diff --git a/continue.md b/continue.md
index a1e26a7..ffce89b 100644
--- a/continue.md
+++ b/continue.md
@@ -9,6 +9,24 @@
   - Old configs in `old_tests/` use multiline inline tables (toml-c17 style) — keep for reference
   - Python's `tomllib` requires single-line inline tables
 
+## Sim Engine Capabilities
+### Implemented
+- Body propagation (elliptical + parabolic via Barker's equation)
+- Orbital↔Cartesian transforms (full z-x-z Euler rotation)
+- Velocity drift detection and element reconstruction
+- Global coordinate computation (hierarchical parent→child)
+- Spacecraft struct, loading, propagation
+- Impulsive burns (prograde, retrograde, normal, antinormal, radial_in, radial_out, custom)
+- TOML 1.0 config parsing
+### NOT Implemented (notify the user before beginning to refactor)
+- SOI transitions
+- Maneuver trigger system (TRIGGER_TIME, TRIGGER_TRUE_ANOMALY)
+- Hohmann transfer calculations
+- Rendezvous planning
+- OrbitTracker
+- Energy functions (KE, PE, total)
+- Hyperbolic propagation
+
 ## 1. Structure
 - One `SCENARIO("description")` per logical test group, with `[tag1][tag2]` annotations
 - Shared fixture: all constants, structs, and variables declared between `SCENARIO` opening and first `SECTION`
@@ -28,19 +46,27 @@
 
 ## 4. Precalc scripts
 - For each test file, create `scripts/precalc_<test_name>.py` that computes expected values.
-- Use `scripts/sim_engine.py` for physics (orbital_to_cartesian, propagate, etc.).
+- **Check the capability matrix first** — only use sim_engine.py for implemented features.
+- **Always output local-frame values** (distances from parent, not global from origin).
+  - C++ tests typically use local coordinates (e.g., `vec3_distance(craft->local_position, (Vec3){0,0,0})`).
+  - Global distances are dominated by parent body positions (e.g., Earth-Sun distance swamps LEO orbit).
 - Output C++-style comments with precalculated constants for embedding in the test.
 - Run with: `python3 scripts/precalc_<test_name>.py`
-- If sim_engine.py lacks a feature (e.g., spacecraft, OrbitTracker), use analytical formulas instead (**but notify the user what feature was missing**)
+- If sim_engine.py lacks a feature, use analytical formulas instead (**but notify the user what feature was missing**)
 
 ## 5. Process per file
+- **Pre-check**: Verify the test file has a TOML config in `old_tests/`. If it doesn't, skip — it's likely hardcoded.
+- **Pre-check**: Check if the test is already in `tests/` (already refactored). Skip if so.
+- **Pre-check**: Check the capability matrix — if the test needs SOI, maneuvers, rendezvous, etc., flag this before starting.
 - Process **one test file at a time**.
 - Create `scripts/precalc_<test_name>.py` and run it to get expected values.
 - Copy from `old_tests/` to `tests/`, rewrite using the pattern from `test_true_anomaly_roundtrip.cpp`.
 - Rewrite TOML configs to TOML 1.0 inline table syntax (single-line `{}`).
 - Build and verify: `make test-build` then `./build/orbit_test '[tag]' -s`.
 - Run full suite: `make test`.
-- Replace broken tests (e.g., OrbitTracker bugs) with working alternatives using propagation checks.
+- **Verify no other tests broke** — the new test shouldn't affect unrelated tests.
+- Note any sim_engine.py features that were missing for this test (add to capability matrix if needed).
+- notify the user about any broken tests (e.g., OrbitTracker bugs)
 - **Always ask for review** before moving to the next file.
 - **Only commit when asked.**