fix: preserve goals across compression session rotation

[Umi4Life]

Context compression ends the active session and forks a child session (new session_id, linked via parent_session_id); /resume then redirects to that child. The goal/objective is keyed goal:<session_id> in state_meta and GoalManager does a flat lookup with no lineage fallback, so the objective read None on the child session after a long/compressed turn (and on resume).

Add migrate_goal(old, new) and call it at the rotation boundary in compress_context. Copy (not move), never overwrites an existing child goal, never raises — a migration failure can't abort the compression split. Flat GoalManager lookup and resume semantics are unchanged.

Tests: goal survives rotation + resume redirect, no-overwrite and no-source no-op invariants, and a focused test that compress_context actually calls migrate_goal during rotation.

Related to #15000: that issue covers compression-created child session chains for message history; this handles another session-scoped state affected by the same rotation.

Fixes #

Type of Change

• 🐛 Bug fix (non-breaking change that fixes an issue)
• ✨ New feature (non-breaking change that adds functionality)
• 🔒 Security fix
• 📝 Documentation update
• ✅ Tests (adding or improving test coverage)
• ♻️ Refactor (no behavior change)
• 🎯 New skill (bundled or hub)

Changes Made

• hermes_cli/goals.py — new module-level migrate_goal(old_session_id, new_session_id) helper beside save_goal/clear_goal. Copy semantics; returns True only when a goal is copied; skips if source absent, if old == new, or if the child already has a goal; swallows + logs errors (never raises).
• agent/conversation_compression.py — call migrate_goal(old_session_id, agent.session_id) at the session-rotation boundary in compress_context (right after the title is propagated to the child), guarded so a migration failure can't abort the compaction split.
• tests/hermes_cli/test_goals.py — TestGoalSurvivesCompressionRotation: goal survives rotation + resume redirect, no-overwrite invariant, no-source no-op invariant.
• tests/run_agent/test_compression_persistence.py — test_compression_rotation_migrates_goal: drives the real compress_context rotation path (summarizer stubbed, no model call) and asserts migrate_goal(old, new) is invoked and the goal resolves under the rotated id.

How to Test

1. Run the targeted suites: venv/bin/pytest tests/hermes_cli/test_goals.py tests/run_agent/test_compression_persistence.py → 60 passed.
2. Regression proof (objective loss): on main, test_goal_survives_session_rotation_and_resume_redirect fails at GoalManager(child).has_goal() (objective lost across compression rotation); with this fix it passes.
3. Guard proof (fix is load-bearing): temporarily neutralize the migrate_goal(...) call in compress_context → test_compression_rotation_migrates_goal fails (spy records no call); restore → passes.
4. Manual: set a /goal, force a long turn that triggers compaction (or /compress), then /resume — the objective is still active under the rotated session instead of reading None.

Checklist

Code

• I've read the Contributing Guide (https://github.com/NousResearch/hermes-agent/blob/main/CONTRIBUTING.md)
• My commit messages follow Conventional Commits (https://www.conventionalcommits.org/) (fix(scope):, feat(scope):, etc.)
• I searched for existing PRs (https://github.com/NousResearch/hermes-agent/pulls) to make sure this isn't a duplicate
• My PR contains only changes related to this fix/feature (no unrelated commits)
• I've run pytest tests/ -q and all tests pass
• I've added tests for my changes (required for bug fixes, strongly encouraged for features)
• I've tested on my platform:

Local platform: Debian GNU/Linux on x86_64, Python 3.11.15.

Documentation & Housekeeping

• I've updated relevant documentation (README, docs/, docstrings) — or N/A
• I've updated cli-config.yaml.example if I added/changed config keys — or N/A
• I've updated CONTRIBUTING.md or AGENTS.md if I changed architecture or workflows — or N/A
• I've considered cross-platform impact (Windows, macOS) per the compatibility guide (https://github.com/NousResearch/hermes-agent/blob/main/CONTRIBUTING.md#cross-platform-compatibility) — or N/A
• I've updated tool descriptions/schemas if I changed tool behavior — or N/A

[teknium1]

Thanks for this — the diagnosis and fix are correct for the rotation path.

Context on why we're closing it: as of #52658 (#38763), compression.in_place now defaults to True. Compaction keeps ONE durable session id for the conversation's whole life — it no longer ends the session and forks a continuation id. Because the id no longer rotates, the active /goal stays attached to the same session automatically — there is no rotation boundary to migrate the goal across.

Rotation now only runs as an explicit opt-out (compression.in_place: false), and the direction is to treat that path as legacy. So this fix hardens a code path that no longer executes by default and that we're not investing further in. Closing as superseded by the in-place default — not a reflection on the code, which was a correct fix for the behavior at the time. Credit preserved in the issue/PR history. Appreciate the contribution.