ln-628-concurrency-auditor
npx machina-cli add skill levnikolaevich/claude-code-skills/ln-628-concurrency-auditor --openclawPaths: File paths (
shared/,references/,../ln-*) are relative to skills repo root. If not found at CWD, locate this SKILL.md directory and go up one level for repo root.
Concurrency Auditor (L3 Worker)
Specialized worker auditing concurrency and async patterns.
Purpose & Scope
- Worker in ln-620 coordinator pipeline
- Audit concurrency (Category 11: High Priority)
- Check race conditions, async/await, thread safety
- Calculate compliance score (X/10)
Inputs (from Coordinator)
Receives contextStore with tech stack, language, codebase root, output_dir.
Workflow
- Parse context + output_dir
- Check concurrency patterns
- Collect findings
- Calculate score
- Write Report: Build full markdown report in memory per
shared/templates/audit_worker_report_template.md, write to{output_dir}/628-concurrency.mdin single Write call - Return Summary: Return minimal summary to coordinator
Audit Rules
1. Race Conditions
What: Shared state modified without synchronization
Detection Patterns:
| Language | Pattern | Grep |
|---|---|---|
| Python | Global modified in async | global\s+\w+ inside async def |
| TypeScript | Module-level let in async | ^let\s+\w+ at file scope + async function modifies it |
| Go | Map access without mutex | map\[.*\].*= without sync.Mutex in same file |
| All | Shared cache | cache\[.*\]\s*= or cache\.set without lock |
Severity:
- CRITICAL: Race in payment/auth (
payment,balance,auth,tokenin variable name) - HIGH: Race in user-facing feature
- MEDIUM: Race in background job
Recommendation: Use locks, atomic operations, message queues
Effort: M-L
2. Missing Async/Await
What: Callback hell or unhandled promises
Detection Patterns:
| Issue | Grep | Example |
|---|---|---|
| Callback hell | \.then\(.*\.then\(.*\.then\( | .then().then().then() |
| Fire-and-forget | async.*\(\) not preceded by await | saveToDb() without await |
| Missing await | return\s+new\s+Promise in async function | Should just return await or return value |
| Dangling promise | \.catch\(\s*\) | Empty catch swallows errors |
Severity:
- HIGH: Fire-and-forget async (can cause data loss)
- MEDIUM: Callback hell (hard to maintain)
- LOW: Mixed Promise styles
Recommendation: Convert to async/await, always await or handle promises
Effort: M
3. Resource Contention
What: Multiple processes competing for same resource
Detection Patterns:
| Issue | Grep | Example |
|---|---|---|
| File lock missing | open\(.*["']w["']\) without flock or lockfile | Concurrent file writes |
| Connection exhaustion | create_engine\(.*pool_size check if pool_size < 5 | DB pool too small |
| Concurrent writes | writeFile or fs\.write without lock check | File corruption risk |
Severity:
- HIGH: File corruption risk, DB exhaustion
- MEDIUM: Performance degradation
Recommendation: Use connection pooling, file locking, asyncio.Lock
Effort: M
4. Thread Safety Violations
What: Shared mutable state without synchronization
Detection Patterns:
| Language | Safe Pattern | Unsafe Pattern |
|---|---|---|
| Go | sync.Mutex with map | map[...] without Mutex in same struct |
| Rust | Arc<Mutex<T>> | Rc<RefCell<T>> in multi-threaded context |
| Java | synchronized or ConcurrentHashMap | HashMap shared between threads |
| Python | threading.Lock | Global dict modified in threads |
Grep patterns:
- Go unsafe:
type.*struct\s*{[^}]*map\[withoutsync.Mutexin same struct - Python unsafe:
global\s+\w+in function +threading.Threadin same file
Severity: HIGH (data corruption possible)
Recommendation: Use thread-safe primitives
Effort: M
5. Deadlock Potential
What: Lock acquisition in inconsistent order
Detection Patterns:
| Issue | Grep | Example |
|---|---|---|
| Nested locks | with\s+\w+_lock:.*with\s+\w+_lock: (multiline) | Lock A then Lock B |
| Lock in loop | for.*:.*\.acquire\(\) | Lock acquired repeatedly without release |
| Lock + external call | .acquire\(\) followed by await or requests. | Holding lock during I/O |
Severity: HIGH (deadlock freezes application)
Recommendation: Consistent lock ordering, timeout locks (asyncio.wait_for)
Effort: L
6. Blocking I/O in Event Loop (Python asyncio)
What: Synchronous blocking calls inside async functions
Detection Patterns:
| Blocking Call | Grep in async def | Replacement |
|---|---|---|
time.sleep | time\.sleep inside async def | await asyncio.sleep |
requests. | requests\.(get|post) inside async def | httpx or aiohttp |
open() file | open\( inside async def | aiofiles.open |
Severity:
- HIGH: Blocks entire event loop
- MEDIUM: Minor blocking (<100ms)
Recommendation: Use async alternatives
Effort: S-M
Scoring Algorithm
MANDATORY READ: Load shared/references/audit_scoring.md for unified scoring formula.
