75
GitHub Stars
1
Bundled Files
2 months ago
Catalog Refreshed
4 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstrat where the catalogue uses aiagentskills.
npx veilstrat add skill getsentry/warden --skill notseer- SKILL.md4.4 KB
Overview
This skill performs high-precision bug detection for TypeScript code, producing only provable bug reports rather than suspicions. Every reported issue includes a concrete proof: file and line, trigger input, observed failure, root cause, and confidence level. The tool targets logic errors, null/undefined access, async mistakes, off-by-one and edge-case failures.
How this skill works
The agent reads code context and traces data flow to identify definite failure points. It applies a five-point proof model: exact location, observable failure, trigger condition, root cause, and consensus confidence. Only defects that can be proven from code and reproduced with a specific input or condition are reported. Reports are concise and include a suggested fix when the remedy is clear.
When to use it
- Before merging TypeScript changes to catch definite bugs that would crash or corrupt data
- When you need bug reports an engineer can act on immediately with no speculation
- During code reviews for critical paths: async flows, boundary checks, and state updates
- When tests are missing for edge cases and you want concrete failing scenarios
- For PRs touching shared state, mutations, or complex control flow
Best practices
- Only report issues you can prove with a concrete trigger and observable failure
- Read surrounding files to understand intent before calling a bug
- Provide exact file path and line number plus the minimal input that reproduces the bug
- Classify severity (critical/high/medium) based on crash, commonness, and data corruption
- Offer a suggested fix only when it is straightforward and unambiguous
Example use cases
- Detecting null access where optional chaining hides a deeper missing validation
- Finding an off-by-one in a loop that drops the last element under certain array lengths
- Reporting a missing await that causes unhandled promise rejections in initialization
- Proving a stale closure captures an old value leading to incorrect state updates
- Identifying division-by-zero or empty-array conditions that will throw at runtime
FAQ
No. It only reports provable bugs that cause wrong output, crashes, or data corruption.
What if I can’t produce all five proof points?
Do not report the issue. The skill requires all five points to avoid speculation.