Audit IC datasheet numeric claims against spec bundles and LaTeX markers.
npx skills add https://github.com/rjwalters/bucket-brigade --skill datasheet-rjwalters
Please help me install this Agent Skill. Skill: datasheet Source: https://github.com/rjwalters/bucket-brigade/tree/main/.anvil/skills/datasheet Command: npx skills add https://github.com/rjwalters/bucket-brigade --skill datasheet-rjwalters
Customer-facing IC datasheets are high-stakes documents where a single wrong number can cause a board spin. Hand-authoring them introduces errors like contradictory specs, double-assigned pins, unrepresentable bus widths, silent spec changes, and pre-silicon values presented as final. This skill encodes the cleanup process distilled from real preliminary datasheets, ensuring every number traces to an authoritative source and passes mechanical integrity checks before reaching a customer.
Use the datasheet skill to produce a customer-facing IC datasheet by providing a thread brief and spec bundle, then run the parallel review and audit commands to validate every number and mechanical constraint before shipping.
High-intent search queries and answers about installing and using this skill.
To audit an IC datasheet, you back-check every numeric claim against a spec bundle and perform mechanical pre-flight checks via LaTeX markers to verify pin-map integrity and bus-width validity before customer release.
A spec source-of-truth audit validates every numeric claim in an IC datasheet against authoritative sources like model exports and RTL parameters, assigning a four-valued verdict to prevent contradictory specs and silent spec changes.
You check for double-assigned pins and bus-width errors by using deterministic mechanical pre-flight checks that read machine-readable LaTeX marker comments to ensure every pin is assigned exactly once and N-bit fields cover claimed ranges.
Yes, the datasheet workflow supports shared-die SKU coherence by cross-reading sibling datasheets during the audit to ensure shared specs like process, die, package, and absolute maximums remain identical across the product family.
The quality threshold for a customer-facing IC datasheet is a 9-dimension, 44-point rubric requiring a minimum score of 39 points to advance from the revision stage to the READY/AUDITED lifecycle state.
You label pre-silicon projected values versus measured specs by enforcing measured-versus-projected provenance labeling during the mandatory audit, ensuring preliminary values are never incorrectly presented as final silicon results.
xingwudaoAudit compiled quant strategy runtime plans against original specifications.
shawnpangCreate customer-facing documentation with structured workflows for help centers and APIs.
flonatGenerates computationally validated master revision plans from referee comments for LaTeX academic papers.
archibateTranslate English source texts into natural Chinese with iterative polish.
