session-archivist
AdvisoryAudited by Static analysis on May 7, 2026.
Overview
No suspicious patterns detected.
Findings (0)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Private conversation details, decisions, file references, and todos may be retained in archives or memory and reused later.
The skill reads Hermes session histories and persists structured summaries locally, with optional storage into Hindsight memory for later cross-session recall.
遍历 ~/.hermes/sessions/*.json ... 本地 markdown 存档（始终） ... Hindsight 存档（如果可用）
Run dry-run first, review the archive directory, and set hindsight_enabled to false if you do not want summaries stored in Hindsight memory.
If the summary misses important details or a session is trimmed unexpectedly, older conversation content may be harder to recover after backups expire.
The workflow intentionally rewrites and reduces session files after archiving, with backups retained for a limited period.
5. 裁剪阶段 ... 保留最近 N 条消息 ... 插入摘要上下文 ... 写回 session 文件 ... 原始 session 备份在 `~/.hermes/session-archives/backups/`，保留 7 天。
Use --dry-run and --list before real runs, confirm backups are being created, and avoid aggressive size thresholds until you trust the summaries.
After setup, session archiving and trimming may happen without a manual command each time.
The setup script can create a persistent scheduled job that automatically runs the archival workflow daily.
hermes cron create "0 3 * * *" --name "session-archivist-daily" --prompt "Run session archivist archival: python3 $ARCHIVER --max-size 1024"
Only run setup_cron.sh if you want automatic daily processing; otherwise run the archiver manually and review existing jobs with hermes cron list.