Agent task: integrate memory mesh with agent sessions, Holmes casefiles, and workspace evidence

[mdheller]

Agent-directed work package.

Target agent: Codex environment for SocioProphet/memory-mesh; Copilot coding agent if available.

Context:
memory-mesh should provide continuity semantics for governed agent/session/workspace memory: casefiles, session memories, evidence references, local-first memory sync, and later mesh/cloud-twin replication. It should consume agent-registry, prophet-workspace, Holmes, guardrail-fabric, and model-governance-ledger contracts without redefining them.

Scope:

1. Inspect the live repository before editing.
2. Add docs/GOVERNED_MEMORY_INTEGRATION.md describing memory scopes, ownership, privacy, revocation, evidence refs, mesh replication boundaries, and casefile/session integration.
3. Add schemas/memory-surface.v1.json covering memory scope, subject, retention, policy refs, agent/session refs, workspace refs, evidence refs, revocation refs, and sync topology.
4. Add example memory records for Holmes casefile memory, SourceOS local agent memory, and project-scoped memory.
5. Add repo.maturity.yaml using repo-maturity.v1 semantics.
6. Add validation for examples.

Acceptance criteria:

• Validation passes using the repository's Makefile or existing validation command.
• Docs clearly state memory-mesh does not own agent authority, model routing, or promotion decisions.
• Examples demonstrate privacy and revocation boundaries.
• No secrets, private content, credentials, or model weights are committed.

Boundary:
One PR only. Do not touch unrelated repositories.

[mdheller]

@codex Please take this memory-mesh issue. Implement the scoped governed memory integration in the issue body and open one PR with validation evidence.

[mdheller]

@copilot Please take this memory-mesh issue if available. Implement the scoped governed memory integration and open one PR with validation evidence.

[chatgpt-codex-connector]

Summary

• Added a new governed-memory integration design doc that defines memory scopes, authority boundaries, privacy, revocation, evidence linking, replication topology, and casefile/session integration, including explicit non-ownership of agent authority/model routing/promotion decisions by memory-mesh. docs/GOVERNED_MEMORY_INTEGRATION.mdL1-L86
• Added memory-surface.v1 schema covering scope, subject, retention, policy refs, agent/session/workspace refs, evidence refs, revocation, sync topology, and structured content payloads. schemas/memory-surface.v1.jsonL1-L130
• Added three governed memory examples for Holmes casefile memory, SourceOS local session memory (with revocation), and project-scoped memory to demonstrate privacy and revocation boundaries. examples/governed-memory/holmes-casefile-memory.example.jsonL1-L64 examples/governed-memory/sourceos-local-agent-memory.example.jsonL1-L56 examples/governed-memory/project-scoped-memory.example.jsonL1-L65
• Added repo.maturity.yaml using repo-maturity.v1 semantics with governance, contract, and validation evidence references. repo.maturity.yamlL1-L27
• Added governed-memory example validation script and wired it into repo validation, plus a unittest entry so it runs in CI/local validation flows. scripts/validate_governed_memory_examples.pyL1-L62 tests/test_governed_memory_examples.pyL1-L12 MakefileL9-L18

Testing

• ✅ make validate
  (Runs upstream lock validation, import plan rendering, Python compile checks, governed memory example validation, existing memoryd tests, new top-level tests, and deploy asset validation.)

View task →