Add cross-runtime ESP32-S3 HIL benchmark - #40

Merged

w1ne merged 48 commits into

mainfrom

bench/twin2silicon-deepseek

Aug 16, 2026

Commits on Aug 13, 2026

test: add first Twin2Silicon repair fixture
w1ne
committed
564bbba

Commits on Aug 14, 2026

fix(agent): request canonical hosted tool names
w1ne
committed
06d3db6
test(agent): resolve canonical tools through public launcher
w1ne
committed
83f0622

Commits on Aug 15, 2026

docs: design ESP32-S3 HIL benchmark
w1ne
committed
4cbf6eb
docs: plan ESP32-S3 HIL benchmark
w1ne
committed
d0ee2ca
feat(bench): add bounded HIL result primitives
w1ne
committed
ee630fa
fix(bench): tighten HIL result contracts
w1ne
committed
ef5cbd7
fix(bench): reap timed-out process groups
w1ne
committed
5c2619e
fix(bench): bound timed-out process reaping
w1ne
committed
78c34f1
fix(bench): report lingering process groups
w1ne
committed
970e7f2
test(bench): add ESP32-S3 GPIO repair fixture
w1ne
committed
a071439
fix(bench): correct ESP-IDF console symbol
w1ne
committed
cf8932b
feat(bench): evaluate ESP32-S3 UART and JTAG evidence
w1ne
committed
3f2d0ab
fix(bench): run bounded ESP32-S3 register checks
w1ne
committed
03cb978
fix(bench): accept Espressif register evidence
w1ne
committed
e8be3a1
feat(bench): orchestrate reproducible ESP32-S3 HIL runs
w1ne
committed
1db308f
fix(bench): harden HIL orchestration evidence
w1ne
committed
5964eb9
fix(bench): synchronize UART readiness before flash
w1ne
committed
98e5df1
fix(bench): redact and account HIL run evidence
w1ne
committed
ccc8ca0
fix(bench): isolate UART capture lifecycle
w1ne
committed
bed71ba
fix(bench): secure HIL process and evidence boundaries
w1ne
committed
448229c
fix(bench): freeze and quarantine agent workspaces
w1ne
committed
d63344e
fix(bench): harden quarantine and evidence freezing
w1ne
committed
b1df5e6
revert(bench): remove overengineered HIL orchestration
w1ne
committed
da631ba
feat(bench): add simple ESP32-S3 HIL runner
w1ne
committed
47ae57c
fix(bench): run ESP32-S3 baseline on hardware
w1ne
committed
6cb6e33
fix(agent): use DeepInfra OpenCode base URL
w1ne
committed
aafa074
docs(bench): design cross-runtime HIL smoke test
w1ne
committed
899858f
docs(bench): plan cross-runtime HIL smoke test
w1ne
committed
968ac04
feat(bench): define native runtime adapters
w1ne
committed
e40948a
fix(bench): make runtime command dispatch explicit
w1ne
committed
2a14165
fix(bench): validate runtime adapter dispatch
w1ne
committed
e100e42
feat(bench): share LabWired instructions across runtimes
w1ne
committed
6af68bc
fix(bench): write isolated Codex MCP config
w1ne
committed
058c74a
fix(bench): isolate runtime MCP configuration
w1ne
committed
f8abcc2
feat(bench): run one native agent trial
w1ne
committed
686d122
fix(bench): isolate public agent inputs
w1ne
committed
b9fdb62
fix(bench): preserve native runtime auth
w1ne
committed
cad9f85
fix(bench): isolate Codex MCP without config merge
w1ne
committed
b2deab1
feat(bench): compare native runtimes with one HIL oracle
w1ne
committed
a6cb3a4
fix(bench): validate HIL usage schema fully
w1ne
committed
21e8b70
fix(bench): retain partial usage and gate HIL failures
w1ne
committed
a8ed94f
docs(bench): add cross-runtime smoke entry points
w1ne
committed
1aa0bca
docs(bench): clarify runtime version evidence
w1ne
committed
50b0d6e
fix(bench): validate board identity in runtime matrix
w1ne
committed
6b40ab2
fix(bench): pin HIL flash to identified UART
w1ne
committed
a51e617
fix(bench): emit complete runtime comparison evidence
w1ne
committed
8b02a03
fix(bench): complete runtime evidence details
w1ne
committed
d13400e