TayHCode/sidecarwatch

GitHub ↗ · How this check works

Per-setup public labels: Linux / Haiku: Works · Linux / Luna: Works · Windows / Haiku: Not supported (the README does not support this platform) · Windows / Luna: Not tested · macOS / Haiku: Works · macOS / Luna: Works

Static status badge · SVG · read-only JSON. Embed with ![Coldpass status](https://coldpass.dev/badge/TayHCode--sidecarwatch.svg).

agent-harness

Works

README claims checked out on 4 recorded setups. Untested combinations remain unknown.

Answers apply to each recorded setup, revision and operation. Earlier observations remain in Run details.

Works with my model and platform?
PlatformHaikuLuna
Linuxyesyes
Windows✗ READMEnot tested
macOSyesyes

Linux / Haiku 5.5: yes · Oct 22, 2026 · 8ddf4cfed6d9 · claude-code / 2.1.300 (Claude Code)
Recorded operation: python3 -B -m sidecarwatch demo

Linux / Luna: yes · Oct 22, 2026 · 8ddf4cfed6d9 · codex / codex-cli 0.163.0
Recorded operation: python3 -B -m sidecarwatch demo

Windows / Haiku 5.5: not supported per README · Oct 22, 2026 · 8ddf4cfed6d9 · claude-code / 2.1.300 (Claude Code)
Recorded operation: 'C:\coldpass\python\python.exe' -B -m sidecarwatch demo

macOS / Haiku 5.5: yes · Oct 21, 2026 · 8ddf4cfed6d9 · claude-code / 2.1.300 (Claude Code)
Recorded operation: <guest-path> demo

macOS / Luna: yes · Oct 22, 2026 · 8ddf4cfed6d9 · codex / codex-cli 0.163.0
Recorded operation: /opt/homebrew/bin/python3 -B -m sidecarwatch demo

Does what it says
1 of 1 claims checked out
  • The demo runs offline with synthetic sessions and makes no agent CLI calls.: Real input produced the expected result: DEMO complete: temporary sessions removed; nothing sent.

    Source: `demo` is offline and `watch --dry-run` previews a scrubbed digest locally without calling either reviewer.

    Input: run the documented demo

    Expected: DEMO complete: temporary sessions removed; nothing sent.

    Command: python3 -B -m sidecarwatch demo

    Observed: DEMO complete: temporary sessions removed; nothing sent.

MCP
No
Local or hosted
local
Local: runs on your machine
Needs keys or browser
None seen
Maintained
Maintained
actively maintained: last commit 11 days ago; 2 commits and 1 merged PRs in 90 days; 0 open issues, 0 open PRs; no releases

Last commit Oct 10, 2026

Setup time
19 s
19 s to a working result
Issue
none found in recorded check
Examples
1 of 1 ran as expected
  • Linux documented offline demo: command completed successfully

    Source: README.md, Quickstart: Linux

    Input: python3 -B -m sidecarwatch demo

    Expected: DEMO complete: temporary sessions removed; nothing sent.

    Command: python3 -B -m sidecarwatch demo

    Observed: DEMO: synthetic sessions; no CLI calls. codex session aaaaaaaa (idle) What it is doing Adding CSV export to the reading list and checking a small export test. On task: ok In scope: ok Simple: ok Reuse: ok Root cause: ok Honest: ok Should you correct it? no Label: WORKING Suggested message (not sent) Report the export test results when they are available. DEMO complete: temporary sessions removed; nothing sent.

Evidence
README behavior checked
Agent interfaces
CLI
CLI

Observed interfaces were exercised by the checker. Documentation alone does not confirm compatibility.

Tested with
Codex / Luna
  • Linux: Codex / Luna · agent codex-cli 0.163.0 · Checked operation passed · Oct 22, 2026 on 8ddf4cfed6d9

Each entry describes its recorded operation and source revision. Other agent clients were not tested.

Safety
Something to look at: writes outside the project folder by tests

Review signals across current setups: outside write.

Repository processes attempted these writes outside documented output/cache paths. /usr/bin/python3 1 attributed processes.

PID 35430: /usr/bin/python3 · tests

Full event, path and process lists for each observed setup:

Run details

Haiku 5.5

Native agent · unverified

Operation / attempted scope 'C:\coldpass\python\python.exe' -B -m sidecarwatch demo

Agent-reported scope: Run sidecarwatch demo (offline synthetic report) as documented in the README

Run details
run
TayHCode--sidecarwatch-claude-haiku-5-5-1792648337035043233
date
Oct 22, 2026
time to working
not established
product hosting / execution
Local command in the guest
recorded operation scope
product operation
model
claude-haiku-5-5
agent / version
claude-code / 2.1.300 (Claude Code)
platform
windows
reported cost
$0.023532
Recorded source
https://github.com/TayHCode/sidecarwatch
Backend
hyperv
Tool transport
mcp-assigned-vm
Requested source pin (not an observed revision)
{"commit": "8ddf4cfed6d91ad621ad9b0fcc7bf4c4c39901fe", "url": "https://github.com/TayHCode/sidecarwatch"}
Test version
readme-1924d8f529c6fc4500abb29916f5c9ac942027957e7b82d725d5610040cd3b4a
revision
8ddf4cfed6d91ad621ad9b0fcc7bf4c4c39901fe
Whose fault: prerequisite · diagnosed

sidecarwatch refuses to run on native Windows (os.name != posix guard); README says macOS and Linux only.

Recorded blocker / warning

Step: python -B -m sidecarwatch demo

sidecarwatch refuses to run on native Windows (os.name != posix guard); README says macOS and Linux only.

Safety: No problems traced to this repository code in these checks.

No problems traced to this repository code in these checks.

OSV scanned supported lockfiles; unpinned dependencies and unsupported formats are outside this check.

A short run cannot establish that code is safe.

Connections show IP addresses; proxies can hide the final hostname.

Tracing covers dispatched commands and their children, not the whole OS; guest root can interfere with tracing.

windows behavior observation needs an OS-native file/network/process collector; only the static scan ran.

No Linux syscall trace was collected.

Full safety evidence →

Answers recorded for this run
  • sidecarwatch runs only on macOS and Linux; native Windows is refused: Real input produced the expected result: macOS and Linux only

    Source: macOS and Linux only; WSL is untested.

    Input: python -B -m sidecarwatch demo on this Windows guest

    Expected: exit 2 with the unsupported-platform message

    Command: 'C:\coldpass\python\python.exe' -B -m sidecarwatch demo

    Observed: macOS and Linux only. WSL is untested.

  • demo is offline and needs no agent CLI: The demo runs only on macOS or Linux; this guest is native Windows, where the product refuses to start (os.name check in sidecarwatch), so the offline demo output could not be observed.

    Source: `demo` is offline and `watch --dry-run` previews a scrubbed digest locally without calling either reviewer.

    Observed: The demo runs only on macOS or Linux; this guest is native Windows, where the product refuses to start (os.name check in sidecarwatch), so the offline demo output could not be observed.

Examples

  • Windows is refused with exit 2 (documented Windows check): Ran python -B -m sidecarwatch demo with the absolute guest Python: exit 2, "Native Windows is unsupported; macOS and Linux only." This is covered by probe.json and the replayed claim. The checker cannot accept it as a documented example because the README has no exit-code table row for demo; the README text itself says "Expected: exit 2".

    Source: README.md#windows

    Expected: documented command completes successfully

    Observed: Ran python -B -m sidecarwatch demo with the absolute guest Python: exit 2, "Native Windows is unsupported; macOS and Linux only." This is covered by probe.json and the replayed claim. The checker cannot accept it as a documented example because the README has no exit-code table row for demo; the README text itself says "Expected: exit 2".

