diff --git a/skills/test-driven-development/SKILL.md b/skills/test-driven-development/SKILL.md
index 4320d8879a..211d3a48f7 100644
--- a/skills/test-driven-development/SKILL.md
+++ b/skills/test-driven-development/SKILL.md
@@ -119,13 +119,17 @@ npm test path/to/test.test.ts
 ```
 
 Confirm:
-- Test fails (not errors)
+- Test fails (not errors) — a **behavior assertion** rejects wrong output
 - Failure message is expected
-- Fails because feature missing (not typos)
+- Fails because the feature's behavior is wrong or missing (not typos)
 
 **Test passes?** You're testing existing behavior. Fix test.
 
-**Test errors?** Fix error, re-run until it fails correctly.
+**Test errors?** (import/module-not-found, missing symbol, syntax, collection,
+setup) Fix the error and re-run until the test fails correctly. Structural
+load failures only prove the surface is not runnable — they are **not**
+evidence that the behavior assertion works. Get an importable negative
+control under the assertion before treating RED as proven.
 
 ### GREEN - Minimal Code
 
diff --git a/skills/writing-plans/SKILL.md b/skills/writing-plans/SKILL.md
index dd2702b8ba..4958812993 100644
--- a/skills/writing-plans/SKILL.md
+++ b/skills/writing-plans/SKILL.md
@@ -103,7 +103,22 @@ def test_specific_behavior():
 - [ ] **Step 2: Run test to verify it fails**
 
 Run: `pytest tests/path/test.py::test_name -v`
-Expected: FAIL with "function not defined"
+Expected: FAIL with a **behavior assertion** rejecting the wrong result
+(e.g. `AssertionError: assert actual == expected`), not a load/import error.
+
+**RED evidence (required for claiming the test works):**
+- **Behavioral RED** — the test runs far enough to execute its assertion and
+  fails because the observable result is wrong. This is the only evidence that
+  the behavior check has teeth.
+- **Structural RED** — `ImportError` / `ModuleNotFoundError` / missing symbol /
+  syntax / collection / setup error. This only proves the surface is absent or
+  not loadable. **Do not** plan or report structural RED as proof that a
+  behavior assertion detects incorrect results.
+- If Step 2 hits structural failure first: get past load/setup (minimal
+  importable stub or negative-control return value as needed), re-run, and
+  record behavioral RED before implementing the real logic — unless the plan
+  explicitly states why behavioral RED is impossible for this technology.
+  Aligns with `test-driven-development` (fails, not errors).
 
 - [ ] **Step 3: Write minimal implementation**