From teamscale
Run pre-commit analysis on local uncommitted changes via Teamscale, then fix the resulting findings.
How this skill is triggered — by the user, by Claude, or both
Slash command
/teamscale:local-fix-findings [<path> ...][<path> ...]The summary Claude sees in its skill listing — used to decide when to auto-load this skill
Send the user's uncommitted working-tree edits to Teamscale for pre-commit
Send the user's uncommitted working-tree edits to Teamscale for pre-commit analysis, then fix the findings introduced by those edits.
Determine which paths to analyse. Default to the repository root. The user may also pass specific paths if they want to narrow the scope.
Invoke pre-commit analysis via the existing Teamscale MCP tool
mcp__plugin_teamscale_teamscale__teamscale-dev_pre-commit. If a tool
with that exact name is not available, use the pre-commit tool exposed
by the teamscale MCP server under whatever name your environment surfaces it.
Pass these arguments:
paths: the paths chosen in step 1uploadScope: ONLY_UNCOMMITTED — we only want findings on changes
that are not yet committed; pre-existing issues are out of scopeonlyPreCommitFindings: true — only report findings actually
introduced by the local edits, not findings already present on the
server sideseverity and categories: leave unset (no extra filtering) unless
explicitly requested by the user.If the MCP tool returns an error, stop the skill and surface the error verbatim to the user. Do not guess findings, fall back to other tools, or retry — the user needs to see the real error (typically a setup, network, or configuration problem).
Triage and fix the returned findings.
To understand finding priorities and fixing strategies, read ../../shared/finding-priorities.md
To get additional information about findings (finding type descriptions) read ../../shared/finding-type-descriptions.md
Do not attempt to tolerate or flag findings.
Summarise. Print two buckets — fixed and skipped — as described in the priorities snippet. The "tolerated" bucket from the snippet does not apply here: this skill does not flag findings, so omit it.
Suggest re-running after the fixes so the user can confirm the introduced findings are gone before committing.
npx claudepluginhub teamscale/claude-code --plugin teamscaleProvides a checklist for code reviews covering functionality, security, performance, maintainability, tests, and quality. Use for pull requests, audits, team standards, and developer training.