Files
memory/patterns/skill-impact.md
T

5.8 KiB

Skill Impact Tracker

Inspired by WikiSkill (arXiv:2608.27454) — tracks skill modifications, their validation outcomes, and acceptance decisions. Prevents repeating failed modifications and provides an audit trail of skill evolution.

How to Use

When a skill is patched, created, or deleted, append an entry to the table below. The entry records WHAT changed, WHY, and WHETHER it helped (if validated).

Entry Format

| Date | Skill | Change Type | Description | Validation | Outcome | Ref |

| 2026-10-08 | smart-home-automation | adopted | EMHASS-References erweitert: Status-Block in defect-analysis (Pointer-Datei) + neue Sektion "CSV Laws & Shadow Testing" in api-and-config | tested | accepted | docs/solutions/workflows/2026-10-08-emhass-phase2-shadow-methodology.md |

  • Change Type: created | patched | deleted | adopted
  • Validation: How was success measured? (manual | tested | benchmark | n/a)
  • Outcome: accepted | rejected | rolled-back | pending
  • Ref: Session ID or solution doc path

Audit Trail

Date Skill Change Type Description Validation Outcome Ref
2026-08-30 compound-learning patched Added WikiMaintainer phase + patterns/ guidance from WikiSkill paper analysis tested accepted this session
2026-08-30 (retrospective) created 4 solution docs + 2 new patterns (PAT-008, PAT-009) from 30-day retrospective tested accepted this session
2026-08-13 hindsight patched Expanded meta-memory automation table (decay scoring, drift detection, skill health) tested accepted session 2026-08-13
2026-08-13 (3 cron jobs) created Decay scoring, drift detection, skill health check — no_agent cron jobs tested accepted session 2026-08-13

Notes

  • Rejected proposals persist here — they are NOT deleted. Future skill updates can consult this log to avoid repeating failed approaches (the WikiSkill paper showed this is critical for effective evolution).
  • Neutral changes (no improvement, no regression) are recorded as accepted with validation: manual — they may enable future improvements.
  • Rollbacks are explicitly tracked — if a skill patch caused regressions and was reverted, the entry stays with outcome: rolled-back. | 2026-10-08 | smart-home | patched | NUT-Absatz ergänzt: Kaskade, Master/slaves, roh-Proto-Details, Watchdog-Cron 4cbf03ccb4de | tested | accepted | docs/solutions/bug-fixes/2026-10-07-nut-ups-ms-a2-1-deployment.md | | 2026-10-08 | (none) | adopted | PAT-016 systemd-path-blindness erstellt (Exit-127-Klasse: ~/.local/bin unsichtbar im Service-PATH; trap-EXIT-Cleanup-Pflicht) | tested | accepted | docs/solutions/bug-fixes/2026-10-08-hermes-offsite-backup-aws-path-rca.md | | 2026-10-11 | devops/seafile-api | patched | references/seafile-bulk-migration.md um Welle-2-Guardrails erweitert: synchronized-as-Gate (Trim-Parser), Status-Tabellen-Truncation unter Last, Heal-Pass, --from0-Exclude-Engineering, Byte-Verification statt du, mv-across-ZFS-Datasets, pkill-in-eigener-Shell-Falle, Locale-abhängige rsync-stats; Disk-Stand 285G/46G aktualisiert | tested | accepted | docs/solutions/workflows/2026-10-11-seafcli-wave2-status-parsing-prefill-race-orchestration.md | | 2026-10-17 | productivity/imap-file-delivery | patched | Pitfall-Tabelle +2: verschachtelte FETCH-Tupel (rekursiv flattening statt dat[0].decode) und RFC2047-q-kodierte Subjects (decode_header+Klartext-Vergleich statt Rohmatch) — sonst FALSE NEGATIVE trotz APPENDUID-OK | tested | accepted | docs/solutions/integrations/2026-10-17-imap-append-verification-rfc2047-tuple-pitfalls.md | | 2026-10-19 | devops/rke2-cluster-administration | patched | Neue Reference references/cilium-networkpolicy-traps-2026-10-19.md (4 stumme CNP-Fallen: Selektion⇒Default-Deny, [{}]=Null-Allow, service.name≠Identity-Label, VIP-SNAT⇒world) + Description-Link; L3 deckte Cilium-NP zuvor nicht ab (nur LB-Pool) | tested | accepted | docs/solutions/bug-fixes/2026-10-19-cilium-networkpolicy-four-traps.md | | 2026-11-02 | devops/ai-noris-inference-api | patched | Context-Fenster-Pitfall ergänzt: Bifrost liefert max_context_length verschachtelt; Hermes-Fallback 128k/1.31M/256k; Fix via model.context_length + providers.noris.models..context_length | tested | accepted | docs/solutions/bug-fixes/2026-11-02-hermes-context-length-bifrost-nested.md | | 2026-11-05 | security/security-exposure-audit | patched | Auth-Bypass-Regel ergänzt: sshd-libc per ldd prüfen (musl-Loader ≠ musl-sshd), fünf AND-Bedingungen für sshhole.com | tested | accepted | docs/solutions/bug-fixes/2026-10-30-cve-2026-98432-fleet-exposure.md | | 2026-11-05 | cronjob | patched | Pitfall ergänzt: monitor_script braucht bloßen Dateinamen (+x, Shebang); bash /abs/pfad macht den Monitor blind; Agent-Prompt-SSH-Zugang vor Scharfstellen live testen | tested | accepted | docs/solutions/bug-fixes/2026-10-30-cve-2026-98432-fleet-exposure.md | | 2026-11-15 | devops/network-reconnaissance | patched | Pitfall ergänzt: zip(ips, as_completed(futures)) misattribuiert Alive-IPs (77/91 Ghosts im Weekly-Scan); Korrektur: pool.map oder {fut: ip}-Dict; Cross-Check gegen ip-map | tested | accepted | docs/solutions/bug-fixes/2026-11-15-network-scan-ip-attribution-bug.md | | 2026-11-15 | openscad-functional-parts | patched | Pitfalls +2: (1) OpenSCAD 2021.01 CLI-Quirks — kein --render, -o relativ zur .scad-Datei, STL default ASCII (--export-format binstl), color() im PNG-Export ignoriert, Framing via --viewall --autocenter; (2) Regel «Passungskonus in fester Außenkontur ⇒ Wand geht gegen 0», Wand an beiden Enden rechnen | tested | accepted | docs/solutions/workflows/2026-11-15-konus-passung-wandstaerke-vorab-pruefen.md |