Claude Code Adds /skill-doctor and Bigger Inline Output Limits
Claude CodeClaude Code 2.1.261 added /skill-doctor, a new command that reports which loaded skills go unused and how much context budget each one costs, so users can prune their setup. The same release added bashOutputMaxChars and taskOutputMaxChars settings that raise how much command and background-task output Claude receives inline, up to 128K characters, before it gets saved to a file instead. Together the two additions target two everyday friction points: bloated, unused skills silently eating context, and long build or test output getting truncated mid-command.
Key Takeaways
/skill-doctorreports which loaded skills go unused, giving users a concrete signal for what to prune from a growing skills setup.- The command also quantifies each skill's context cost, turning an abstract "this feels bloated" feeling into a specific token number.
bashOutputMaxCharsandtaskOutputMaxCharsraise the inline output ceiling to up to 128K characters, up from Claude Code's previous default limits.- The change specifically helps verbose build and test tooling, where important output previously risked landing past the truncation point.
- Both settings are opt-in and session-configurable, not new defaults imposed on every user.
- The pairing targets two forms of accumulated friction: skill bloat over a project's lifetime, and output loss during any single long-running command.
/skill-doctor Surfaces Skills That Aren't Pulling Their Weight
As users accumulate more skills in a Claude Code project, some inevitably go unused while still contributing to the fixed overhead every session pays to load them. Version 2.1.261 introduces /skill-doctor, a command that reports which loaded skills are going unused and quantifies what each one costs in context tokens. Rather than guessing which of a growing skills folder is worth keeping, a user can run the command and get a concrete basis for pruning: skills that never get invoked but still consume budget on every turn are exactly the ones this feature is built to surface.
Longer Command Output Stays Inline Instead of Getting Cut Off
The second addition addresses a more common annoyance: verbose commands, such as build tools or test runners with chatty logs, getting truncated before Claude can see the useful part. The new bashOutputMaxChars and taskOutputMaxChars settings let a user raise the ceiling on how much command output and background-task output Claude Code will keep inline, up to 128K characters, before falling back to saving the rest to a file. Users working with unusually verbose tooling can now tune this limit up front instead of hitting the default ceiling mid-task and having to re-run a command or dig through a saved file to find the relevant lines.
Why It Matters
Neither change reshapes how Claude Code works day to day, but both quietly reduce friction that accumulates the longer a project runs: unpruned skills quietly taxing every conversation, and long-running commands whose most important output lands past the truncation point. Both are opt-in, session-level improvements rather than defaults that change existing workflows.