What does this PR do?

This PR fixes a context compression bug where Pass 3 could corrupt tool_calls.function.arguments by raw string truncation (args[:200] + "...[truncated]"), producing invalid JSON and causing downstream json.loads(arguments) failures (for example Unterminated string).

The fix keeps Pass 3, but makes it JSON-safe:

• Parse arguments as JSON before modifying.
• Shrink large payloads structurally (recursive trim of long strings/lists/dicts).
• Re-serialize with json.dumps(...) so output stays valid JSON.
• Preserve pre-existing invalid argument strings unchanged (no mutation).

This approach preserves the intent of Pass 3 (reduce oversized args) without violating the tool-call JSON contract.

Related Issue

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

• Updated agent/context_compressor.py:
  • Added _shrink_json_for_tool_args(...) recursive helper.
  • Replaced raw-truncate Pass 3 logic with JSON-safe parse/shrink/serialize flow.
  • Preserved invalid pre-existing arguments verbatim (skip mutation on parse failure).
  • Updated method docstring to reflect JSON-safe argument shrinking behavior.
• Updated tests/agent/test_context_compressor.py:
  • Added regression test ensuring large args are shrunk but remain valid JSON.
  • Added regression test ensuring malformed existing args are not modified.

How to Test

1. Reproduce old behavior (before fix): run a long session with large tool arguments and compaction; observe JSON parse failures on truncated args.
2. Run tests:
   • pytest -q tests/agent/test_context_compressor.py
3. Verify fixed behavior:
   • Large tool_calls.function.arguments are reduced in size.
   • json.loads(arguments) succeeds for shrunk payloads.
   • Already-invalid argument strings are unchanged after pruning.

Checklist

Code

• I've read the Contributing Guide
• My commit messages follow Conventional Commits (fix(scope):, feat(scope):, etc.)
• I searched for existing PRs 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: macOS

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 — or N/A
• I've updated tool descriptions/schemas if I changed tool behavior — or N/A

Screenshots / Logs

• N/A (logic + tests change; no UI changes).