Skip to content

Releases: nedcodes-ok/cursor-doctor

v1.7.7 — Smarter Lint Rules

28 Feb 19:58

Choose a tag to compare

Lint accuracy improvements (tested against real-world repos):

Reduced false positives:

  • Vague rule detection is now context-aware: qualified phrases like 'be consistent with WordPress conventions' are no longer flagged
  • Fixed false positives on empty description values and description-in-body check

New checks:

  • Dead rule detection: Rules with alwaysApply: false and empty/missing globs now flagged as ERROR ('Rule will never load')
  • Empty frontmatter values: Catches description: and globs: with no value (common copy-paste mistake)

All 80 tests passing.

v1.7.6 — Windows + UX

28 Feb 19:32

Choose a tag to compare

Windows compatibility:

  • Normalize CRLF line endings across all source files (was breaking line counting on Windows)

Edge cases:

  • Detect and skip binary files in .cursor/rules/ with clear warning
  • Exit code 2 for system/internal errors (1 = lint issues, 0 = clean)

UX:

  • Added 6 missing commands to --help: doctor, order, diff, init, generate, verify
  • Invalid --days values now show a warning instead of silently defaulting

v1.7.5

28 Feb 19:23

Choose a tag to compare

Fix plugin.js walkFiles: add node_modules exclusion and depth limit (5) to prevent unbounded directory scanning

v1.7.4 — Security Hardening

28 Feb 19:15

Choose a tag to compare

Security release — 7 vulnerabilities fixed

Critical:

  • Sanitize numeric CLI parameters to prevent command injection via shell interpolation
  • Block SSRF: validate URLs and reject private IPs/localhost in team baseline fetches

High:

  • Path traversal protection on all rule file write operations (generate + team import)
  • ReDoS mitigation: skip oversized lines in conflict directive extraction

Medium:

  • Skip symlinks during file traversal to prevent arbitrary file reads
  • Detect dangerous command patterns in MCP server configurations

v1.7.3

28 Feb 19:07

Choose a tag to compare

Bug fix release — 5 bugs fixed from audit pass 4

Critical:

  • Rewrite LSP server to properly read multi-line JSON-RPC messages using Content-Length byte counting (was reading only first line)

Medium:

  • Fix .cursorrules sort priority — now correctly sorts last within always tier (was incorrectly first)
  • Fix prototype pollution risk in conflict deduplication (use Object.create(null))
  • Fix prototype pollution risk in heading duplicate detection (2 locations)

Low:

  • Clean up regex property access spacing

v1.7.2

28 Feb 18:59

Choose a tag to compare

Bug fix release — 10 bugs fixed from deep audit pass 3

Fixes:

  • Fix execSync deadlock on large repos (add 10MB maxBuffer) in performance tracking
  • Fix division by zero in overlap ratio calculation
  • Add try-catch guards around lint modules to prevent cascading failures
  • Add 15s request timeout to URL fetches in team-sync and generate
  • Fix false positive conflict detection on short subjects
  • Fix version parser to handle prerelease tags
  • Improve token estimation accuracy with 10% metadata overhead factor

v1.7.0 — Critical Bug Fixes

28 Feb 18:16

Choose a tag to compare

14 bugs fixed from deep audit. See commit for full details.

Critical: YAML array globs now parsed correctly, ReDoS vulnerability fixed, alwaysApply logic fixed.
High: --json output works for check/lint/stats.
Medium: mcp.json allowlisted, agent files no longer need frontmatter, paywall exits correctly.
Low: TTY color detection, dead code cleanup, updated paywall message.

v1.6.0 — CLAUDE.md/AGENTS.md + MCP Validation

28 Feb 17:44

Choose a tag to compare

Phase 4: Selective Platform Expansion

New Free Commands

cursor-doctor agents — Validate CLAUDE.md, AGENTS.md, and .cursor/agents/

  • Structure validation, anti-pattern detection, missing sections
  • Duplicate heading detection, file size warnings

cursor-doctor mcp — Validate MCP config files

  • JSON syntax validation, required field checks, type consistency
  • Hardcoded secret detection (catches API keys in config files)
  • Placeholder value warnings, hook conflict checks

Health Scan Integration

Both checks integrated into cursor-doctor scan with dynamic scoring.

npx cursor-doctor@1.6.0

v1.5.0: Rule Performance Tracking, Rule Testing, Team Sync

28 Feb 17:27

Choose a tag to compare

Phase 3: New Capabilities

Rule Performance Tracking (npx cursor-doctor perf) — PRO

Answers the question every Cursor user has: are my rules actually doing anything?

  • Analyzes git history to determine which rules fire
  • Shows active, always-on, low activity, and dead rules
  • Calculates wasted tokens on dead rules
  • Suggests removing rules with zero file matches
  • Supports VS Code extension activity data for richer metrics

Rule Testing (npx cursor-doctor test) — PRO

Give it a rule + code. It checks whether an AI model follows the rule.

  • A/B comparison: run with rule vs without, shows the diff
  • Adherence scoring (0-100) with violation and improvement lists
  • Supports Gemini (free tier), OpenAI, Anthropic
  • Single rule or batch test all rules against a code file
# Test one rule
GEMINI_API_KEY=... npx cursor-doctor test error-handling.mdc src/api.ts

# Test all rules
GEMINI_API_KEY=... npx cursor-doctor test src/api.ts

Team Sync (npx cursor-doctor team) — PRO

Share rules across teams. Detect when someone's gone rogue.

  • team export — Portable JSON config of all rules
  • team import <file-or-url> — Import with --overwrite, --dry-run
  • team baseline <source> — Set team standard (file or URL)
  • team drift — Shows added (personal overrides), modified, deleted rules vs baseline

Install

npx cursor-doctor perf
npx cursor-doctor test <code-file>
npx cursor-doctor team export --out=team-rules.json
npx cursor-doctor team drift

v1.4.0: Smart Token Budget + Cross-Format Conflict Detection

28 Feb 17:12

Choose a tag to compare

Phase 2: Deepen the Moat

Smart Token Budget (npx cursor-doctor budget)

FREE:

  • Context window usage visualization (% of 120K context window)
  • Always-loaded vs conditional token split
  • Top 5 rules ranked by token cost

PRO (--pro):

  • Per-file-type breakdown
  • Per-rule cost ranking (full list)
  • Waste detection with suggested globs and savings estimates
  • Historical tracking with trend analysis

Cross-Format Conflict Detection (npx cursor-doctor conflicts) — PRO

Detects contradictory instructions across .cursor/rules/, CLAUDE.md, AGENTS.md, .cursorrules, .cursor/agents/, and hooks.json.

Enhanced Existing Commands

  • scan: Token budget now shows context window percentage
  • audit: Includes file-type breakdown, waste detection, cross-format conflicts

Install

npx cursor-doctor budget
npx cursor-doctor budget --pro
npx cursor-doctor conflicts