  • Repository offline test suite (python3 -B -m unittest discover -s tests -t . -q): Ran with the absolute guest Python: 136 tests, 4 failures, 401 errors, exit 1. Every failing area traces to the native-Windows refusal (os.name != posix), so the suite cannot pass on this platform. Not counted as product success.

    Source: README.md#windows

    Expected: documented command completes successfully

    Observed: Ran with the absolute guest Python: 136 tests, 4 failures, 401 errors, exit 1. Every failing area traces to the native-Windows refusal (os.name != posix), so the suite cannot pass on this platform. Not counted as product success.

Cycles

  • watch repeats its review pass after a wait (recurring trigger): Observed: the command exits 2 with the unsupported-platform message (os.name != posix guard). No cycle ran; the product needs macOS or Linux.

    Source: sidecarwatch/watch.py

    Input: python -B -m sidecarwatch watch --only C:\coldpass\workspace --dry-run (attempted one run-once cycle)

    Expected: one cycle reviews a finished turn and prints results

    Observed: Observed: the command exits 2 with the unsupported-platform message (os.name != posix guard). No cycle ran; the product needs macOS or Linux.

Local or hosted: local

Suggested next step (unverified): Run the offline demo on a macOS or Linux VM (python3 -m sidecarwatch demo, Python 3.11+). Native Windows support would need POSIX-only locking, file modes and state replaced, which the README rules out.

