rho-cli 0.1.25

Rho CLI tools for encrypted agent collaboration, dataset publishing, controlled runs, and result release workflows
Documentation
# V1 Scope

## Build First

Rho v1 should build the smallest architecture that proves:

- identity works
- repo policy works
- messages work
- tools are declarative and runnable
- agents can propose but not bypass policy
- middleware can require human approval
- mock data can be generated from private data shapes

## In Scope

### `rho id`

- GitHub-linked identity
- local public/private key generation
- trust/export/import basics

### `rho crypto`

- sign and verify files
- encrypt and decrypt files for known recipients
- manage crypto algorithm metadata
- provide a clean crypto boundary for other commands

### `rho auth`

- evaluate whether an actor may perform an action
- manage repo-level authorization metadata
- support trust-on-first-use pairing
- support first-connection approval hooks
- support permanent block and deny records

### `rho repo`

- initialize repo layout
- manage participants
- manage permissions policy
- manage repo root manifest metadata
- create required visible `rho/` structure

### `rho tools`

- define tool manifests
- validate schemas
- list available tools for a repo

### `rho messages`

- write and read text-backed structured messages inside the repo
- route inbox and outbox files
- basic human and agent addressing

### `rho run`

- run tools through sandbox profiles
- validate requests
- write run status and outputs

### `rho agent`

- launch an agent with repo context
- expose repo tools and message protocol
- require `rho run` for protected execution

### `rho data`

- inspect source data
- infer schemas and common shapes
- generate mock/synthetic outputs
- register mock/private/released paths

### `rho doctor`

- check required dependencies
- guide installation steps
- validate identity and key setup
- validate trust and auth setup
- validate repo layout and policy files
- check agent backend readiness
- check sandbox readiness

### `rho middleware`

- intercept approval-worthy events
- execute policy hooks
- record decisions

### `rho telegram`

- adapter that lets middleware notify or query a human operator

## Explicitly Out Of Scope For V1

- secure multi-party computation
- daemon-based sync substrate
- notebook-specialized collaboration runtime
- broad flow engine with many backends
- realtime transport beyond normal Git/GitHub usage
- desktop-first orchestration
- rich auto-materialized encrypted shadow stores
- many identity backends beyond GitHub + keypairs
- key rotation and recovery workflows
- custom envelope evolution and migration
- automatic background dependency management services
- complex federated policy distribution

## What To Borrow From The Old System

- approval-mediated execution from Beaver / Rho relay ideas
- explicit mock/private split from Twin concepts
- file-backed request and response protocols
- a separate crypto boundary rather than mixing encryption into every command
- explicit trust and authorization records instead of implicit allow behavior
- strict separation between product semantics and transport adapters
- explicit operator diagnostics instead of hidden setup magic

## What To Leave Behind For Now

- multi-repo workspace orchestration
- many transport modes
- heavy path and config indirection
- publisher ACL plus subscriber materialization model
- notebook runtime magic
- Nextflow-style workflow backends
- bioscript-style domain packaging and generation
- secure-compute packaging burden
- broad distributed trust frameworks beyond TOFU plus explicit blocks