Add personal-data skills; document ETL/infra procedures

[oinkspook]

Pull Request Description

Type of Change

• Bug fix
• New feature
• Enhancement
• Documentation update
• Other (please specify)

Description of Changes

Codifies procedures that were documented (in corneel_notes.txt and the instroom/1cijferho implementations) but not yet captured as skills. All content is grounded in real, verified code — not planning notes.

New knowledge skills

• browser-pseudonymize — in-browser Pyodide/WebAssembly encryption of sensitive columns (BSN/PGN/onderwijsnummer) before upload, in a Streamlit app. Captures the encrypt-vs-pseudonymize distinction, the "never keyless-hash a BSN" rule, the Pyodide bootstrap/caching, base64 JS-payload injection, and the st_js_blocking session-state latching gotcha. Grounded in 1cijferho/src/frontend/Files/.
• identifier-mapping — split a record into a derived UUID + dual Postgres tables (secret_sensitive/secret_regular, FK cascade) + a MinIO object. Server-side keyed-HMAC UUID derivation; three read-back resolution levels. Grounded in 1cijferho/src/eencijferho/io/personal_data.py.

Extended skills

• surf-sdp-helm-flux — new "Batch workloads" section: scheduled CronJob + on-demand Job, the unique-name-per-release mechanism, image-by-digest + config/secret wiring. Flags field bugs in the reference instroom chart.
• etl-pipeline — cross-pipeline trigger-token setup (INSTROOM_CONFIG_TRIGGER_TOKEN).
• surfdrive — public-share creation (where SURFDRIVE_SHARE_TOKEN/PASSWORD come from).

Related Issues

None.

Comparison: Before and After

Before:

Personal-data handling (browser encryption, UUID/dual-store) and the batch-job / trigger-token / surfdrive-link procedures lived only in code and corneel_notes.txt; no skill surfaced them.

After:

Two new skills + three extended skills codify these patterns, with the security-sensitive parts (encrypt vs pseudonymize, keyed server-side hashing) made explicit.

Testing Instructions

Skills are markdown (no executable change). To validate:

1. Confirm each SKILL.md frontmatter description triggers on its intended prompts.
2. Cross-references (/browser-pseudonymize, /identifier-mapping, /sdp-secrets-management, /etl-pipeline) resolve to existing skills.

Validation

• styler::style_active_file() has been run — N/A: no R code

Dependencies

None.

Additional Information

• identifier-mapping explicitly marks the restricted DB user, stored procedures, and INT-9/unique constraints as design intent, not implemented — it does not claim code that is absent.
• Known chart bugs in instroom-config (duplicate successfulJobsHistoryLimit, misused ttlSecondsAfterFinished, CronJob-only keys on a Job) are documented in the skill but not fixed here — that is a separate change in the instroom-config repo.
• Relates to the merge backlog noted in docs/skill-gaps.md: please review promptly rather than letting it queue.

Checklist

• I have tested these changes locally
• I have updated the documentation accordingly
• My code follows the project's coding standards
• I have added/updated tests to cover my changes (if applicable) — N/A: docs-only
• The code is accompanied by comments
• I have linked this PR to relevant issues (if applicable)

🤖 Generated with Claude Code