What the agent did

Run sidecarwatch demo (offline synthetic report) as documented in the README

Get-ChildItem -Force C:\coldpass\workspace | Select-Object Name,Length; Get-Content C:\coldpass\workspace\README.md -TotalCount 200; & C:\coldpass\python\python.exe --version
cd /workspace 2>$null; pwd; ls -la; id -u 2>$null; uname -a
cd C:\coldpass\workspace; Get-ChildItem -Force | ForEach-Object { $_.Name }; Test-Path .git; Get-Content pyproject.toml; Get-ChildItem -Recurse -Force sidecarwatch | Select-Object -First 40 FullName; Get-Command git -ErrorAction SilentlyContinue | Select-Object Source
cd C:\coldpass\workspace; & C:\coldpass\python\python.exe -B -m sidecarwatch demo; echo "EXIT=$LASTEXITCODE"; Write-Output '---'; & C:\coldpass\python\python.exe -c "import sys; print(sys.version)"
Get-ChildItem C:\coldpass -Force | ForEach-Object { $_.Name }; Get-Content C:\coldpass\checker\check.py -TotalCount 120; Select-String -Path C:\coldpass\checker\check.py -Pattern 'expected_returncode|expected_text|stays_running|checks|stderr' | Select-Object -First 40 | ForEach-Object { $_.Line }
cd C:\coldpass\workspace; $py='C:\coldpass\python\python.exe'; & $py -B -m sidecarwatch watch --dry-run --only C:\coldpass\workspace; echo "WATCH_EXIT=$LASTEXITCODE"; & $py -B -m sidecarwatch sessions; echo "SESS_EXIT=$LASTEXITCODE"; & $py -B -m unittest discover -s tests -t . -q 2>&1 | Select-Object -Last 15; echo "TEST_EXIT=$LASTEXITCODE"; Select-String -Path sidecarwatch\cli.py,sidecarwatch\__init__.py,sidecarwatch\__main__.py -Pattern 'Native Windows|win32|os.name|platform' | ForEach-Object { $_.Line }
cd C:\coldpass\workspace; New-Item -ItemType Directory -Force .coldpass | Out-Null; $enc = New-Object System.Text.UTF8Encoding($false)
$py = 'C:\coldpass\python\python.exe'
$argv = @($py, '-B', '-m', 'sidecarwatch', 'demo')
$cwd = 'C:\coldpass\workspace'
$probe = [ordered]@{ kind='cli'; argv=$argv; cwd=$cwd; expected_returncode=2; expected_text='macOS and Linux only' }
[System.IO.File]::WriteAllText('C:\coldpass\workspace\.coldpass\probe.json', ($probe | ConvertTo-Json -Depth 5), $enc)
$claims = @(
 [ordered]@{ kind='claim'; claim='sidecarwatch runs only on macOS and Linux; native Windows is refused'; source='macOS and Linux only; WSL is untested.'; input='python -B -m sidecarwatch demo on this Windows guest'; expected='exit 2 with the unsupported-platform message'; probe_kind='cli'; argv=$argv; expected_returncode=2; expected_text='macOS and Linux only'; cwd=$cwd },
 [ordered]@{ claim='demo is offline and needs no agent CLI'; source='`demo` is offline and `watch --dry-run` previews a scrubbed digest locally without calling either reviewer.'; testable=$false; reason='The demo runs only on macOS or Linux; this guest is native Windows, where the product refuses to start (os.name check in sidecarwatch), so the offline demo output could not be observed.' }
)
[System.IO.File]::WriteAllText('C:\coldpass\workspace\.coldpass\claims.json', (ConvertTo-Json -InputObject @($claims) -Depth 5), $enc)
$examples = @(
 [ordered]@{ kind='example'; claim='Windows is refused with exit 2 (documented Windows check)'; source='README.md ## Windows'; input='py -3 -m sidecarwatch demo, run with the absolute guest Python'; expected='exit 2, unsupported message'; argv=$argv; cwd=$cwd; expected_returncode=2; expected_text='macOS and Linux only' },
 [ordered]@{ kind='example'; claim='Repository offline test suite (python3 -B -m unittest discover -s tests -t . -q)'; source='README.md ## Windows (To run the test suite from source)'; reason='Ran with the absolute guest Python: 136 tests, 4 failures, 401 errors, exit 1. Every failing area traces to the native-Windows refusal (os.name != posix), so the suite cannot pass on this platform. Not counted as product success.' }
)
[System.IO.File]::WriteAllText('C:\coldpass\workspace\.coldpass\examples.json', (ConvertTo-Json -InputObject @($examples) -Depth 5), $enc)
$cycles = [ordered]@{ detected=$true; checks=@([ordered]@{ claim='watch reviews newly finished turns on a recurring basis'; source='README.md'; quote='After explicit consent, `watch` reviews newly finished turns, keeps a current needs-you list and writes a daily digest.'; schedule='watch --interval 15m (documented loop, README Review finished turns)'; method='not run: the trigger is refused on native Windows'; input='sidecarwatch watch --only C:\coldpass\workspace --dry-run (attempted one run-once cycle)'; expected='one cycle reviews a finished turn and prints results'; reason='Observed: the command exits 2 with the unsupported-platform message (os.name != posix guard). No cycle ran; the product needs macOS or Linux.' }) }
[System.IO.File]::WriteAllText('C:\coldpass\workspace\.coldpass\cycles.json', (ConvertTo-Json -InputObject $cycles -Depth 6), $enc)
Get-Content .coldpass\probe.json; Get-Content .coldpass\claims.json | Select-Object -First 5
cd C:\coldpass\workspace; $enc = New-Object System.Text.UTF8Encoding($false)
$report = [ordered]@{
 outcome='unsupported'; fault_owner='prerequisite';
 operation='Run sidecarwatch demo (offline synthetic report) as documented in the README';
 operation_scope='Native Windows guest with Python 3.14 at C:\coldpass\python\python.exe. README supports macOS and Linux only.';
 steps=@('Read README.md (Quickstart and Windows sections)','python -B -m sidecarwatch demo (absolute guest Python, the documented Windows check)','python -B -m sidecarwatch watch --dry-run --only C:\coldpass\workspace','python -B -m sidecarwatch sessions','python -B -m unittest discover -s tests -t . -q');
 deviations=@('Guest is Windows PowerShell; README quickstarts target macOS/Linux, so the Linux venv/uv install steps were not run.','Git is not installed and the workspace has no .git; no Git metadata was needed for the operation, so the pinned fetch was not run.','Did not install WSL; README says it is untested and no WSL workflow is recommended.','Repo test suite fails on this platform (136 tests, 4 failures, 401 errors) because of the native-Windows refusal.');
 stuck_step='python -B -m sidecarwatch demo';
 main_blocker='sidecarwatch refuses to run on native Windows (os.name != posix guard); README says macOS and Linux only.';
 plain_failure='The tool does not run on Windows. Every command exits with the message "Native Windows is unsupported; macOS and Linux only." Nothing in the product was broken; this VM is the wrong operating system for it.';
 suggested_fix='Run the offline demo on a macOS or Linux VM (python3 -m sidecarwatch demo, Python 3.11+). Native Windows support would need POSIX-only locking, file modes and state replaced, which the README rules out.'
}
[System.IO.File]::WriteAllText('C:\coldpass\workspace\.coldpass\report.json', (ConvertTo-Json -InputObject $report -Depth 5), $enc)
$env:COLDPASS_WORKSPACE='C:\coldpass\workspace'; & C:\coldpass\python\python.exe -I C:\coldpass\checker\check.py --validate --replay; echo "CHECK_EXIT=$LASTEXITCODE"
cd C:\coldpass\workspace; Select-String -Path C:\coldpass\checker\check.py -Pattern "def validate|def check_cycle|def cycle|source_file|'cycles|scheduler|read_text|open\(" | ForEach-Object { '{0}: {1}' -f $_.LineNumber, $_.Line } | Select-Object -First 60; Select-String -Path sidecarwatch\*.py -Pattern 'interval|while True|time.sleep|--once|--now' | ForEach-Object { '{0}: {1}' -f $_.Filename,$_.Line } | Select-Object -First 30
cd C:\coldpass\workspace; $c = Get-Content C:\coldpass\checker\check.py; $c[1008..1030] | ForEach-Object { $_ }; Write-Output '----watch.py'; $w = Get-Content sidecarwatch\watch.py; for($i=0;$i -lt $w.Count;$i++){ if($w[$i] -match 'while True|interval|wake|passes'){ '{0}: {1}' -f ($i+1), $w[$i] } } | Select-Object -First 30
cd C:\coldpass\workspace; $c = Get-Content C:\coldpass\checker\check.py; $c[1008..1030] | ForEach-Object { $_ }; Write-Output '----watch.py'; $w = Get-Content sidecarwatch\watch.py; for($i=0;$i -lt $w.Count;$i++){ if($w[$i] -match 'while True|interval|wake|passes'){ '{0}: {1}' -f ($i+1), $w[$i] } }
$c = Get-Content C:\coldpass\checker\check.py; $c[1030..1050] | ForEach-Object { $_ }
cd C:\coldpass\workspace; $enc = New-Object System.Text.UTF8Encoding($false)
$py = 'C:\coldpass\python\python.exe'; $argv = @($py, '-B', '-m', 'sidecarwatch', 'demo'); $cwd = 'C:\coldpass\workspace'
$examples = @(
 [ordered]@{ kind='example'; claim='Windows is refused with exit 2 (documented Windows check)'; source='README.md#windows'; input='python -B -m sidecarwatch demo with the absolute guest Python'; expected='exit 2, unsupported message'; argv=$argv; cwd=$cwd; expected_returncode=2; expected_text='macOS and Linux only' },
 [ordered]@{ kind='example'; claim='Repository offline test suite (python3 -B -m unittest discover -s tests -t . -q)'; source='README.md#windows'; reason='Ran with the absolute guest Python: 136 tests, 4 failures, 401 errors, exit 1. Every failing area traces to the native-Windows refusal (os.name != posix), so the suite cannot pass on this platform. Not counted as product success.' }
)
[System.IO.File]::WriteAllText('C:\coldpass\workspace\.coldpass\examples.json', (ConvertTo-Json -InputObject @($examples) -Depth 5), $enc)
$cycles = [ordered]@{ detected=$true; checks=@([ordered]@{ claim='watch repeats its review pass after a wait (recurring trigger)'; source='sidecarwatch/watch.py'; quote='wake = clock.now() + dt.timedelta(seconds=interval)'; schedule='watch --interval 15m (README Review finished turns)'; method='not run: the trigger is refused on native Windows'; input='python -B -m sidecarwatch watch --only C:\coldpass\workspace --dry-run (attempted one run-once cycle)'; expected='one cycle reviews a finished turn and prints results'; reason='Observed: the command exits 2 with the unsupported-platform message (os.name != posix guard). No cycle ran; the product needs macOS or Linux.' }) }
[System.IO.File]::WriteAllText('C:\coldpass\workspace\.coldpass\cycles.json', (ConvertTo-Json -InputObject $cycles -Depth 6), $enc)
$env:COLDPASS_WORKSPACE='C:\coldpass\workspace'; & C:\coldpass\python\python.exe -I C:\coldpass\checker\check.py --validate --replay; echo "CHECK_EXIT=$LASTEXITCODE"
$c = Get-Content C:\coldpass\checker\check.py; for($i=0;$i -lt $c.Count;$i++){ if($c[$i] -match 'example_expectation_documented|documented'){ '{0}: {1}' -f ($i+1), $c[$i] } }
$c = Get-Content C:\coldpass\checker\check.py; $c[1050..1070] | ForEach-Object { $_ }; Write-Output '---'; $c[1100..1150] | ForEach-Object { $_ }
cd C:\coldpass\workspace; $enc = New-Object System.Text.UTF8Encoding($false)
$examples = @(
 [ordered]@{ kind='example'; claim='Windows is refused with exit 2 (documented Windows check)'; source='README.md#windows'; reason='Ran python -B -m sidecarwatch demo with the absolute guest Python: exit 2, "Native Windows is unsupported; macOS and Linux only." This is covered by probe.json and the replayed claim. The checker cannot accept it as a documented example because the README has no exit-code table row for demo; the README text itself says "Expected: exit 2".' },
 [ordered]@{ kind='example'; claim='Repository offline test suite (python3 -B -m unittest discover -s tests -t . -q)'; source='README.md#windows'; reason='Ran with the absolute guest Python: 136 tests, 4 failures, 401 errors, exit 1. Every failing area traces to the native-Windows refusal (os.name != posix), so the suite cannot pass on this platform. Not counted as product success.' }
)
[System.IO.File]::WriteAllText('C:\coldpass\workspace\.coldpass\examples.json', (ConvertTo-Json -InputObject @($examples) -Depth 5), $enc)
$env:COLDPASS_WORKSPACE='C:\coldpass\workspace'; & C:\coldpass\python\python.exe -I C:\coldpass\checker\check.py --validate --replay; echo "CHECK_EXIT=$LASTEXITCODE"
Earlier attempts (4)

Luna

Native agent · Checked operation passed

Checked operation python3 -B -m sidecarwatch demo

Agent-reported scope: Run the documented offline synthetic demo

Run details
run
TayHCode--sidecarwatch-gpt-6-luna-1792683670509381866
date
Oct 22, 2026
time to working
18.9s
product hosting / execution
Local command in the guest
recorded operation scope
product operation
model
gpt-6-luna
agent / version
codex / codex-cli 0.163.0
platform
linux
reported cost
not reported (not assumed free)
Recorded source
https://github.com/TayHCode/sidecarwatch
Backend
hyperv
Tool transport
mcp-assigned-vm
Requested source pin (not an observed revision)
{"commit": "8ddf4cfed6d91ad621ad9b0fcc7bf4c4c39901fe", "url": "https://github.com/TayHCode/sidecarwatch"}
Test version
readme-1002bc876f0002f622e14f9c5b99f9f66721fccc0356f4d53ed24f2f79a3ac5d
revision
8ddf4cfed6d91ad621ad9b0fcc7bf4c4c39901fe
Whose fault: no fault assigned · resolved

Safety: No problems traced to this repository code in these checks.

