zymi-core
VerifiedAuditable YAML-based backend for secure agent tool pipelines.
What is zymi-core?
Zymi-core serves as the backend layer for agent tools rather than providing planners or chat interfaces. It converts YAML pipelines into executable DAGs that support connectors, approvals, and Python-based steps while validating everything upfront.
Every run produces a hash-chained event log stored in a local database. Agents or operators can query past executions, inspect exact inputs and outputs per step, and resume interrupted flows without additional logging setup.
The project targets teams that need reliable tool execution for agents in production or regulated environments. It works with MCP-compatible hosts such as Claude Desktop or frameworks like LangGraph while also supporting direct Telegram bot deployments.
What you can build with zymi-core
MCP tool exposure
Expose YAML pipelines as callable tools to any MCP host so agents gain governed access to shell, HTTP, and file operations with built-in approval prompts.
Standalone Telegram agent
Initialize a project, configure a bot token, and run a two-step assistant-reviewer pipeline that handles messages while logging every decision.
Run auditing and debugging
Enable observability mode to let agents or developers list historical runs, retrieve event traces, and examine step-level I/O without parsing log files.
Install zymi-core
uv tool install zymi-coreuv tool install zymi-core # one-time; puts `zymi` on PATH globally
mkdir telegram-agent && cd telegram-agent
zymi init --example telegram
# 1. Create a bot via @BotFather in Telegram; copy the token.
# 2. Fill .env:
cp .env.example .env # edit TELEGRAM_BOT_TOKEN + OPENAI_API_KEY
# 3. Open project.yml, replace "your_username_here" with your actual
# Telegram username (no @). Keeps strangers out of the bot.
zymi fetch # uv sync — builds ./.venv from pyproject.toml
zymi serve chat # .env is auto-loaded; pipeline runs in ./.venv- 1Install the CLI globally with uv tool install zymi-core.
- 2Create a new directory and run zymi init with an example template such as telegram.
- 3Copy the environment example, add required tokens, and update any username restrictions in the project file.
- 4Execute zymi fetch to synchronize dependencies into a local virtual environment.
- 5Start the service with zymi serve to activate the pipeline and begin handling requests.
zymi-core: pros & cons
Pros
- +Full event sourcing enables replayable and auditable runs without extra setup.
- +Declarative YAML approach makes pipelines easy to version and review like infrastructure code.
- +Interactive approval gates render directly in supported agent UIs for risky actions.
- +Built-in observability lets agents introspect their own execution history.
Cons
- –Requires familiarity with YAML and DAG concepts for effective use.
- –Depends on external tools like uv for dependency management.
- –Focuses only on the tool layer so users must pair it with separate agent frameworks.
Frequently asked questions
No, it supports standalone operation such as running a Telegram bot directly from the YAML configuration.
User reviews
Verified reviews from the community shape this listing's rating.
Loading reviews…
Similar agents
Other general-purpose options worth comparing.
