[FEATURE] Pipe / chaining for slash commands

Preflight Checklist

• [x] I have searched existing requests and this feature hasn't been requested yet
• [x] This is a single feature request (not multiple features)

Problem Statement

Problem
Slash commands inside an interactive Claude Code session can only be run sequentially. There is no native way to compose them — e.g. taking the output/result of one command as input/context for the next.

The output (or summary) of the left-hand command would be passed as context to the right-hand command, similar to how claude -p already accepts piped stdin in non-interactive mode.

Why

• Brings well-known Unix composition semantics into the REPL.
• Reduces the need to write a custom .claude/commands/*.md file for every multi-step workflow.
• Lets users combine built-in commands, custom commands, and MCP-provided commands fluently.

Thanks!

Proposed Solution

Proposed solution
Support Unix-style piping for slash commands inside the REPL:

/review src/auth.cs | /commit
/test | /fix
/lint --fix | /commit

Alternative Solutions

Alternatives considered

• Custom slash commands that hardcode the steps (works, but doesn't compose).
• Subagents (heavier, separate context window).
• External bash scripting with claude -p (loses the interactive session state).

Priority

Critical - Blocking my work

Feature Category

CLI commands and flags

Use Case Example

_No response_

Additional Context

_No response_