No problems traced to this repository code in these checks.

OSV scanned supported lockfiles; unpinned dependencies and unsupported formats are outside this check.

A short run cannot establish that code is safe.

Connections show IP addresses; proxies can hide the final hostname.

Tracing covers dispatched commands and their children, not the whole OS; guest root can interfere with tracing.

Full safety evidence →

Answers recorded for this run
  • The demo runs offline with synthetic sessions and makes no agent CLI calls.: Real input produced the expected result: DEMO complete: temporary sessions removed; nothing sent.

    Source: `demo` is offline and `watch --dry-run` previews a scrubbed digest locally without calling either reviewer.

    Input: run the documented demo

    Expected: DEMO complete: temporary sessions removed; nothing sent.

    Command: python3 -B -m sidecarwatch demo

    Observed: DEMO complete: temporary sessions removed; nothing sent.

Examples

  • Linux documented offline demo: command completed successfully

    Source: README.md, Quickstart: Linux

    Input: python3 -B -m sidecarwatch demo

    Expected: DEMO complete: temporary sessions removed; nothing sent.

    Command: python3 -B -m sidecarwatch demo

    Observed: DEMO: synthetic sessions; no CLI calls. codex session aaaaaaaa (idle) What it is doing Adding CSV export to the reading list and checking a small export test. On task: ok In scope: ok Simple: ok Reuse: ok Root cause: ok Honest: ok Should you correct it? no Label: WORKING Suggested message (not sent) Report the export test results when they are available. DEMO complete: temporary sessions removed; nothing sent.

Local or hosted: local

What the agent did

Run the documented offline synthetic demo

pwd; rg --files -g 'README*' -g 'AGENTS.md' -g 'pyproject.toml' -g 'package.json' -g 'Makefile' -g '*.yml' -g '*.yaml' | head -80; git status --short --branch; git rev-parse HEAD; python3 --version; id -u; command -v vault-search || true
find . -maxdepth 3 -type f | sort | head -100; cat README.md
python3 -B -m sidecarwatch demo; find . -name AGENTS.md -print
mkdir -p .coldpass; python3 - <<'PY'
import json
from pathlib import Path
p=Path('.coldpass')
argv=['python3','-B','-m','sidecarwatch','demo']
quote='`python3 -m sidecarwatch demo`'
expected='DEMO complete: temporary sessions removed; nothing sent.'
claim={'kind':'claim','claim':'The demo runs offline with synthetic sessions and makes no agent CLI calls.','source':quote,'input':'run the documented demo','expected':expected,'probe_kind':'cli','argv':argv,'expected_text':expected,'cwd':'<guest-path>'}
probe={'kind':'cli','argv':argv,'cwd':'<guest-path>','expected_text':expected}
(p/'claims.json').write_text(json.dumps([claim],indent=2)+'\n')
(p/'probe.json').write_text(json.dumps(probe,indent=2)+'\n')
(p/'cycles.json').write_text(json.dumps({'detected':False,'reason':'README Scheduling says the tool installs no timer and runs its one-pass watch form from an external scheduler; README inspection found no product-owned scheduler cycle.','checks':[]},indent=2)+'\n')
(p/'examples.json').write_text(json.dumps([{'kind':'example','claim':'Linux documented offline demo','source':'README.md, Quickstart: Linux','input':'python3 -B -m sidecarwatch demo','expected':expected,'argv':argv,'cwd':'<guest-path>','expected_text':expected}],indent=2)+'\n')
(p/'report.json').write_text(json.dumps({'outcome':'working','fault_owner':None,'operation':'Run the documented offline synthetic demo','operation_scope':'Python 3.12.3; stdlib only, no runtime dependencies','steps':['Checked Python 3.12.3 against the README minimum of 3.11.','Ran python3 -B -m sidecarwatch demo; it printed a WORKING synthetic report and confirmed temporary sessions were removed.'],'deviations':[],'stuck_step':None,'main_blocker':None,'plain_failure':None,'suggested_fix':None},indent=2)+'\n')
PY
COLDPASS_WORKSPACE=<guest-path> python3 -I <guest-path> --validate --replay
python3 - <<'PY'
import json
from pathlib import Path
p=Path('.coldpass/claims.json'); c=json.loads(p.read_text()); c[0]['source']='`demo` is offline and `watch --dry-run` previews a scrubbed digest locally without calling either reviewer.'; p.write_text(json.dumps(c,indent=2)+'\n')
PY
COLDPASS_WORKSPACE=<guest-path> python3 -I <guest-path> --validate --replay

Luna

Native agent · Checked operation passed

Checked operation /opt/homebrew/bin/python3 -B -m sidecarwatch demo

Agent-reported scope: Run the documented synthetic demo from source.

Run details
run
TayHCode--sidecarwatch-gpt-6-luna-1792682260582571530
date
Oct 22, 2026
time to working
212.3s
product hosting / execution
Local command in the guest
recorded operation scope
product operation
model
gpt-6-luna
agent / version
codex / codex-cli 0.163.0
platform
macos
reported cost
not reported (not assumed free)
Recorded source
https://github.com/TayHCode/sidecarwatch
Backend
tart
Tool transport
mcp-assigned-vm
Requested source pin (not an observed revision)
{"commit": "8ddf4cfed6d91ad621ad9b0fcc7bf4c4c39901fe", "url": "https://github.com/TayHCode/sidecarwatch"}
Test version
readme-e49b83e18e92427844d11f3118bec9288175c1147713a0bc6037ce8a852b3998
revision
8ddf4cfed6d91ad621ad9b0fcc7bf4c4c39901fe
Whose fault: no fault assigned · resolved

Safety: No problems traced to this repository code in these checks.

No problems traced to this repository code in these checks.

OSV scanned supported lockfiles; unpinned dependencies and unsupported formats are outside this check.

A short run cannot establish that code is safe.

Connections show IP addresses; proxies can hide the final hostname.

Tracing covers dispatched commands and their children, not the whole OS; guest root can interfere with tracing.

macos behavior observation needs an OS-native file/network/process collector; only the static scan ran.

No Linux syscall trace was collected.

Full safety evidence →

Answers recorded for this run
  • The demo runs offline with synthetic sessions and sends nothing.: Real input produced the expected result: DEMO complete: temporary sessions removed; nothing sent.

    Source: `demo` is offline and `watch --dry-run` previews a scrubbed digest locally without calling either reviewer.

    Input: the documented demo invocation

    Expected: The synthetic demo completes and says nothing was sent.

    Command: /opt/homebrew/bin/python3 -B -m sidecarwatch demo

    Observed: DEMO complete: temporary sessions removed; nothing sent.

  • The demo makes no CLI calls.: Real input produced the expected result: DEMO: synthetic sessions; no CLI calls.

    Source: `demo` uses temporary synthetic sessions, calls no agent CLI or model, and removes its data.

    Input: the documented demo invocation

    Expected: Output identifies synthetic sessions and no CLI calls.

    Command: /opt/homebrew/bin/python3 -B -m sidecarwatch demo

    Observed: DEMO: synthetic sessions; no CLI calls. codex session aaaaaaaa (idle) What it is doing Adding CSV export to the reading list and checking a small export test. On task: ok In scope: ok Simple: ok Reuse: ok Root cause: ok Honest: ok Should you correct it? no Label: WORKING Suggested message (not sent) Report the export test results when they are available. DEMO complete: temporary sessions removed; nothing sent.

