Perenna v0.2.0

Automated release for Perenna v0.2.0.

Upgrade notes

Perenna 0.2.0 removes the process-level source label from the committed
memory format, MCP results, startup configuration, and generated plugin
configuration.

Before upgrading a repository created by an earlier version:

1. Stop or avoid concurrent Perenna mutations.
2. Remove source from every memory in the current Git tree.
3. Validate the five-field memory format and commit that exact conversion.
4. Upgrade Perenna and restart connected clients.
5. Read every affected memory again before patching, replacing, or deleting
   it because the conversion changes its revision.

Remove the --source argument and PERENNA_SOURCE environment variable from
startup configuration. Perenna 0.2.0 rejects the earlier six-field memory
format and no longer accepts either setting.

Other changes include action-specific MCP input schemas, stage-specific index
recovery guidance, memory consolidation guidance in the bundled Skill, and a
documented Glama-to-ChatGPT connection workflow.

Changelog

Changes since v0.1.3:

• refactor: split MCP schemas, deploy keys, and CLI output into modules (#11) (144a467)
• docs(skill): guide memory consolidation (#12) (15fbc79)
• fix!: align memory and index contracts (#14) (a7d2ad3)
• docs: document hosted connectivity workflows (#13) (f0e29cf)
• chore: release 0.2.0 (6c68271)

Full diff