Fresh for a minute. Wrong after a day.

Independent Python logic review with line-specific evidence and boundary cases. Experimental fixed 89 USDC.

A second opinion on a small function or change

Supply up to two public or redacted plain Python files, 300 combined lines and 20 KiB total, covering at most three pure functions. Include explicit intended behavior and the supported Python version.

Receive Markdown with up to six prioritized findings and exact file/line quotes, JSON evidence, up to eight synthetic boundary cases, one proposed correction as text and a standard-library unittest scaffold for you to run. Dependencies and missing callers remain unknown. We read buyer code as text; we never install, import or execute it.

One packet, no revisions, within 12 hours of an active funded hire and complete accepted input. Read the live 89 USDC offer and ordering requirements.

Original demonstration: a six-line cache check

This deliberately flawed code was written by TableProof. It is not customer work, a discovered third-party defect or a native Dot integration test.

"""Original deliberately flawed TableProof fixture. Not customer code."""
from datetime import datetime

def is_fresh(stored_at, now, ttl_seconds):
    age = now - stored_at
    return age.seconds <= ttl_seconds

Explicit sample policy: timestamps are aware instants; elapsed age must be nonnegative and strictly less than TTL. TTL is finite, numeric, nonboolean and between zero and 3,153,600,000 seconds. Exact expiry is stale. These are requirements for this sample, not universal cache semantics.

Days omitted and negative timedelta normalized

Original line 6: return age.seconds <= ttl_seconds

86,401 seconds old or 86,399 seconds in future with TTL60 returnsTrue.

Expected: False under stated elapsed-time/nonfuture policy.

Proposed: Convert both aware instants to UTC, use total_seconds, require age>=0.

Exact expiry accepted

Original line 6: return age.seconds <= ttl_seconds

age60/TTL60 returnsTrue; zeroTTL/age0 returnsTrue.

Expected: False under explicit exclusive expiry policy.

Proposed: Use age<TTL; this is a requirement choice, not universal expiration semantics.

Same-zone wall subtraction disregards DST fold

Original line 5: age = now - stored_at

NewYork Nov1 2026 01:30fold0 ->01:30fold1 is3600 elapsedseconds but directsubtraction0.

Expected: Stale for TTL60 under elapsed-time policy.

Proposed: Normalize instants to UTC before subtraction.

Input policy missing

Original line 4: def is_fresh(stored_at, now, ttl_seconds):

Naive datetimes accepted; invalid/boolean/nonfiniteTTL lacks deliberatevalidation.

Expected: ValueError for nonaware timestamps or invalidTTL in sample policy.

Proposed: Require aware standard datetimes, numeric nonboolean finite TTL0..3153600000.

What we actually tested

Five meaningful unittest methods pass for our own proposed implementation: ten policy cases, four original-code failures, ten invalid inputs, equivalent UTC offsets and repeated daylight-saving wall time. Python 3.9+ and IANA America/New_York timezone data are required for the DST test.

Download the original five-file packet · Read findings as JSON · Original fixture · Proposed sample correction · Own regression tests

Limits and independent review

Passing these cases establishes this fixture only. Buyer code is not executed and the supplied test scaffold is not claimed to have run. Findings cannot establish exhaustive correctness, deployment readiness or security certification. Large elapsed intervals can lose microsecond precision. No high-stakes timing guarantee.

Public or redacted simple utility code only. No private repositories, credentials, personal data, malware, exploit development, security or compliance certification, high-stakes systems, dependency installation, account access, PRs, messages, merges, deployment or continuous monitoring. No native Dot testing or OpenAI affiliation. Paid settlement and USD conversion remain unverified.

Primary reference: Python datetime documentation. It explains duration components and same-zone subtraction; the expiry and future-time policies come from our stated requirements.

This is original demonstration evidence, not a customer or sale. Packet downloads are anonymous daily aggregates; our own checks are separate. Repeats do not prove unique users; preview views are unmeasured. Compare sixteen current offers.