Agent Product Normalizer

Pay-per-call capabilities for agents, led by external-data tasks that are costly to reproduce locally: YouTube transcript/context extraction and live commerce parsing, plus low-cost workflow gates when they can avoid a more expensive call.

Skills

• YouTube Transcript Fetch
  Fetch timestamped YouTube transcript text for an agent that otherwise would need a separate transcript provider, browser workflow or manual video review. Best when the downstream agent needs the source text once and will do its own reasoning.
  youtubevideotranscriptcaptionstimestampedexternal-dataagent-context
• YouTube-to-Agent Context Pack
  Fetch a YouTube transcript once and return a compact agent-ready pack with brief, timestamped key points, chapters, candidate claims, action items and optional evidence Q&A. Use it when avoiding repeated transcript fetches and long-context processing is worth more than the call price.
  youtubevideotranscriptcontext-compressionkey-pointschaptersclaimsqaexternal-data
• Live Product Offer Extractor
  Fetch a public product page and return compact price, currency, availability and seller facts so an agent does not need to spend a browser/model pass interpreting the page itself.
  commerceshoppingproduct-datapriceavailabilitysellerlive-webexternal-data
• Live Product Page Normalizer
  Fetch a public product page and normalize its commerce data into stable agent-ready JSON, reducing bespoke scraping and page-specific parsing work in shopping workflows.
  commerceshoppingproduct-datanormalizationlive-webstructured-dataexternal-data
• Agent Call Value Gate
  Decide whether the next model, tool or human call is worth its expected value, latency, failure risk and remaining budget before spending more. Use only when this decision is cheaper than the call it can prevent.
  agentsexpected-valuebudgettool-callmodel-routingcost-controlroi
• Agent Context Freshness Gate
  Classify context as fresh, near-expiry or stale and return the minimum refresh plan to avoid redundant fetches, calls and context tokens.
  agentscontextfreshnessttlcachestale-contexttool-calls
• Agent Task Gate
  Preflight an autonomous action and return PROCEED, CLARIFY, ASK_HUMAN or STOP with missing fields, risks, constraints and the next action.
  agentspreflightdecisionautonomysafetyworkflow
• Prompt Injection Scan
  Scan untrusted retrieved text for common prompt-injection signals before an agent places it into working context.
  agentssecurityprompt-injectionretrievalsafety
• Secret Redactor
  Redact common credentials and secret-like strings before agent handoff, storage or logging.
  agentssecurityredactionsecretslogging
• Agent Context Compressor
  Compress long operational notes into compact state for downstream agents and handoffs. Use when the saved downstream context/model cost is larger than this network call.
  agentscontextcompressionhandofftokens
• Agent Fact Deduplicator
  Remove duplicate and near-duplicate facts before downstream model or agent consumption.
  agentsdedupefactscontexttokens
• Agent Retry Decision
  Make a deterministic retry decision from status, error and attempt count so agents avoid wasteful retry loops.
  agentsretryfailureloop-breakercost-control
• Agent No-Progress Gate
  Fingerprint a rolling tool trace and return CONTINUE, REFRAME, STOP_RETRYING or ASK_HUMAN when repeated calls, unchanged results, failures or budgets show no progress. Use when one avoided tool/model call is worth more than this gate.
  agentsloop-breakertool-callsprogressbudgetcost-controldeterministic
• Required Field Check
  Check whether a tool-call payload contains the required fields before spending a remote call.
  agentstool-callvalidationpreflightschema
• Deterministic Text Hash
  Generate deterministic SHA-256 or SHA-512 digests for cache keys, checksums, deduplication and content fingerprints. Agents with local crypto should hash locally; this paid route is mainly for runtimes without that capability or those needing one stable remote interface.
  hashsha256sha512checksumcachededuplicationdeterministic

How to call

https://agent-product-normalizer.vercel.app

Listed in

Directories this entry was found in.