Pre-screen analysis outputs (tables, figures, logs) built on restricted or confidential data for statistical-disclosure-limitation problems before any release. Scans for small cell counts, complementary-suppression gaps, dominance (p-percent / (n,k)), re-identifiable exact counts, PII leakage, and unrounded sensitive statistics; classifies each finding CRITICAL / WARNING / OK and gates on any CRITICAL. Use before depositing or sharing restricted-data results, or when the user says "disclosure check", "SDL scan", "is this output safe to release", "check for small cells", "disclosure avoidance", "pre-screen for the RDC", or "can I export this from the enclave".
/disclosure-check — Statistical-Disclosure-Limitation pre-screenScan analysis outputs built on restricted or confidential data (Census FSRDC, IRS SOI, administrative registers, linked health records, proprietary firm panels) for the disclosure-avoidance problems that get an export request rejected — before it reaches the data provider's official disclosure review. The skill is a pre-screen, not a substitute for that review.
Core principle: A single un-suppressed n=3 cell, an exact count that pins down one firm, or a p-percent dominance failure can re-identify a person or establishment. Catch it on your machine, not in the rejection email from the RDC analyst.
$0 — outputs directory to scan. Defaults to output/, where every language's pipeline writes; if it does not exist but a pre-v2.6 scripts/<lang>/_outputs/ does, scan that and say so. Also accepts any export-staging directory (e.g., a to_review/ folder the analyst stages for the RDC).--provider — selects which disclosure-rule profile to load (Phase 0). One of census / irs / irb / generic. Providers differ — thresholds and rules are not interchangeable; default generic is deliberately conservative.--threshold N — override the minimum cell count (default n<10). Census FSRDC commonly uses 10 for establishments; IRS and many IRBs differ. Always reconcile with your provider's written rules..claude/rules/confidential-data.md for the project's restricted-data handling contract and the rule-profile placeholder.--provider profile (a placeholder config the forker fills in from their signed agreement — Census, IRS, and IRB rules differ and supersede any default here):
n<10),p-percent (a cell is unsafe if the largest respondents contribute > p% of the total) and (n,k) (top n units > k% of total),generic profile and flag prominently in the report that real provider thresholds must be substituted.Glob the outputs dir for .tex, .csv, .txt, .log, .smcl, .out, .md tables and figure-data files. For each:
0 < n < threshold that is not already suppressed.p-percent and (n,k) rules.\d{3}-\d{2}-\d{4}), exact dates of birth, addresses, exact lat/long or fine geocodes, record IDs that survived into an output.| Disposition | Meaning | Examples |
|---|---|---|
| CRITICAL | Would fail the provider's disclosure review; blocks release. | n=3 cell un-suppressed; complementary-suppression hole; p-percent dominance failure; any PII; an exact count identifying ≤2 units. |
| WARNING | Plausibly safe but needs a human judgment call. | Cell at exactly the threshold; unrounded total just over a rounding base; geographic statistic near the min-population floor. |
| OK | Within the loaded rules, no action needed. | Counts ≥ threshold and rounded; dominance passes; no PII. |
When two findings interact (a suppressed cell + a recoverable margin), report them together — the gate cares about the joint disclosure risk, not each cell in isolation. Be economics-aware: DiD / event-study cell counts per (cohort × period), IV first-stage subsamples, RCT arm × stratum balance tables, and panel firm-counts are the usual offenders.
For each CRITICAL / WARNING, propose the standard SDL fix, in order of preference:
Each suggestion names the file, the cell/location, the rule it violates, and the concrete edit — never auto-applies it (the analyst owns the disclosure decision).
Exit non-zero on any CRITICAL. WARNINGs surface but do not block. See Exit behavior.
Write quality_reports/disclosure_check_[outputs-dir-slug].md:
# Disclosure Check: [outputs dir]
**Date:** [YYYY-MM-DD]
**Provider profile:** census | irs | irb | generic (rules source: confidential-data.md)
**Min cell count:** [N] **Dominance:** p=[p]%, (n,k)=([n],[k]%) **Rounding base:** [b]
## Summary
| Disposition | Count |
|---|---|
| CRITICAL | M |
| WARNING | W |
| OK | P |
| **Verdict** | **PASS / FAIL** (FAIL iff M > 0) |
## CRITICAL (blocks release)
| File | Location | Rule violated | Observed | Suggested remediation |
|---|---|---|---|---|
| tab3_by_cohort.tex | row "2008", col "n" | min cell (n<10) | n=4 | suppress cell + suppress complement in margin |
## WARNING (human judgment)
| File | Location | Concern | Suggested action |
|---|---|---|---|
## OK
[counts only, or a short list]
## Next steps
1. Resolve every CRITICAL — suppress / round / top-code / aggregate, then re-run.
2. Review WARNINGs with the agreement's written rules in hand.
3. Re-run until zero CRITICAL, THEN submit to the provider's OFFICIAL disclosure review./audit-reproducibility's gate semantics: WARNING ≠ FAIL, only CRITICAL blocks.--provider <name> — Load that data provider's disclosure rules (e.g. census, irs, irb). Default: the generic small-cell ruleset.--threshold <n> — Override the minimum cell-count threshold (default n<10); match your data-use agreement's actual rule..claude/rules/confidential-data.md — restricted-data handling contract + the provider-rule profiles this skill loads..claude/rules/replication-protocol.md — for restricted-data papers the replication package ships code + access path, not the microdata; screen every released output first..claude/skills/audit-reproducibility/SKILL.md — numeric paper↔code verification: run it on the retained values, this skill on the released ones..claude/skills/data-analysis/SKILL.md, .claude/skills/stata-replication/SKILL.md — produce the R / Stata / Python outputs this skill screens.--provider profile is wrong, the scan is wrong. Reconcile with the written agreement, not a default.pedrohcgs/claude-code-my-workflow
fe70774
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.