Skip to content

Fix dataclaw issue #16 — prevent automatic deletion of old sessions, create PR #211

@xliry

Description

@xliry

Fork repo: peteromallet/dataclaw
Workspace: ~/dataclaw (needs to be cloned first: gh repo clone peteromallet/dataclaw ~/dataclaw)
Create a PR to the upstream repo (peteromallet/dataclaw).


Issue #16 — Prevent automatic deletion of old sessions

Summary: Some agentic coding tools (Claude Code, Gemini CLI) automatically clean up sessions older than 30 days. This causes data loss for dataclaw which needs to preserve session data.

Issue body:
Some agentic coding tools may automatically clean up sessions older than 30 days. I've confirmed Claude Code and Gemini CLI do so by default. We may tell users to configure these tools not to do so, which helps preserve more data.

Or we may incrementally collect data in dataclaw and make a cron job to periodically do so. But for now I think incremental processing is not strictly needed for data < 100 MB.

Comment from @wjessup (2026-03-01):

Another solution is a proxy to catch and log original sessions.
I could add this into: https://github.com/wjessup/context-compactor


Approach

  • Read the existing codebase to understand how dataclaw collects and stores session data
  • Implement a solution to prevent/mitigate automatic deletion (e.g., incremental collection via cron, documentation for tool config, or backup mechanism)
  • Consider the proxy approach mentioned in comments as an alternative
  • Keep it simple — no over-engineering for < 100 MB data

Deliverables

  1. Clone repo if not present
  2. Create a branch (e.g. fix/issue-16-prevent-session-deletion)
  3. Implement the fix
  4. Open PR to peteromallet/dataclaw referencing the issue (Fixes Remove duplicate API docs from CLAUDE.md — MCP description already covers it #16)

Metadata

Metadata

Assignees

No one assigned

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions