Callee CLI
A CLI and plugin: agent workflows live in the repository as Markdown and YAML with roles, scripts, pauses and branching.
- 78 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- `Script` nodes run shell commands and roles call coding agents with project access
Install
- Works in
Manual install
npx --yes @baldaworks/callee@latest setup codexFor another agent replace `codex` with `claude`, `cursor`, `opencode`, `copilot` or `grok`.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- `Script` nodes run shell commands and roles call coding agents with project access
This is third-party code. Review the repository files before installing.
What it does
Callee describes repeatable agent work as versioned Markdown and YAML resources. The graph includes model roles (`Role`, `DynamicRole`), local `Script` steps, a `Human` pause and typed evaluations through TypeSafe or OpenRouter. Nodes combine through Sequential, Parallel, Loop and Router patterns. The whole graph is validated before running, one explicit state object passes through every step and one final artifact comes out. The command `npx --yes @baldaworks/callee@latest setup codex` installs six starter resources and the integration, and `claude`, `grok`, `copilot`, `opencode` or `cursor` work instead of `codex`.
Who it is for. Teams that run the same multi-step agent process, such as multi-role review, and want to keep it next to the code.
Good fit when
- Several reviewers work in parallel while an architect compiles the result.
- The process is reviewed like code: prompts, models and permissions live in the repository.
- Midway through a chain the agent stops and waits for a human answer.
- The same scenario runs from a terminal and from Claude Code.
Not a fit when
- A one-off task: an ordinary prompt is enough.
- There is no coding agent or provider keys: roles will not run without them.
- A ready interface without YAML is expected: resources are written by hand.
Example request
Run workflows/investigate and explain the project architecture and main entry points.Limitations
The resource API version is still `v1alpha1`, so formats may change. Roles need separately installed provider executables and their credentials. Scripts and human pauses need no model, and typed evaluations call their own HTTP APIs. Scripts run commands on your machine, so resources from other repositories are read before running. Parallel nodes run unattended and only `Human` pauses for a person. The license is MIT.
How to disable. Remove the added resources and the agent integration, then the `@baldaworks/callee` package if needed. Delete artifacts from past runs by hand.
FAQ
Which node kinds exist?
Ten: Role, DynamicRole, Script, Human, TypeSafeJev, OpenRouterDecision, Sequential, Parallel, Loop and Router.
How do I install it for Claude Code?
With `npx --yes @baldaworks/callee@latest setup claude`. For Codex, Cursor and others change the agent name at the end of the command.
What is validated before running?
The whole graph: resource references, node kinds and connections. A schema error stops the run before any model call.
Does every step need a model?
No. `Script` and `Human` work without a model.
Where are examples?
In the repository docs: the Quickstart and Browse runnable examples sections.
Discussion
No questions or reviews yet.
Related
CLI toolsAn orchestrator for agents: the workflow is defined in code, every step is enforced, key points need human approval and decisions go into an immutable journal
CLI for parallel agents: splits a task into sections, runs Codex, Claude Code or OpenCode together and merges into one branch
The squads CLI runs teams of agents defined in markdown files in git: a lead briefs, workers act, a verifier checks.
One runtime for Claude Code, Codex, OpenCode and more: a run counts as done only when checks the agent did not write pass
The ee CLI gives agents durable local memory: rules, decisions and anti-patterns with scores and provenance
A plugin that syncs Claude Code, Codex CLI and Antigravity on one project and lets them hand work to each other.
README in short
The README explains workflows as a resource graph, the node kinds, a YAML example, one-command setup and documentation links.