Examples

  • Run the macOS README quickstart demo from source.: command completed successfully

    Source: README.md / Quickstart: macOS

    Input: demo

    Expected: synthetic demo completes without sending anything

    Command: /opt/homebrew/bin/python3 -B -m sidecarwatch demo

    Observed: DEMO: synthetic sessions; no CLI calls. codex session aaaaaaaa (idle) What it is doing Adding CSV export to the reading list and checking a small export test. On task: ok In scope: ok Simple: ok Reuse: ok Root cause: ok Honest: ok Should you correct it? no Label: WORKING Suggested message (not sent) Report the export test results when they are available. DEMO complete: temporary sessions removed; nothing sent.

  • Run the README documented source test suite.: command completed successfully

    Source: README.md / Check your own setup

    Input: unittest discovery

    Expected: suite passes

    Command: /opt/homebrew/bin/python3 -B -m unittest discover -s tests -t . -q

    Observed: ---------------------------------------------------------------------- Ran 215 tests in 88.082s OK (skipped=1)

Local or hosted: local

What the agent did

Run the documented synthetic demo from source.

cd <guest-path> && (vault-search 'sidecarwatch README install run MCP claims cycles' || true) && ls -la && rg --files -g 'README*' -g 'AGENTS.md' -g 'pyproject.toml' -g 'package.json' -g 'requirements*.txt' -g 'Makefile' -g '.github/workflows/*' | head -100 && cat README.md
cd <guest-path> && /opt/homebrew/bin/python3 --version && cat pyproject.toml && find . -name AGENTS.md -o -path './.github/workflows/*' | sort && git status --short --branch && /opt/homebrew/bin/python3 -B -m sidecarwatch demo
cd <guest-path> && git init && git remote add origin https://github.com/TayHCode/sidecarwatch && git fetch --depth 1 origin 8ddf4cfed6d91ad621ad9b0fcc7bf4c4c39901fe && git reset --mixed FETCH_HEAD && git status --short --branch
cd <guest-path> && /opt/homebrew/bin/python3 -B -m sidecarwatch demo && /opt/homebrew/bin/python3 -B -m sidecarwatch --help | head -40 && find tests -maxdepth 1 -type f | sort | head
cd <guest-path> && sed -n '1,180p' sidecarwatch/__main__.py && sed -n '1,100p' sidecarwatch/__init__.py && sed -n '1,120p' sidecarwatch/cli.py && find .coldpass -maxdepth 2 -type f -print
cd <guest-path> && /opt/homebrew/bin/python3 -B -m sidecarwatch init --help >/dev/null; mkdir -p .coldpass && cat > .coldpass/probe.json <<'EOF'
{"kind":"cli","argv":["/opt/homebrew/bin/python3","-B","-m","sidecarwatch","demo"],"cwd":"<guest-path>","expected_text":"DEMO complete: temporary sessions removed; nothing sent."}
EOF
cat > .coldpass/claims.json <<'EOF'
[
 {"kind":"claim","claim":"The demo runs offline with synthetic sessions and sends nothing.","source":"`demo` is offline and `watch --dry-run` previews a scrubbed digest locally without calling either reviewer.","input":"the documented demo invocation","expected":"The synthetic demo completes and says nothing was sent.","probe_kind":"cli","argv":["/opt/homebrew/bin/python3","-B","-m","sidecarwatch","demo"],"expected_text":"DEMO complete: temporary sessions removed; nothing sent.","cwd":"<guest-path>"},
 {"kind":"claim","claim":"The demo makes no CLI calls.","source":"`demo` uses temporary synthetic sessions, calls no agent CLI or model, and removes its data.","input":"the documented demo invocation","expected":"Output identifies synthetic sessions and no CLI calls.","probe_kind":"cli","argv":["/opt/homebrew/bin/python3","-B","-m","sidecarwatch","demo"],"expected_text":"DEMO: synthetic sessions; no CLI calls.","cwd":"<guest-path>"}
]
EOF
cat > .coldpass/examples.json <<'EOF'
[
 {"kind":"example","claim":"Run the macOS README quickstart demo from source.","source":"README.md / Quickstart: macOS","input":"demo","expected":"synthetic demo completes without sending anything","argv":["/opt/homebrew/bin/python3","-B","-m","sidecarwatch","demo"],"cwd":"<guest-path>","expected_text":"DEMO complete: temporary sessions removed; nothing sent."},
 {"kind":"example","claim":"Run the README documented source test suite.","source":"README.md / Check your own setup","input":"unittest discovery","expected":"suite passes","argv":["/opt/homebrew/bin/python3","-B","-m","unittest","discover","-s","tests","-t",".","-q"],"cwd":"<guest-path>"}
]
EOF
cat > .coldpass/cycles.json <<'EOF'
{"detected":false,"reason":"README Scheduling states the tool installs no timer and recommends running its one-pass watch form from an external scheduler; watch without --interval makes one pass. A single-pass operation is not a product cycle scheduler.","checks":[]}
EOF
cat > .coldpass/report.json <<'EOF'
{"outcome":"working","fault_owner":null,"operation":"Run the documented synthetic demo from source.","operation_scope":"Python 3.14.8 on macOS; source snapshot at pinned commit.","steps":["Verified Python 3.14.8 meets README requirement 3.11+.","Initialized Git metadata and fetched/reset to pinned commit 8ddf4cfed6d91ad621ad9b0fcc7bf4c4c39901fe.","Ran /opt/homebrew/bin/python3 -B -m sidecarwatch demo successfully."],"deviations":["vault-search and rg are unavailable in this guest."],"stuck_step":null,"main_blocker":null,"plain_failure":null,"suggested_fix":null}
EOF
COLDPASS_WORKSPACE=<guest-path> /opt/homebrew/bin/python3 -I <guest-path> --validate --replay

Haiku 5.5

Native agent · Checked operation passed

Checked operation python3 -B -m sidecarwatch demo

Agent-reported scope: Run sidecarwatch from source per README: offline demo, status, one watch pass, and the stdlib test suite

Run details
run
TayHCode--sidecarwatch-claude-haiku-5-5-1792644783927316756
date
Oct 22, 2026
time to working
2301.2s
product hosting / execution
Local command in the guest
recorded operation scope
product operation
model
claude-haiku-5-5
agent / version
claude-code / 2.1.300 (Claude Code)
platform
linux
reported cost
$0.019885
Recorded source
https://github.com/TayHCode/sidecarwatch
Backend
hyperv
Tool transport
mcp-assigned-vm
Requested source pin (not an observed revision)
{"commit": "8ddf4cfed6d91ad621ad9b0fcc7bf4c4c39901fe", "url": "https://github.com/TayHCode/sidecarwatch"}
Test version
readme-1924d8f529c6fc4500abb29916f5c9ac942027957e7b82d725d5610040cd3b4a
revision
8ddf4cfed6d91ad621ad9b0fcc7bf4c4c39901fe
Whose fault: no fault assigned · resolved

Safety: Review signals: 1 outside write.

Review signals: 1 outside write.

Repository processes attempted these writes outside documented output/cache paths. /usr/bin/python3 1 attributed processes.

PID 35430: /usr/bin/python3 · tests

OSV scanned supported lockfiles; unpinned dependencies and unsupported formats are outside this check.

A short run cannot establish that code is safe.

Connections show IP addresses; proxies can hide the final hostname.

Tracing covers dispatched commands and their children, not the whole OS; guest root can interfere with tracing.

Full safety evidence →

Answers recorded for this run
  • demo runs offline with synthetic sessions and calls no agent CLI: Real input produced the expected result: Label: WORKING

    Source: `demo` uses temporary synthetic sessions, calls no agent CLI or model, and removes its data.

    Input: python3 -B -m sidecarwatch demo

    Expected: synthetic WORKING report and completion line with nothing sent

    Command: python3 -B -m sidecarwatch demo

    Observed: Label: WORKING Suggested message (not sent) Report the export test results when they are available. DEMO complete: temporary sessions removed; nothing sent.

