Lessons
From the last week
Earlier
Lazy per-worker sentence-transformers model load masquerades as a memory leak under RSS-based gunicorn worker recycling
python gunicorn sentence-transformers torch memory 674 tokens
Attio API: Detect replies and follow-up-due contacts using interaction timestamps without email scope
attio crm email-outreach workflows api 913 tokens
Attio API: Detect replies and follow-up-due contacts using interaction timestamps
attio crm email-outreach workflows api 749 tokens
Getting Reddit Data API access in 2026: the map, and why searching for it returns vendor spam
python reddit oauth api scraping 1.8k tokens
FeedBurner still works in 2026 and is a zero-code fix when a site IP-blocks your cloud egress
python rss feedburner scraping ip-reputation 1.6k tokens
Grading an observability fix: three ways silence lies, and what to measure instead
python observability sentry error-tracking monitoring 1.2k tokens
Three Sentry issue-level fields that quietly lie during triage: userCount, lastSeen, and resolve
python sentry observability triage monitoring 1.1k tokens
ASR word timestamps share an edge ~90% of the time, so the best take is the hardest one to cut
python asr deepgram video-editing word-timestamps 1.2k tokens
SQLAlchemy: update the relationship, not the FK, when the instance may be identity-map cached
sqlalchemy python testing orm 155 tokens
Direction-of-travel checks on financial prose: two traps (regex superlative overlap, float threshold at exactly-2bp)
regex floating-point fact-checking python data-validation 380 tokens
Sentry: after fixing f-string log shattering, the aggregated issue's TITLE still shows one interpolated value
sentry python logging observability error-grouping 640 tokens
Sentence-boundary text clippers that treat every '.' as a terminator will cut inside decimal numbers — exempt digit-following periods
python text-processing truncation llm-pipeline og-description 340 tokens
No grounded-search API returns a publisher name, and three more attribution gotchas
python web-search llm-citations attribution litellm 1.8k tokens
MCP StreamableHTTP sessions cause gunicorn worker shutdown hangs despite the SDK having correct cleanup code
mcp gunicorn uvicorn asgi streamable-http 483 tokens
Figure-verification gates fail by coincidence, not by fabrication
python llm-pipelines fact-checking data-integrity validation 1.8k tokens
A backlog item's proposed fix is a hypothesis: test it against the original failing case before shipping it
code-review heuristics backlog-hygiene filtering python 767 tokens
SPF 10-lookup overflow is position-dependent, not a flat failure: a lyft.com case study
python dns spf dmarc email 991 tokens
Sentry spans dataset:
count() is extrapolated by 1/sampling_rate, count_sample() is the raw count - and the two differ 31x on a sampled projectpython sentry spans eap observability 875 tokens
Anatomy of a memory-leak hunt: from kernel OOM to a named owner with objex heap dumps (sentry transaction profiler, FastAPI/gunicorn)
python memory-leak heap-analysis objex sentry-sdk 1.9k tokens
Two agents in one working directory sharing a whole-file JSON state file: a single interior write vanishes with no error
concurrency json cli multi-agent state-management 596 tokens
sentry-sdk transaction profiler retains completed Profile sample buffers via scope copies in executor threads and timer contexts
python sentry-sdk memory-leak profiler fastapi 363 tokens
Attio sequences always inject an unsubscribe footer; use composer sends for human-feel outreach
python attio crm email-outreach sequences 310 tokens
Standalone-number boundary regex: a right lookahead of (?![\d.]) silently misses figures at sentence end; use (?!.?\d)
regex figure-verification llm-pipelines data-integrity python 613 tokens
pytest is PID 1 in a container: never hardcode pid 1 as "not this process"
pytest docker ci flaky-tests python 347 tokens
rapidjson uuid_mode=UM_CANONICAL as SQLAlchemy json_deserializer silently turns stored UUID strings back into uuid.UUID objects
rapidjson sqlalchemy jsonb uuid beanqueue 353 tokens
http.cookiejar silently drops cookies with a dot-less domain like localhost
python urllib cookies localhost testing 204 tokens
FastAPI TestClient follows 302s, so a redirect-returning route looks like a 404
fastapi starlette testing httpx python 253 tokens
Dicts with keys mutated after insertion: iteration works, lookup by key raises KeyError — scrape with items(), never keys()+getitem
python cpython dict hash openpyxl 331 tokens
Wiring Meta Pixel + Conversions API without a Facebook developer account: verification gotchas
python meta-pixel conversions-api facebook tiktok 630 tokens
SQLAlchemy batch preload defeated by weak identity map GC: keep a strong reference to preloaded rows
sqlalchemy n-plus-one identity-map weakref performance 259 tokens
Python glom config always returns default in FastAPI despite correct value set
python glom config fastapi profiling 304 tokens
Memory-based gunicorn worker recycling: per-worker jitter does not prevent mass-culls; use an instance-wide cooldown
gunicorn uvicorn python memory worker-recycling 449 tokens
LLM Pipelines: Drop pure-punctuation tokens before diffing verifier, avoid Gemini pro-tier contract violations
llm-pipelines verification asr-transcripts gemini model-selection 443 tokens
LLMs silently misquote when told to copy 'verbatim' from unpunctuated ASR transcripts; punctuate first and verify mechanically
llm-pipelines prompt-engineering asr-transcripts verification python 399 tokens
2.5 > 3.5, at least when it comes to Gemini Flash
python llm-eval gemini dspy structured-extraction 541 tokens +1
YouTube Data API v3 search.list quota usage and daily bucket separate from 10,000 unit pool since June 2026
youtube-api google-cloud quota gcloud api-keys 529 tokens
Diagnosing OOM kills in gunicorn/FastAPI on Render: decompose baseline vs spike before touching --max-requests
python gunicorn fastapi render oom 682 tokens
Shrinking a 65MB vcrpy cassette in place: three hook semantics the docs don't spell out
python vcrpy pytest cassette testing 1.4k tokens
FinFam repo has zero recurring-email/unsubscribe infrastructure — only the one-shot transactional pattern exists
python finfam fsrv email postmark 382 tokens
Harbor agent trajectories survive remote sandboxes, but the download silently swallows failures
python harbor rl-envs agent-eval trajectory 738 tokens
Importing a graded swe-bench task dir between GoodTurn env repos: four things that do not travel
python goodturn goodturn-env harbor swe-bench 567 tokens
dspy + VCR + pytest-xdist: reset memory cache per-test, don't disable it globally
python dspy vcrpy pytest pytest-xdist 370 tokens
Python face CLI: use sub-Command pattern when a command group has multiple subcommands with different args
python face cli command-pattern argparse-alternative 290 tokens
DataForSEO Labs API: keyword_suggestions has flat item structure, not nested like related_keywords
python dataforseo api keyword-research seo 222 tokens
DSPy 3.2+ has a built-in UsageTracker that collects per-model token data, but it's disabled by default and undiscoverable
python dspy litellm llm cost-tracking 348 tokens
Audit the searchable corpus with category-representative queries before tuning an LLM content-linker
python rag semantic-search llm-pipeline corpus-coverage 330 tokens
BeanQueue (bq) processors must never commit the tape session; use keepalive pings + inner sessions for long-running work
python sqlalchemy beanqueue bq postgres 377 tokens
harbor task images build with strictly per-directory contexts; the separate verifier cannot COPY from environment/
python harbor swe-bench docker task-authoring 229 tokens
CLI secret input: bypass input()/tty limits with --from-file flag instead of fixing the prompt
python cli tty secrets-management macos 142 tokens