tax-reference-currency
Skill1 starsChanged 34 days agoMITView on GitHub
Refresh tax reference data and dependent applications with explicit years, official sources, historical preservation, deployment evidence and protected taxpayer reconciliation.
In this repository43 other files it gives its agents.
- ansible-hardeningSkill
- cavetechSkill
- ci-cdSkill
- claude-to-cursorSkill
- code-consistencySkill
- code-qualitySkill
- config-auditorSkill
- container-sandboxSkill
- cpa-tax-specialistSkill
- database-designSkill
- definition-of-doneSkill
- differential-reviewSkill
- exact-head-reviewSkill
- fast-ciSkill
- grill-meSkill
- inbound-syncSkill
- infosec-architectSkill
- insecure-defaultsSkill
- lonely-sysadminSkill
- mcp-createSkill
- mcp-hardeningSkill
- model-evaluationSkill
- modern-pythonSkill
- ms-office-suiteSkill
- mutation-checkSkill
- plan-to-execution-handoffSkill
- prd-to-planSkill
- property-based-testingSkill
- pr-truth-triageSkill
- repo-hardening-workflowSkill
- resume-writerSkill
- skill-updaterSkill
- software-developerSkill
- software-development-lifecycleSkill
- solutions-architectSkill
- stay-readySkill
- supply-chain-risk-auditorSkill
- systematic-debuggingSkill
- tddSkill
- token-miserSkill
- triage-issueSkill
- verification-before-completionSkill
- write-a-prdSkill
The file
We split this skill into the rules it gives an agent
10 rules. About 0 are common, 10 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈10
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 21
Do not apply a next-year deduction or reporting threshold to the current year merely because it is the newest publication.
- Distinctive: line 28
Do not assume a repository name, model alias, health check or README proves currency.
- Distinctive: line 32
Start from official agency sources, such as:
- Distinctive: line 44
Check state conformity provision by provision.
- Distinctive: line 49
Keep draft and final forms distinct.
- Distinctive: line 71
Do not retrain or promote a model automatically because reference material changed.
- Distinctive: line 72
Do not use private tax documents, financial exports or sessions as training data without separate appropriate authorization and admission.
- Distinctive: line 78
Use original protected exports and genuine filing/payment evidence.
- Distinctive: line 88
Keep credentials, identifiers, genuine transaction rows, source documents, private prompts and raw backups out of repositories and ordinary …
- Distinctive: line 94
Run full checks at exact proposed heads on supported runtimes, independent source/ calculation examples, real service contracts, browser …
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Tax-reference currency
A source-file update is not a deployed-service update or a reconciliation of a person's financial records. Report each layer independently. This procedural skill contains no authoritative runtime tables, private context or professional credentials.
Establish scope and preserve history
Confirm the as-of date, tax year, jurisdiction, required forms and actual application entry points. Distinguish current-year rules from published future-effective values. Do not apply a next-year deduction or reporting threshold to the current year merely because it is the newest publication. Preserve prior returns, original documents, closed periods, accepted rule versions and historical-year examples.
Inventory source, installed and running identities separately. Preserve dirty and unpushed work. Discover dependent calculators, document extractors, retrieval indexes, model-serving wrappers, caches, training/evaluation consumers and user-facing exports. Do not assume a repository name, model alias, health check or README proves currency.
Verify public-law records
Start from official agency sources, such as:
- https://www.irs.gov/forms-pubs
- https://www.irs.gov/newsroom
- https://www.ssa.gov/cola/
- https://www.ftb.ca.gov/tax-pros/law/conformity.html
- https://tax.hawaii.gov/news/announce/
Record source identity, tax/form year, effective interval, jurisdiction, publication revision, verification time, content hash and independent review. A hash of a JSON summary is not the hash of its underlying source document. Keep missing raw archives explicitly missing. A downloaded source is not automatically accepted as applicable.
Check state conformity provision by provision. Do not copy federal treatment into a state return or treat income tax, gross-receipts tax and lodging tax as one obligation. Information-return issuance thresholds are not taxable-income exclusions. Missing source documents, basis, income or withholding are unknown values, never assumed zero.
Keep draft and final forms distinct. Do not change the printed year on an old form, extrapolate brackets using a guessed inflation factor, or quietly use the latest table for an unsupported year. Stop unsupported material calculations with a specific reason.
Update code and dependent services
Trace actual API, tool, document, graph and export paths to the calculation/source used. Fix duplicate executable constants and stale seed material under source-backed tests; classify valid historical occurrences before editing. Do not mass-replace old numbers in archived evidence or question banks. Preserve an explicit version boundary.
Pin shared reference releases and test consumer parity. Include data files in built packages/images. Test strict year/status selection, boundary amounts, unsupported scope, missing inputs and source-to-output mappings with independent expected values. A changed contract requires caller tests, not silently suppressed old failures.
Stage retrieval updates in a new reviewed snapshot. Enforce year, jurisdiction, source identity and applicability through real graph/API/streaming/tool paths. Similarity scores, reranking, model confidence and fluent citations are not proof of legal applicability. Invalidate affected caches and incompatible qualification artifacts by their exact source/model/prompt bindings; preserve historical audits.
Do not retrain or promote a model automatically because reference material changed. Do not use private tax documents, financial exports or sessions as training data without separate appropriate authorization and admission. Preserve held-out evaluation custody; a model cannot approve its own labels or declare its own tax answer authoritative.
Reconcile protected taxpayer records separately
Use original protected exports and genuine filing/payment evidence. Establish the latest reconciled cutoff rather than inferring it from a code commit. Import new records idempotently, link corrections/reversals, and reconcile gross receipts, settlements, bank activity and each tax authority independently.
A platform withholding, scheduled payment, uploaded screenshot or generated packet must not automatically mark a return accepted or tax settled. Match authority, account, period, amount and authenticated evidence. Changes to reviewed facts create revisions and invalidate downstream approvals; they do not overwrite historic filings.
Keep credentials, identifiers, genuine transaction rows, source documents, private prompts and raw backups out of repositories and ordinary logs. Return only safe summaries and protected evidence references to implementation coordinators.
Acceptance and reporting
Run full checks at exact proposed heads on supported runtimes, independent source/ calculation examples, real service contracts, browser flows, egress/privacy tests, backup restore and rollback. Use the existing authorized deployment process; neither a plan nor a reference update grants permission to submit a return or pay tax.
Report separately: reference records updated; raw sources archived and reviewed; executable paths corrected; corpus staged; code installed; live behavior verified; protected records reconciled through a stated date; and filing readiness for the supported scope. Missing evidence stays unknown or blocked. No blanket 'all current' claim follows from matching JSON, passing helper tests or a successful health probe.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