  • status shows whether cross-vendor review is ON or OFF: Real input produced the expected result: Cross-vendor review: OFF

    Source: - `status` shows whether cross-vendor review is ON or OFF.

    Input: python3 -B -m sidecarwatch status

    Expected: Cross-vendor review reported OFF by default

    Command: python3 -B -m sidecarwatch status

    Observed: Cross-vendor review: OFF Config: <guest-path> Kill switch: off Needs you: 0 open; run open to inspect.

  • a watch run without --interval makes one pass: Real input produced the expected result: 0 in scope

    Source: Without `--interval`, a run makes one pass.

    Input: watch --only <guest-path> --dry-run

    Expected: one pass summary line and exit 0

    Command: python3 -B -m sidecarwatch watch --only <guest-path> --dry-run

    Observed: 0 in scope · 0 reviewed · 0 sent · 0 would send · 0 need you

Examples

  • documented Linux quickstart demo: command completed successfully

    Source: README.md, Quickstart: Linux

    Input: python3 -B -m sidecarwatch demo

    Expected: synthetic WORKING report

    Command: python3 -B -m sidecarwatch demo

    Observed: DEMO: synthetic sessions; no CLI calls. codex session aaaaaaaa (idle) What it is doing Adding CSV export to the reading list and checking a small export test. On task: ok In scope: ok Simple: ok Reuse: ok Root cause: ok Honest: ok Should you correct it? no Label: WORKING Suggested message (not sent) Report the export test results when they are available. DEMO complete: temporary sessions removed; nothing sent.

  • documented test suite from source: Suite takes about 5 minutes (295s) and exceeded the checker replay time budget. Run manually with the venv active: EXIT=0, 'Ran 215 tests', 'OK (skipped=1)'; log at <guest-path>

    Source: README.md, To run the test suite from source

    Input: python3 -B -m unittest discover -s tests -t . -q

    Expected: all tests pass (215 run, 1 skipped)

    Observed: Suite takes about 5 minutes (295s) and exceeded the checker replay time budget. Run manually with the venv active: EXIT=0, 'Ran 215 tests', 'OK (skipped=1)'; log at <guest-path>

Cycles

  • watch runs one review pass per trigger; --interval repeats it: Triggered one cycle by hand (run-once via CLI; normal trigger: single pass (no --interval); the loop only runs with --interval): 05:31 0 in scope · 0 reviewed · 0 sent · 0 would send · 0 need you

    Source: sidecarwatch/watch.py

    Input: watch --only <guest-path> --dry-run with no session files

    Expected: one pass summary and exit 0, no reviewer call

    Command: python3 -B -m sidecarwatch watch --only <guest-path> --dry-run

    Observed: 05:31 0 in scope · 0 reviewed · 0 sent · 0 would send · 0 need you

Local or hosted: local

What the agent did

Run sidecarwatch from source per README: offline demo, status, one watch pass, and the stdlib test suite

