review/odysseus workspace fixes

[KaikiDaemon]

• fix(skills): validate serialized list fields
• fix(mcp): bound connections and redact errors
• fix(io): harden atomic file replacement
• fix(tools): make agent file writes atomic
• fix(runtime): update compatibility and resource lifecycles
• test: close resources and enforce clean async tests
• fix(tools): stop batches at user prompts
• fix(research): persist generated report links
• fix(mcp): close replaced server connections
• fix(mcp): cancel pending connections on shutdown
• fix(mcp): clear completed HTTP connection tasks
• fix(mcp): validate route input payloads

[github-actions]

⚠️ PR description — action needed

The following required sections are missing or incomplete. Please update the PR description to address them:

• Summary is empty or too short — describe what changed and why.
• Linked Issue — add a reference like Fixes #NNN, a bare #NNN, or a link to the issue.
• Type of Change — check at least one box.
• Checklist — check the duplicate-search box to confirm you searched existing issues and PRs.
• How to Test — explain how a reviewer can verify this change. Numbered steps, the commands you ran, or a short code block all work — give a sentence or two of real detail (not just "tested locally").

---

This comment is deleted automatically once all sections are complete.

[RaresKeY]

Thanks for the work here. I am closing this version because it is an umbrella workspace-fixes branch rather than one reviewable change.

The PR contains 12 separate commits across 39 files, covering skills serialization, MCP connection limits and lifecycle cleanup, error redaction, route validation, atomic file replacement, agent file writes, tool-batch behavior, runtime resource cleanup, research-link persistence, and broad test changes. These are independent concerns that should be reviewed, tested, and reverted separately.

The branch also conflicts with dev, fails the required title and description checks, and has no linked issue or reproducible test plan tying the changes to one defined problem.

Some individual fixes may still be useful. If they remain relevant, please bring them back as fresh, focused PRs based on current dev, with one coherent problem per PR and its corresponding regression tests. Please do not reuse this bundled branch.