script creates the virtual environments of both MCP servers once, then, for every agent CLI it finds, registers `creatio-kb`, `creatio-tests`, `context7`, `microsoft-learn` and (with the .NET 10 SDK) `roslyn-codelens` at user scope with the interpreter of each venv, so the paths follow the clone, and copies the four Creatio skills into the agent's personal skills folder (`~/.claude/skills` for Claude Code, `~/.agents/skills` for Codex and Copilot). Claude Code also gets the ponytail plugin and graphify. Re-running it is safe. Then:
cd automatic-creatio-tests && dotnet build && dotnet test # the framework, no browser needed powershell -File src/CreatioAutoTests.Cli/bin/Debug/net8.0/playwright.ps1 install chromium # once
What was scanned
The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.
README.md · README · 72 lines
- # Creatio agent kit
- Everything an AI agent needs to build and prove Creatio customisations as package files: a knowledge base of
- recipes and verified platform facts, Python toolkits that write business processes, rules, objects, pages, cases,
- data bindings and SQL scripts byte for byte, a describe layer that answers questions about a package without
- reading its files, a UI and backend test framework with its own DSL, two MCP servers that give all of it to an
- agent as tools, and four agent skills (Claude Code, Codex, Copilot) that turn the tools into roles.
- ## Layout
- | Folder | What it holds | Start at |
- |---|---|---|
- | [creatio-knowledge-base/](creatio-knowledge-base/README.md) | 177 code recipes (Classic UI, Freedom UI, C#, SQL), the clio reference, `no-code/tools/` (17 package toolkits, the describe layer, the environment layer), the `creatio-kb` MCP server (44 tools), the skills `creatio-kb`, `creatio-author`, `creatio-tester`, `creatio-loop`, the platform facts register | its README |
- | [automatic-creatio-tests/](automatic-creatio-tests/README.md) | the test framework: `.act` cases in a small DSL over Playwright for Freedom UI and Classic UI pages, process cases and OData backend cases without a browser, the `creatio-tests` MCP server (11 tools) | its README, [docs/dsl.md](automatic-creatio-tests/docs/dsl.md) |
- | [ci-cd/](ci-cd/README.md) | the pipeline of a package repository: `pipeline.py` with the stages check, build, register, deploy (IsChanged cleared, push-pkg, compile, restart, Redis), unit tests inside Creatio, UI cases; templates for GitHub Actions, GitLab CI and Jenkins; `clio-ext.py`, the developer's commands around clio (push-all in dependency order, hierarchy, localization and process checks, OAuth service); the check, hierarchy, lint, build, deploy and push-all stages are also `ci_*` tools of the knowledge-base MCP server | its README |
- | [postman/](postman/) | a Postman collection for manual API checks against a site; a helper, not part of the pipeline | - |
- | [AGENTS.md](AGENTS.md) | the rules for any AI that changes this repository | - |
- ## Quick start
- Requires Python 3.10+, .NET 8 SDK, at least one agent CLI (Claude Code, Codex CLI or GitHub Copilot CLI), and
- clio for the package round trip.
- ```bash
- git clone <this repository>
- cd creatio-agent-kit
- py -3 agent_setup.py # python3 on macOS and Linux; --dry-run prints the commands, --check plans only
- py -3 agent_setup.py --agents codex # one agent instead of every one found: claude, codex, copilot
- ```
- The script creates the virtual environments of both MCP servers once, then, for every agent CLI it finds, registers
- `creatio-kb`, `creatio-tests`, `context7`, `microsoft-learn` and (with the .NET 10 SDK) `roslyn-codelens` at user
- scope with the interpreter of each venv, so the paths follow the clone, and copies the four Creatio skills into the
- agent's personal skills folder (`~/.claude/skills` for Claude Code, `~/.agents/skills` for Codex and Copilot). Claude
- Code also gets the ponytail plugin and graphify. Re-running it is safe. Then:
- ```bash
- cd automatic-creatio-tests && dotnet build && dotnet test # the framework, no browser needed
- powershell -File src/CreatioAutoTests.Cli/bin/Debug/net8.0/playwright.ps1 install chromium # once
- ```
- ## How a task runs
- On a project the tools run as two roles under one task-folder contract: the **author** builds the task from
- `project.json` (the environment, the package, the paths) with the toolkits, pushes with `clio push-pkg`, probes
- and runs the unit tests; the **tester** proves every acceptance criterion with the tier it needs (unit, probe,
- OData backend case, process case, `.act` UI case), keeps the evidence and writes the report. The contract, the
- files of a task folder and the proof tiers are the skill
- [creatio-loop](creatio-knowledge-base/skill/creatio-loop/SKILL.md); the roles are
- [creatio-author](creatio-knowledge-base/skill/creatio-author/SKILL.md) and
- [creatio-tester](creatio-knowledge-base/skill/creatio-tester/SKILL.md); the recipes come through
- [creatio-kb](creatio-knowledge-base/skill/creatio-kb/SKILL.md).
- ## Tests
- | What | Command |
- |---|---|
- | a package toolkit | `python creatio-knowledge-base/no-code/tools/<toolkit>/tests/test_<toolkit>_tools.py` (plain functions, no pytest; `describe/tests/test_describe.py`) |
- | the knowledge base index and the MCP tools | `creatio-knowledge-base/mcp-server/.venv/Scripts/python.exe creatio-knowledge-base/mcp-server/test_kb_index.py`, `... test_server_tools.py` |
- | the tool docs in sync | `python creatio-knowledge-base/no-code/tools/tests/test_sync_tool_docs.py` |
- | every CLI writes UTF-8; the record history and timer tools | `python creatio-knowledge-base/no-code/tools/tests/test_utf8_cli.py`, `python creatio-knowledge-base/no-code/tools/env/tests/test_env_history.py` |
- | the test framework | `dotnet test` in `automatic-creatio-tests/` |
- | the framework's MCP server | `automatic-creatio-tests/mcp-server/.venv/Scripts/python.exe automatic-creatio-tests/mcp-server/test_server.py` |
- ## Licence
- MIT. See [LICENSE](LICENSE).
- ## What is not in the repository
- Kept outside the repository: `project.json` files with credentials, the MCP servers' virtual environments, `.creatio-index/` stores written next to packages,
- and the framework's `Reports/` and `Configuration/creatio.env.config.json`.
Ratings and comments
Sign in to rate this skill.
Sign inFree to use. If it’s helped, pass it on.