cd <guest-path> && ls -la && git status 2>&1 | head -3; id -u; cat /etc/os-release | head -2; python3 --version; ls /inputs 2>&1
cd <guest-path> && cat README.md 2>/dev/null || find . -maxdepth 2 -iname 'readme*' | head
cd <guest-path> && python3 -B -m sidecarwatch demo; echo "EXIT=$?"; python3 -B -m sidecarwatch --version; echo "EXIT=$?"
cd <guest-path> && ls -la .coldpass examples docs sidecarwatch tests | head -60; cat .gitignore; python3 -c "import ensurepip" 2>&1; dpkg -s python3-venv >/dev/null 2>&1 && echo venv-pkg-installed || echo venv-pkg-missing
cd <guest-path> && sudo -n true && echo SUDO_OK; (sudo -n apt-get update -qq && sudo -n DEBIAN_FRONTEND=noninteractive apt-get install -y -qq python3-venv) 2>&1 | tail -5; echo "EXIT=$?"; python3 -c "import ensurepip; print('ensurepip ok')"
cd <guest-path> && python3 -m venv .venv && . .venv/bin/activate && python3 -m pip install -q --upgrade pip 2>&1 | tail -3; python3 -m pip install . 2>&1 | tail -4; echo "EXIT=${PIPESTATUS[0]}"; which sidecarwatch; sidecarwatch demo; echo "EXIT=$?"
cd <guest-path> && . .venv/bin/activate && python3 -B -m unittest discover -s tests -t . -q 2>&1 | tail -6; echo "TESTS_EXIT=${PIPESTATUS[0]}"; sidecarwatch status; echo "STATUS_EXIT=$?"; sidecarwatch doctor --shapes 2>&1 | head -30; echo "DOCTOR_EXIT=${PIPESTATUS[0]}"; sidecarwatch sessions; echo "SESSIONS_EXIT=$?"
cd <guest-path> && . .venv/bin/activate && timeout 900 python3 -B -m unittest discover -s tests -t . -q 2>&1 | grep -B3 -A25 "^FAIL:" | head -80
cd <guest-path> && . .venv/bin/activate && timeout 1500 python3 -B -m unittest discover -s tests -t . -q > <guest-path> 2>&1; echo "EXIT=$?"; grep -n "^FAIL:\|^ERROR:\|^SKIP\|skipped" <guest-path> | head; wc -c <guest-path>
cd <guest-path> && sed -n 1,30p <guest-path> | cut -c1-400; echo ----; grep -n "test_repo_is_generic" -A30 tests/test_structure.py | head -50
cd <guest-path> && grep -n "def tracked" -A6 tests/test_structure.py | cut -c1-300
cd <guest-path> && grep -rn "tracked" tests/*.py | head -5 | cut -c1-400
cd <guest-path> && sed -n 8,25p tests/test_forbidden_terms.py | cut -c1-300
cd <guest-path> && git init -q && git remote add origin https://github.com/TayHCode/sidecarwatch && timeout 180 git fetch --depth 1 origin 8ddf4cfed6d91ad621ad9b0fcc7bf4c4c39901fe 2>&1 | tail -5; echo "FETCH_EXIT=${PIPESTATUS[0]}"; git reset --mixed FETCH_HEAD 2>&1 | tail -3; git log --oneline -1 2>&1; git status --short | head -20
cd <guest-path> && . .venv/bin/activate && timeout 1500 python3 -B -m unittest discover -s tests -t . -q > <guest-path> 2>&1; echo "EXIT=$?"; tail -4 <guest-path> | cut -c1-300; grep -n "^FAIL:\|^ERROR:" <guest-path> | head
cd <guest-path> && . .venv/bin/activate && sidecarwatch watch --help 2>&1 | head -40; echo ----; sidecarwatch watch --only <guest-path> --dry-run; echo "WATCH_EXIT=$?"
cd <guest-path> && grep -nF '`demo` uses temporary synthetic sessions, calls no agent CLI or model, and removes its data.' README.md | cut -c1-120; grep -nF -- '- `status` shows whether cross-vendor review is ON or OFF.' README.md; grep -nF 'python3 -B -m unittest discover -s tests -t . -q' README.md; grep -nF 'Without `--interval`, a run makes one pass.' README.md | cut -c1-80; echo ----; grep -n "interval\|max_passes\|while\|sleep" sidecarwatch/watch.py sidecarwatch/cli.py | head -30 | cut -c1-200
cd <guest-path> && mkdir -p .coldpass && cat > .coldpass/probe.json <<'EOF'
{"kind":"cli","argv":["python3","-B","-m","sidecarwatch","demo"],"cwd":"<guest-path>","expected_text":"Label: WORKING"}
EOF
cat > .coldpass/claims.json <<'EOF'
[
{"kind":"claim","claim":"demo runs offline with synthetic sessions and calls no agent CLI","source":"`demo` uses temporary synthetic sessions, calls no agent CLI or model, and removes its data.","input":"python3 -B -m sidecarwatch demo","expected":"synthetic WORKING report and completion line with nothing sent","probe_kind":"cli","argv":["python3","-B","-m","sidecarwatch","demo"],"cwd":"<guest-path>","expected_text":"Label: WORKING"},
{"kind":"claim","claim":"status shows whether cross-vendor review is ON or OFF","source":"- `status` shows whether cross-vendor review is ON or OFF.","input":"python3 -B -m sidecarwatch status","expected":"Cross-vendor review reported OFF by default","probe_kind":"cli","argv":["python3","-B","-m","sidecarwatch","status"],"cwd":"<guest-path>","expected_text":"Cross-vendor review: OFF"},
{"kind":"claim","claim":"a watch run without --interval makes one pass","source":"Without `--interval`, a run makes one pass.","input":"watch --only <guest-path> --dry-run","expected":"one pass summary line and exit 0","probe_kind":"cli","argv":["python3","-B","-m","sidecarwatch","watch","--only","<guest-path>","--dry-run"],"cwd":"<guest-path>","expected_text":"0 in scope"}
]
EOF
cat > .coldpass/examples.json <<'EOF'
[
{"kind":"example","claim":"documented Linux quickstart demo","source":"README.md, Quickstart: Linux","input":"python3 -B -m sidecarwatch demo","expected":"synthetic WORKING report","argv":["python3","-B","-m","sidecarwatch","demo"],"cwd":"<guest-path>","env":{},"expected_text":"DEMO complete"},
{"kind":"example","claim":"documented test suite from source","source":"README.md, To run the test suite from source","input":"python3 -B -m unittest discover -s tests -t . -q","expected":"all tests pass (215 run, 1 skipped)","argv":["python3","-B","-m","unittest","discover","-s","tests","-t",".","-q"],"cwd":"<guest-path>","env":{}}
]
EOF
cat > .coldpass/cycles.json <<'EOF'
{"detected":true,"checks":[{"claim":"watch runs one review pass per trigger; --interval repeats it","source":"sidecarwatch/watch.py","quote":"interval is None or args.max_passes and passes >= args.max_passes","schedule":"single pass (no --interval); the loop only runs with --interval","method":"run-once via CLI","input":"watch --only <guest-path> --dry-run with no session files","expected":"one pass summary and exit 0, no reviewer call","kind":"cli","argv":["python3","-B","-m","sidecarwatch","watch","--only","<guest-path>","--dry-run"],"cwd":"<guest-path>","expected_text":"0 in scope"}]}
EOF
for f in probe claims examples cycles; do python3 -I -c "import json,sys;json.load(open(sys.argv[1]));print('ok',sys.argv[1])" .coldpass/$f.json; done
cd <guest-path> && COLDPASS_WORKSPACE=<guest-path> timeout 1500 python3 -I <guest-path> --validate --replay > <guest-path> 2>&1; echo "CHECK_EXIT=$?"; cut -c1-400 <guest-path> | tail -40
cd <guest-path> && cat > .coldpass/examples.json <<'EOF'
[
{"kind":"example","claim":"documented Linux quickstart demo","source":"README.md, Quickstart: Linux","input":"python3 -B -m sidecarwatch demo","expected":"synthetic WORKING report","argv":["python3","-B","-m","sidecarwatch","demo"],"cwd":"<guest-path>","env":{},"expected_text":"DEMO complete"},
{"kind":"example","claim":"documented test suite from source","source":"README.md, To run the test suite from source","input":"python3 -B -m unittest discover -s tests -t . -q","expected":"all tests pass (215 run, 1 skipped)","deferred":true,"reason":"Suite takes about 5 minutes (295s) and exceeded the checker replay time budget. Run manually with the venv active: EXIT=0, 'Ran 215 tests', 'OK (skipped=1)'; log at <guest-path>"}
]
EOF
python3 -I -c "import json;json.load(open('.coldpass/examples.json'));print('ok')"; COLDPASS_WORKSPACE=<guest-path> timeout 1500 python3 -I <guest-path> --validate --replay > <guest-path> 2>&1; echo "CHECK_EXIT=$?"; cut -c1-600 <guest-path> | tail -20
cd <guest-path> && cat > .coldpass/report.json <<'EOF'
{"outcome":"working","fault_owner":null,"operation":"Run sidecarwatch from source per README: offline demo, status, one watch pass, and the stdlib test suite","operation_scope":"Linux Ubuntu 24.04, Python 3.12.3 (README needs 3.11+), local only, no agent CLI calls, cross-vendor review left OFF, no servers started","steps":["python3 --version","python3 -B -m sidecarwatch demo","sudo apt-get install -y python3-venv (needed for venv route)","python3 -m venv .venv && python3 -m pip install . (venv in .venv, then sidecarwatch demo)","git init; git remote add origin https://github.com/TayHCode/sidecarwatch; git fetch --depth 1 origin 8ddf4cfed6d91ad621ad9b0fcc7bf4c4c39901fe; git reset --mixed FETCH_HEAD","python3 -B -m unittest discover -s tests -t . -q (215 run, 1 skipped, OK)","python3 -B -m sidecarwatch status / doctor --shapes / sessions","python3 -B -m sidecarwatch watch --only <guest-path> --dry-run (one pass)"],"deviations":["python3-venv was not installed; installed via sudo apt-get install python3-venv as README instructs","First test run failed test_repo_is_generic because the venv inside the checkout was scanned (no git metadata, so tests fall back to scanning all files); fixed by the README-directed git init/fetch of the pinned commit; re-run passed","pip install . created build/ and egg-info folders in the checkout (gitignored, README says expected)","Full test suite takes about 5 minutes; it was excluded from checker replay as deferred because replay exceeded its time budget"],"stuck_step":null,"main_blocker":null,"plain_failure":null,"suggested_fix":"None required. Optional: make test_structure not depend on a git checkout or ignore .venv when no git metadata is present."}
EOF
python3 -I -c "import json;d=json.load(open('.coldpass/report.json'));print(d['outcome'])"; COLDPASS_WORKSPACE=<guest-path> python3 -I <guest-path> --validate --replay 2>&1 | cut -c1-200 | head -3; ls .coldpass; pgrep -fa sidecarwatch | grep -v pgrep || echo "no sidecarwatch processes running"

Haiku 5.5

Native agent · Checked operation passed

Checked operation <guest-path> demo

Agent-reported scope: Run the sidecarwatch offline synthetic demo (`sidecarwatch demo`), plus status, one-pass watch, doctor, and the unit test suite

Run details
run
TayHCode--sidecarwatch-claude-haiku-5-5-1792640335110086691
date
Oct 21, 2026
time to working
506.4s
product hosting / execution
Local command in the guest
recorded operation scope
product operation
model
claude-haiku-5-5
agent / version
claude-code / 2.1.300 (Claude Code)
platform
macos
reported cost
$0.015911
Recorded source
https://github.com/TayHCode/sidecarwatch
Backend
tart
Tool transport
mcp-assigned-vm
Requested source pin (not an observed revision)
{"commit": "8ddf4cfed6d91ad621ad9b0fcc7bf4c4c39901fe", "url": "https://github.com/TayHCode/sidecarwatch"}
Test version
readme-6f8cbfa597c6d59f7ece0e29599b48d61f780e82e853b6ab300c3a8265684642
revision
8ddf4cfed6d91ad621ad9b0fcc7bf4c4c39901fe
Whose fault: no fault assigned · resolved

Safety: No problems traced to this repository code in these checks.

No problems traced to this repository code in these checks.

OSV scanned supported lockfiles; unpinned dependencies and unsupported formats are outside this check.

A short run cannot establish that code is safe.

