Priority: Medium
Context
When providing the dry run flag, validation steps should still happen, we shouldn't just be parroting what the user has inputted.
As an example, the tw thread delete command does validation checks prior to deleting the thread, these checks should still happen even if it's dry run, that's the whole point of a dry run.
Practical example. npm pack --dry-run will build a package, and run as if it's packing the package up, the only thing it won't actually do is create the package, because it's a dry run.
Additional: Improve dry-run output for agent consumption
Based on 7 Principles for Agent-Friendly CLIs, Principle 3 (Fail Fast with Actionable Errors):
Failed commands should teach agents how to succeed on retry.
Problem
Current dry-run messages are vague and don't tell an agent what to do next:
Dry run: would delete thread 123
Dry run: would create thread in channel 456
An agent seeing this doesn't know the exact command to run to execute for real.
Proposed improvement
Dry-run output should include the exact command to execute:
Dry run: would delete thread 123
To execute: tw thread delete 123 --yes
Dry run: would create thread in channel 456
Title: My thread title
To execute: remove --dry-run flag to create
This gives agents (and humans) a clear next step. The "to execute" line acts as a confirmation receipt and a copy-pasteable command.
Scope
All commands with --dry-run support should be updated:
thread create, thread reply, thread done, thread delete, thread mute, thread unmuteconversation reply, conversation done, conversation mute, conversation unmutemsg update, msg deletecomment update, comment deleteaway set, away clearreact, unreact
Context
When providing the dry run flag, validation steps should still happen, we shouldn't just be parroting what the user has inputted.
As an example, the
tw thread deletecommand does validation checks prior to deleting the thread, these checks should still happen even if it's dry run, that's the whole point of a dry run.Practical example.
npm pack --dry-runwill build a package, and run as if it's packing the package up, the only thing it won't actually do is create the package, because it's a dry run.Additional: Improve dry-run output for agent consumption
Based on 7 Principles for Agent-Friendly CLIs, Principle 3 (Fail Fast with Actionable Errors):
Problem
Current dry-run messages are vague and don't tell an agent what to do next:
An agent seeing this doesn't know the exact command to run to execute for real.
Proposed improvement
Dry-run output should include the exact command to execute:
This gives agents (and humans) a clear next step. The "to execute" line acts as a confirmation receipt and a copy-pasteable command.
Scope
All commands with
--dry-runsupport should be updated:thread create,thread reply,thread done,thread delete,thread mute,thread unmuteconversation reply,conversation done,conversation mute,conversation unmutemsg update,msg deletecomment update,comment deleteaway set,away clearreact,unreact