diff --git a/ci/measure_baseline_205.py b/ci/measure_baseline_205.py
index 458282ea..41a64ee4 100644
--- a/ci/measure_baseline_205.py
+++ b/ci/measure_baseline_205.py
@@ -57,7 +57,10 @@
 # in the fixture project. teensyLC, teensy30, teensy41, stm32f103c8 all ship a
 # Blink-class sketch as ``src/main.ino``.
 TARGETS = [
-    ("teensyLC", "tests/platform/teensylc", "Blink"),
+    # WHY: env names must match the [env:...] keys in each fixture's
+    # platformio.ini exactly (case-sensitive). The teensy LC env is
+    # `teensylc` lowercase per tests/platform/teensylc/platformio.ini.
+    ("teensylc", "tests/platform/teensylc", "Blink"),
     ("teensy30", "tests/platform/teensy30", "Blink"),
     ("teensy41", "tests/platform/teensy41", "Blink"),
     ("stm32f103c8", "tests/platform/stm32f103c8", "Blink"),
diff --git a/tasks/baseline-205.md b/tasks/baseline-205.md
index 4d37312a..eb9e9639 100644
--- a/tasks/baseline-205.md
+++ b/tasks/baseline-205.md
@@ -1,68 +1,82 @@
-# Baseline measurements for #205 — DEFERRED
+# Baseline measurements for #205
 
-Captured: 2026-04-24
-Git SHA: (this PR's foundation commit — see PR description)
-Branch: main
-Tooling: `uv run python ci/measure_baseline_205.py`
+Captured: 2026-05-10T15:14:08Z
+Git SHA: fffb1b4cc6389942bf63cc2ae0f01e65484e10fe
+Branch: chore/fix-baseline-205-teensylc-env
+Tooling: cargo 1.94.1 (29ea6fb6a 2026-03-24), size tool: arm-none-eabi-size.exe
 
-## Status
+Generated by `uv run python ci/measure_baseline_205.py`. See module docstring for methodology.
 
-The capture script (`ci/measure_baseline_205.py`) is implemented and
-runnable. The actual data capture is **deferred to a follow-up step**
-because Teensy/STM32 builds against the foundation-landed resolver are
-heavyweight (multi-minute per board on a cold cache) and the build
-infrastructure on the development workstation could not complete all
-four boards within the agent run-window. Running the script in a clean
-CI environment with all toolchains pre-warmed will populate the table
-below.
+## teensylc / Blink
 
-## How to run
+- Project: `tests/platform/teensylc`
+- ELF: `tests/platform/teensylc/.fbuild/build/teensylc/release/firmware.elf`
+- TU count: 68
+- .text: 6,980 bytes
+- .data: 380 bytes
+- .bss: 1,068 bytes
+- .dmabuffers: 192 bytes
+- Excluded library hits in compile_commands.json:
+  - FNET: 68 entries
+  - Snooze: 68 entries
+  - RadioHead: 68 entries
+  - mbedtls: not present
 
-```bash
-uv run python ci/measure_baseline_205.py --out tasks/baseline-205.md
-uv run python ci/measure_baseline_205.py --targets teensyLC teensy41
-```
+## teensy30 / Blink
 
-The script:
+- Project: `tests/platform/teensy30`
+- ELF: `tests/platform/teensy30/.fbuild/build/teensy30/release/firmware.elf`
+- TU count: 68
+- .text: 6,684 bytes
+- .data: 272 bytes
+- .bss: 1,120 bytes
+- .dmabuffers: 248 bytes
+- Excluded library hits in compile_commands.json:
+  - FNET: 68 entries
+  - Snooze: 68 entries
+  - RadioHead: 68 entries
+  - mbedtls: not present
 
-1. Builds `tests/platform/<board>` for `teensyLC`, `teensy30`,
-   `teensy41`, `stm32f103c8` via the existing `fbuild` CLI.
-2. Counts distinct `file` entries in the resulting
-   `compile_commands.json` (TU count).
-3. Probes `firmware.elf` section sizes (`.text`, `.data`, `.bss`,
-   `.dmabuffers`) via `arm-none-eabi-size` (preferred) or `llvm-size`.
-4. Scans `compile_commands.json` for `FNET` / `Snooze` / `RadioHead`
-   / `mbedtls` references — the four libraries that were wrongly
-   selected before the foundation phases of #205 landed.
+## teensy41 / Blink
 
-## Expected once captured
+- Project: `tests/platform/teensy41`
+- ELF: `tests/platform/teensy41/.fbuild/build/teensy41/release/firmware.elf`
+- TU count: 85
+- .text: section absent or size tool unavailable
+- .data: 3,776 bytes
+- .bss: 1,664 bytes
+- .dmabuffers: section absent or size tool unavailable
+- Excluded library hits in compile_commands.json:
+  - FNET: 85 entries
+  - Snooze: 85 entries
+  - RadioHead: 85 entries
+  - mbedtls: not present
 
-| env          | TU count | .text | .data | .bss | .dmabuffers | excluded libs |
-|--------------|----------|-------|-------|------|-------------|----------------|
-| teensyLC     | (≤ 250 per AC#1) | … | … | (≤ 3 KB per AC#1) | — | none of FNET/Snooze/RadioHead/mbedtls present |
-| teensy30     | … | … | … | … | (≤ 1 KB per AC#2) | none |
-| teensy41     | … | … | … | … | … | (regression baseline) |
-| stm32f103c8  | … | … | … | … | — | (must include SPI per AC#4) |
+## stm32f103c8 / Blink
 
-## Why not just ship the placeholder and call it done
+- Project: `tests/platform/stm32f103c8`
+- ELF: `tests/platform/stm32f103c8/.fbuild/build/stm32f103c8/release/firmware.elf`
+- TU count: 201
+- .text: 12,192 bytes
+- .data: 240 bytes
+- .bss: 1,012 bytes
+- Excluded library hits in compile_commands.json:
+  - FNET: not present
+  - Snooze: not present
+  - RadioHead: not present
+  - mbedtls: not present
 
-Phase 6 (acceptance gates) needs *real* numbers to anchor the
-"+1%" / "≤ 250" / "≤ 3 KB" thresholds in the issue body. A guess will
-be argued about during Phase 6 reviews. The capture has to happen on a
-host that can actually link these four ELFs, which means either (a) a
-clean CI runner with the Teensy/STM32 toolchains pinned, or (b) a
-warmed local install where every framework download has already
-completed. Neither was ready inside this PR's window.
+## Build status
 
-## Tracking
-
-This file is replaced wholesale on the next successful run of
-`measure_baseline_205.py`. The non-empty rows above will be filled in
-with measured numbers + an ISO timestamp + the exact git SHA the
-measurement was taken against.
+| env | build | TU count | size tool | notes |
+|---|---|---|---|---|
+| teensylc | ok | 68 | arm-none-eabi-size.exe |  |
+| teensy30 | ok | 68 | arm-none-eabi-size.exe |  |
+| teensy41 | ok | 85 | arm-none-eabi-size.exe |  |
+| stm32f103c8 | ok | 201 | arm-none-eabi-size.exe |  |
 
 ## Run command
 
-```bash
+```
 uv run python ci/measure_baseline_205.py --out tasks/baseline-205.md
 ```