Calculate a one-dimensional layer and interface thermal-resistance scenario from supplied inputs. No hotspot, convection, transient or validated package-safety claim.

Compare supplied structures with explicit salt, protonation and stereochemical handling. For dataset reconciliation; no empirical identity or purity confirmation.

Relate supplied film thickness, resistivity and sheet resistance under stated uniform-film assumptions. For arithmetic checks; no measured device-performance claim.

Convert electrode potentials with supplied SHE offsets only when declared conditions match. Explicit sign convention; no reference-data or area validation.

Compare supplied molecules using a declared fingerprint and similarity metric. For dataset analysis; chemical similarity does not establish equivalent function or safety.

Fit supplied substrate/rate pairs by reciprocal Lineweaver-Burk OLS; report Km, Vmax and residuals. Exposes noise bias; no parameter confidence interval.

Summarize supplied particle-size distributions while preserving number, mass or volume weighting. For metrology-data analysis; no toxicology or safety inference.

Check declared property, unit, temperature, method, uncertainty, source and revision against supported unit conventions. No independent measurement verification.

Calculate a bounded set of molecular descriptors using a pinned toolkit version. For research dataset preparation; descriptors do not certify drug suitability or toxicity.

Check supplied reaction declarations for atom and charge accounting. For record quality control; no reaction feasibility, synthesis procedure or safety claim.

Estimate differential strain from supplied expansion coefficients and temperature changes. For preliminary scenarios; no stress, delamination or reliability certification.

Evaluate supplied absorption observations under a declared band-gap model with fit-window sensitivity. No universal band-gap or integrated-device performance claim.

Revisit up to three approved, rights-reviewed public text sources. Check exact quoted passages in versioned extracted text, compare declared prior text hashes and return bounded locators, retrieval times and hashes. No paywall bypass, PDF/JavaScript execution, factual certification or completeness guarantee.

Check up to 250 supplied product records for duplicate SKUs, nonpositive or malformed decimal prices, currency-code shape, and optional GTIN check digits. Return bounded row-level findings. No registry lookup, identifier ownership check, automatic repair or marketplace approval guarantee.

Compare stored, dated unpaid observations of one approved endpoint. Report quote, network, schema and reachability changes with sample counts, largest sampling gap and snapshot age. Sparse observations are not continuous uptime, customer calls, settlement or delivery evidence.

Check up to 500 supplied CSV records for quoting, column counts, duplicate headers, spreadsheet-formula prefixes and caller-declared scalar types. Return bounded row/column findings and input integrity hashes. No cell execution, automatic repair, PII detection or import-success guarantee.

Fetch one approved public endpoint without payment. Inspect Base and Arbitrum native-USDC quotes against caller-declared payee and spending caps, plus bounded discovery-schema observations and latency. No signing, settlement test, paid delivery, uptime guarantee or safety clearance.

Match supplied Raman peak positions by nearest unused reference within an absolute tolerance. No linewidth analysis or unique material identification.

Check declared process temperature and corrosivity against substrate constraints and supplied evidence records. Flags conflicts; no fabrication recipe or compatibility approval.

Score supplied material candidates by weighted cohort min-max criteria; withhold incomplete scores without imputation. No qualification or engineering approval.