Output Format
MANDATORY READ: Load shared/templates/audit_worker_report_template.md for file format.
Write report to {output_dir}/628-concurrency.md with category: "Concurrency" and checks: race_conditions, missing_await, resource_contention, thread_safety, deadlock_potential, blocking_io.
Return summary to coordinator:
Report written: docs/project/.audit/ln-620/{YYYY-MM-DD}/628-concurrency.md
Score: X.X/10 | Issues: N (C:N H:N M:N L:N)
Reference Files
- Worker report template:
shared/templates/audit_worker_report_template.md - Audit scoring formula:
shared/references/audit_scoring.md - Audit output schema:
shared/references/audit_output_schema.md
Critical Rules
- Do not auto-fix: Report only, concurrency fixes require careful human review
- Language-aware detection: Use language-specific patterns (Go sync.Mutex, Python asyncio.Lock, Java synchronized)
- Effort realism: S = <1h, M = 1-4h, L = >4h
- Critical path escalation: Race conditions in payment/auth = CRITICAL, regardless of other factors
- Exclusions: Skip test files, skip single-threaded CLI tools, skip generated code
Definition of Done
- contextStore parsed (language, concurrency model, output_dir)
- All 6 checks completed (race conditions, missing await, resource contention, thread safety, deadlock potential, blocking I/O)
- Findings collected with severity, location, effort, recommendation
- Score calculated per
shared/references/audit_scoring.md - Report written to
{output_dir}/628-concurrency.md(atomic single Write call) - Summary returned to coordinator
Version: 3.0.0 Last Updated: 2025-12-23
Source
git clone https://github.com/levnikolaevich/claude-code-skills/blob/master/ln-628-concurrency-auditor/SKILL.mdView on GitHub Overview
Concurrency Auditor (L3) analyzes code for race conditions, missing async/await usage, resource contention, and deadlock risks. It collects findings with severity, location, and recommended fixes, then writes a Markdown report to the designated output directory.
How This Skill Works
The auditor receives context from the coordinator, scans code against language-specific detection patterns (race conditions, missing async/await, resource contention, thread safety), and aggregates findings. It then computes a compliance score and writes a single Markdown report to {output_dir}/628-concurrency.md, returning a concise summary to the coordinator.
When to Use It
- Auditing a codebase for race conditions in worker pipelines.
- Validating proper async/await usage in async-heavy modules.
- Identifying resource contention like file locks or DB pool exhaustion.
- Checking thread safety in shared mutable state across goroutines/threads.
- Generating a formal audit report for stakeholders.
Quick Start
- Step 1: Provide contextStore and target output_dir from the coordinator to the worker.
- Step 2: Run the concurrency auditor to collect findings and compute the score.
- Step 3: Review {output_dir}/628-concurrency.md and apply fixes; re-run to verify.
Best Practices
- Apply strict synchronization around shared state using mutexes, atomic ops, or message queues.
- Ensure all asynchronous operations are awaited or properly handled; avoid fire-and-forget without checks.
- Use resource pools and file locking to prevent contention and data corruption.
- Adopt language-appropriate thread-safety primitives (e.g., Mutex, Arc<Mutex<T>>, synchronized blocks).
- Validate findings with tests and generate the in-memory report before writing to disk.
Example Use Cases
- Python: global variable modified inside an async def leading to race on a shared cache.
- TypeScript: module-level let mutated by an async function without proper synchronization.
- Go: map access without a mutex inside the same struct, causing data races.
- JavaScript/Node: fire-and-forget call (e.g., saveToDb()) without awaiting, risking ordering issues.
- JavaScript/Promises: dangling promise with an empty .catch() swallowing errors.