chore(claude-code): add PostToolUse hook to enforce lint and prettier after file edits

Description

Add a Claude Code PostToolUse hook in .claude/settings.json (core-web) that automatically runs Prettier formatting and Nx lint checks every time Claude edits or writes a file. This ensures lint and style errors are caught and reported before Claude gives its final response — preventing back-and-forth correction cycles and keeping the codebase consistently formatted.

A companion Stop hook runs nx affected -t lint --fix on all modified files before Claude finishes a session, acting as a final safety net.

Why

When Claude edits multiple files in a session, lint or formatting violations can accumulate silently. Developers then discover errors only after reviewing Claude's output, requiring extra round-trips. Automating the check at the tool-use level shifts detection left and keeps code quality gates inside the AI-assisted workflow.

Current Implementation (core-web)

The following hooks are already configured in core-web/.claude/settings.json:

{
  "hooks": {
    "PostToolUse": [
      {
        "matcher": "Edit|Write",
        "hooks": [
          {
            "type": "command",
            "command": "FILE=$(jq -r '.tool_input.file_path // empty'); [ -z \"$FILE\" ] || { yarn nx format:write --files=\"$FILE\" 2>/dev/null; LINT_OUT=$(yarn nx affected -t lint --files=\"$FILE\" 2>&1); LINT_EXIT=$?; if [ $LINT_EXIT -ne 0 ]; then jq -n --arg msg \"Lint errors in $FILE:\\n$LINT_OUT\" '{\"hookSpecificOutput\": {\"hookEventName\": \"PostToolUse\", \"additionalContext\": $msg}}'; fi; }"
          }
        ]
      }
    ],
    "Stop": [
      {
        "hooks": [
          {
            "type": "command",
            "command": "MODIFIED=$(git diff --name-only HEAD 2>/dev/null | grep -E '\\.(ts|html|scss|js)$'); [ -z \"$MODIFIED\" ] && exit 0; LINT_OUT=$(yarn nx affected -t lint --fix 2>&1); LINT_EXIT=$?; [ $LINT_EXIT -eq 0 ] && exit 0; jq -n --arg msg \"Lint errors that could not be auto-fixed. Please address before finishing:\\n$LINT_OUT\" '{\"continue\": false, \"stopReason\": $msg}'"
          }
        ]
      }
    ]
  }
}

PostToolUse (Edit | Write matcher):

• Extracts the edited file path from the tool input
• Runs yarn nx format:write --files=$FILE (Prettier)
• Runs yarn nx affected -t lint --files=$FILE
• If lint fails, injects the error output as additionalContext back into Claude's context so it can self-correct

Stop hook:

• Collects all .ts|.html|.scss|.js files modified in the current session
• Runs yarn nx affected -t lint --fix across them
• If unfixable errors remain, blocks Claude from stopping and surfaces the details

Acceptance Criteria

• PostToolUse hook is present in core-web/.claude/settings.json with matcher: "Edit|Write"
• Hook runs yarn nx format:write --files=$FILE (Prettier) on every edited file
• Hook runs yarn nx affected -t lint --files=$FILE on every edited file
• Lint failures are fed back as additionalContext so Claude self-corrects before responding
• Stop hook runs nx affected -t lint --fix on all session-modified files before Claude finishes
• Stop hook blocks session completion when unfixable lint errors remain and surfaces the error output
• Hook configuration is documented in core-web/CLAUDE.md under a Hooks section

Additional Context

• Tooling: Nx monorepo, Yarn, Angular 19+
• Affected files: .ts, .html, .scss, .js
• Claude Code hooks docs: https://docs.anthropic.com/en/docs/claude-code/hooks
• PostToolUse hookSpecificOutput.additionalContext injects text back into Claude's next prompt turn, enabling self-correction without user intervention

Start with core-web/.claude/settings.json and core-web/CLAUDE.md, then consult the linked Claude Code hooks documentation and inspect the existing Nx commands. Verify the PostToolUse and Stop hooks against the listed acceptance criteria, including formatting, lint feedback, session-wide lint fixing, and documentation; the work is done when those behaviors are reproducible for the affected file types.