baronguyen001
User8 production-tested Claude skills: automation, Gemini cost/structured output, OSS bounty scouting, ML validation.
Categories
Indexed Skills (25)
backtest-comparator
Compare strategy variants across multiple years or folds and flag the ones that only look good on average - rank by mean fold score but surface dispersion and worst-fold so a variant that overfits one lucky period does not win silently. Use for compare backtest variants, flag overfit strategy, per-fold consistency, year by year backtest, or which parameter set is robust.
cron-dispatch
Schedule any script to run on a recurring schedule on Windows (Task Scheduler) or Linux (cron) - register, list, and remove jobs from one command, with logging to a file and a guard against overlapping runs. Use for schedule a script, run nightly, set up a cron job, windows task scheduler, or run on a timer.
gemini-flash-budget
Run high-volume, low-cost Gemini extraction on a Flash model with thinking_budget=0 so simple, well-scoped prompts skip thinking tokens entirely - cheapest and fastest path for bulk classification and field extraction. Use for cheap gemini extraction, disable gemini thinking, thinking_budget 0, bulk classify with flash, or high volume gemini calls.
gemini-prompt-cache
Cache a long, stable Gemini system-prompt prefix once and reuse the handle across many calls so repeated instruction tokens bill at the cheaper cached rate instead of full input. Use for cache gemini system prompt, context caching, cut gemini input cost, cachedContent, or reuse a long prompt prefix.
github-label-scout
Scan a public GitHub repo for open issues carrying a given label and drop the ones already taken - skips pull requests returned by the issues endpoint and any issue with assignees - using the public REST API with no token required. Use for find github issues by label, good first issue scout, skip assigned issues, github issue triage, or open source contribution finder.
gmail-imap-digest
Pull recent email over IMAP, keep only messages from an allowlist of trusted sender domains, and render a daily digest - credentials from env using a provider app-password, never a login password. Use for email digest, fetch newsletters by IMAP, summarize inbox daily, gmail app password fetch, or build a morning email roundup.
pipeline-orchestrator
Chain a scrape -> AI -> alert pipeline where each stage is a plain callable, with exponential-backoff retry per stage and a JSON state checkpoint between stages so a crashed or rate-limited run resumes instead of restarting. Use for chain pipeline steps, scrape then summarize then notify, retry between stages, resumable pipeline, or orchestrate a job.
shortform-script
Draft a short-form TikTok/Reels/Shorts or long-form video script from a topic and creator persona file - hook variants, beat-structured body, CTA, b-roll cues, caption overlays, and [STORY]/[NUMBER] slots so the model never fabricates personal facts. Use for write a video script, tiktok script, youtube script, or /shortform-script.
gemini-vision-extract
Extract typed JSON from an image with Gemini - pass a receipt, invoice, screenshot, or chart and get fields back validated against a Pydantic schema, not prose. Key from the environment only. Use for read a receipt, parse an invoice image, OCR into structured data, extract fields from a screenshot, or image to JSON with gemini.
playwright-login-session
Log in once in a real browser, save the Playwright storage_state, then start every later run already signed in - no re-login, no committed cookies. Use for reuse a logged-in session, skip login on each run, save browser cookies, storage_state, or keep a scraper authenticated.
proxy-rotator
Rotate a pool of HTTP/SOCKS proxies with round-robin selection, failure tracking, and a cooldown that benches dead proxies before retrying. BYO proxy list via env - none are shipped. Use for rotate proxies, spread requests across proxies, avoid IP bans, retry through a different proxy, or proxy health check.
webhook-receiver
Receive inbound webhooks with no web framework - a tiny stdlib HTTP server that verifies an HMAC signature in constant time and queues each payload to disk for a worker to process. Use for receive a webhook, handle an incoming POST, trigger a script from a webhook, verify a webhook signature, or stripe/github/telegram webhook intake.
csv-report-writer
Turn a run's list of result dicts into a schema'd CSV and a Markdown table from one column spec - declare columns once, emit both, with stable ordering and safe escaping, stdlib only, no pandas. Use when the user asks to write results to CSV, export a report, make a markdown summary table, or save a run's output as a spreadsheet.
env-config-loader
Load a .env file, validate that required keys are present, and read typed values (int/bool/float/list) with clear error messages - stdlib only, no python-dotenv, no pydantic. Use when the user asks to load a .env, validate config, fail fast on missing env vars, read an env var as an int or bool, or parse a comma-separated env list.
http-retry-session
A hardened requests.Session with urllib3 Retry - exponential backoff, a status-forcelist for 429/5xx, honored Retry-After, and a hard connect/read timeout on every call. Use when the user asks to make HTTP calls more reliable, add retries to requests, handle 429 rate limits, set request timeouts, or stop a scraper from hanging on a slow endpoint.
rate-limit-budget
A token-bucket pacer that keeps API calls under a per-second rate AND a hard per-run budget - acquire() blocks just enough to smooth bursts, and a spend cap stops the run before it blows a free-tier quota. Generic and stdlib-only. Use when the user asks to rate-limit API calls, throttle requests, stay under a quota, cap calls per run, or avoid 429s by pacing instead of retrying.
slack-webhook-alerter
Post run results to a Slack incoming webhook from Python - webhook URL from env, Block Kit sections plus a colored attachment, no Slack SDK. Use when the user asks to send a slack alert, notify slack, post to a slack channel, mirror telegram alerts to slack, or wire a script to push a status message to Slack.
sqlite-state
Give scheduled scripts memory between runs with one SQLite file - a seen-set for dedup, a key/value cursor to resume where you left off, and order-preserving new-item filtering. Use for dedup across runs, don't re-alert the same item, remember the last id, resume a scraper, or persist state between cron runs.
algora-bounty-scout
Scout public Algora org bounty boards at algora.io/<org>/bounties?status=open - parse amount and linked GitHub issue, dedupe, then filter already-assigned issues, saturated races, joke bounties, dead-lane orgs, and maintainer comments that say the bounty is inactive. Use for scout algora, find OSS bounties, or open source bounty hunting.
config-audit-checklist
Audit a project for config drift before a backtest or deploy - compare the documented config snapshot against .env and README defaults, flag mismatches, and check the snapshot last-verified date. Use for config audit, params drift, pre-deploy checklist, or production config check.
gemini-cost-tracker
Wrap Gemini API calls to log token usage and USD cost per call, including input, output, cached, and thinking tokens, with per-model and per-session summaries. Use when the user asks to track gemini cost, inspect gemini token usage, or estimate repeated LLM call spend.
gemini-structured-output
Get validated JSON out of Gemini using a Pydantic model - convert the model to responseSchema, set responseMimeType, then parse and validate the response, with a salvage parser for JSON wrapped in prose. Use for gemini structured output, gemini json schema, pydantic to gemini, or responseSchema.
pr-body-formatter
Generate a complete GitHub PR body with Summary, Changes, How tested, Test plan, and issue link from a branch diff plus issue reference, and never ship a placeholder body. Use for pr body template, format pr description, write a PR description, or API/CLI PR creation.
telegram-alerter
Send a Telegram message/alert from Python - bot token and chat id from env, HTML formatting, inline buttons, photo attach, retry on 429. Use when the user asks to send a telegram alert, notify telegram, telegram bot send, or wire a script to push notifications to Telegram.
walk-forward-runner
Set up a leakage-free walk-forward rolling split for any time-series model or strategy and flag overfit parameters where a value ranks high on train but low on validation. Use for walk forward validation, time series cross validation, no leakage split, rolling backtest, or parameter overfit checks.
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.