Hi Hacker News! I built TimeCodeSecurity (TCS), an open-source static security engine for Python.

Most existing open-source SAST tools rely on regex patterns or LLMs, leading to high false-positive rates and alert fatigue. TCS takes a compiler-level deterministic approach: it parses Python's Abstract Syntax Tree (AST) and tracks dataflow from function parameters directly into sensitive execution sinks (subprocess.run, raw SQL cursors, and path operations).

Key highlights: - Deterministic AST taint tracking (Zero regex / Zero AI guessing) - Closed-loop remediation: Patches are validated via AST syntax compilation and re-scanned in-memory before touching disk - Chronological visual proof graphs: Source -> Taint -> Sink - Sub-second execution (~700ms)

You can run it directly in your terminal:

  pip install git+https://github.com/kushigaur3103-svg/time-code-security.git
  tcs scan ./your_project --fix

Would love feedback on the AST traversal and in-memory verification approach! https://github.com/kushigaur3103-svg/time-code-security And that's my linkdin link https://www.linkedin.com/in/ayush-gaur-1488b1420?utm_source=share_via&utm_content=profile&utm_medium=member_android

Please cite claude or llms when they write code.

“Most existing open-source SAST tools rely on regex patterns or LLMs”, this may be the case with new tools in the last 7 months but before then has definitely not been. (Bandit, semgrep, black duck, etc…)

I do like this as an intro project, but why did you choose to use Python source code as your validators and remediation? I initially assumed for llm.

[dead]

[dead]