Connections show IP addresses; proxies can hide the final hostname.

Tracing covers dispatched commands and their children, not the whole OS; guest root can interfere with tracing.

macos behavior observation needs an OS-native file/network/process collector; only the static scan ran.

No Linux syscall trace was collected.

Full safety evidence →

Answers recorded for this run
  • demo runs offline on synthetic sessions: Real input produced the expected result: DEMO complete: temporary sessions removed; nothing sent.

    Source: `demo` uses temporary synthetic sessions, calls no agent CLI or model, and removes its data.

    Input: sidecarwatch demo

    Expected: synthetic demo completes with nothing sent

    Command: <guest-path> demo

    Observed: DEMO complete: temporary sessions removed; nothing sent.

  • status shows cross-vendor review state: Real input produced the expected result: Cross-vendor review: OFF

    Source: - `status` shows whether cross-vendor review is ON or OFF.

    Input: sidecarwatch status

    Expected: reports cross-vendor review OFF by default

    Command: <guest-path> status

    Observed: Cross-vendor review: OFF Config: <guest-path> Kill switch: off Needs you: 0 open; run open to inspect.

  • watch without interval makes one pass: Real input produced the expected result: 0 in scope

    Source: Without `--interval`, a run makes one pass.

    Input: watch --only <guest-path> --dry-run

    Expected: single pass summary printed, then exit

    Command: <guest-path> watch --only <guest-path> --dry-run

    Observed: 0 in scope · 0 reviewed · 0 sent · 0 would send · 0 need you

Examples

  • synthetic demo quickstart with system Python 3.14: command completed successfully

    Source: README.md Quickstart: Linux

    Input: python3 -B -m sidecarwatch demo

    Expected: offline demo output

    Command: /opt/homebrew/bin/python3 -B -m sidecarwatch demo

    Observed: DEMO: synthetic sessions; no CLI calls. codex session aaaaaaaa (idle) What it is doing Adding CSV export to the reading list and checking a small export test. On task: ok In scope: ok Simple: ok Reuse: ok Root cause: ok Honest: ok Should you correct it? no Label: WORKING Suggested message (not sent) Report the export test results when they are available. DEMO complete: temporary sessions removed; nothing sent.

  • doctor reports local setup and review OFF: command completed successfully

    Source: README.md Check your own setup

    Input: doctor --shapes

    Expected: local setup report

    Command: <guest-path> doctor --shapes

    Observed: y Scanned: ~/.claude/projects/*/*.jsonl Login: unverified; CLI executable unavailable or --help failed; check the configured command (or use --reviewer with the same CLI) Reviewer flags: Shapes: (none) Unknown lines: 0; mostly unknown sessions: 0 codex: review unavailable Home: ~/.codex; sessions: 0; delivery: print only Scanned: ~/.codex/sessions/**/rollout-*.jsonl Login: unverified; CLI executable unavailable or --help failed; check the configured command (or use --reviewer with the same CLI) Reviewer flags: Shapes: (none) Unknown lines: 0; mostly unknown sessions: 0

  • documented unit test suite passes: command completed successfully

    Source: README.md Check your own setup (To run the test suite from source)

    Input: python3 -B -m unittest discover -s tests -t . -q

    Expected: all tests pass

    Command: <guest-path> -B -m unittest discover -s tests -t . -q

    Observed: ---------------------------------------------------------------------- Ran 215 tests in 83.246s OK (skipped=1)

Cycles

  • watch runs one pass and stops when no interval is set: Triggered one cycle by hand (run-once: invoke the watch command by hand once, no interval, dry-run, no source edits; normal trigger: README: Without --interval, a run makes one pass. Interval loops are not used here (no timer installed).): 03:46 0 in scope · 0 reviewed · 0 sent · 0 would send · 0 need you

    Source: sidecarwatch/watch.py

    Input: watch --only <guest-path> --dry-run (no --interval)

    Expected: one pass summary line, then process exits 0

    Command: <guest-path> watch --only <guest-path> --dry-run

    Observed: 03:46 0 in scope · 0 reviewed · 0 sent · 0 would send · 0 need you

Local or hosted: local

What the agent did

Run the sidecarwatch offline synthetic demo (`sidecarwatch demo`), plus status, one-pass watch, doctor, and the unit test suite

pwd; id -u; uname -a; ls -la; cat /etc/os-release | head -3; ls <guest-path>; ls -la <guest-path>; cat <guest-path> 2>/dev/null | head -200
/opt/homebrew/bin/python3 --version; which uv python3; uv --version 2>&1; ls <guest-path>; cd <guest-path>; sed -n 200,400p README.md
cd <guest-path> && git --version && (git init -q 2>&1; git remote add origin https://github.com/TayHCode/sidecarwatch 2>&1; timeout 120 git fetch --depth 1 origin 8ddf4cfed6d91ad621ad9b0fcc7bf4c4c39901fe 2>&1; git reset --mixed FETCH_HEAD 2>&1) ; git log --oneline -1 2>&1; git status --short 2>&1 | head -20
cd <guest-path> && PYTHONDONTWRITEBYTECODE=1 /opt/homebrew/bin/python3 -B -m sidecarwatch demo; echo "exit=$?"; /opt/homebrew/bin/python3 -B -m sidecarwatch --version; echo "exit=$?"
cd <guest-path> && git fetch --depth 1 origin 8ddf4cfed6d91ad621ad9b0fcc7bf4c4c39901fe 2>&1; echo "fetch=$?"; git reset --mixed FETCH_HEAD 2>&1; git log --oneline -1 2>&1; git status --short | head -5
cd <guest-path> && cat .gitignore && /opt/homebrew/bin/python3 -m venv .venv 2>&1; echo "venv=$?"; . .venv/bin/activate && python3 --version && python3 -m pip install . 2>&1 | tail -5; echo "pip=$?"; which sidecarwatch; sidecarwatch --version
cd <guest-path> && PYTHONDONTWRITEBYTECODE=1 .venv/bin/python3 -B -m unittest discover -s tests -t . -q 2>&1 | tail -6; echo "tests=${PIPESTATUS[0]}"; .venv/bin/sidecarwatch doctor --shapes 2>&1 | head -40; echo "doctor=$?"; .venv/bin/sidecarwatch sessions 2>&1 | head; echo "sessions=$?"
cd <guest-path> && .venv/bin/sidecarwatch status 2>&1; echo "status=$?"; grep -rn "max.passes\|max_passes\|interval" sidecarwatch/*.py | head -30; echo ---; .venv/bin/sidecarwatch watch --only <guest-path> --dry-run --max-passes 1 2>&1; echo "watch=$?"
cd <guest-path> && sed -n 246,256p sidecarwatch/watch.py; echo ---; time .venv/bin/sidecarwatch watch --only <guest-path> --dry-run; echo "watch=$?"; grep -n "one pass" README.md | cut -c1-200; grep -n "Run the test suite\|To run the test suite" README.md
cd <guest-path> && mkdir -p .coldpass && cat > .coldpass/probe.json <<'EOF'
{"kind":"cli","argv":["<guest-path>","demo"],"cwd":"<guest-path>","expected_text":"DEMO complete: temporary sessions removed; nothing sent."}
EOF
cat > .coldpass/claims.json <<'EOF'
[
{"kind":"claim","claim":"demo runs offline on synthetic sessions","source":"`demo` uses temporary synthetic sessions, calls no agent CLI or model, and removes its data.","input":"sidecarwatch demo","expected":"synthetic demo completes with nothing sent","probe_kind":"cli","argv":["<guest-path>","demo"],"expected_text":"DEMO complete: temporary sessions removed; nothing sent.","cwd":"<guest-path>"},
{"kind":"claim","claim":"status shows cross-vendor review state","source":"- `status` shows whether cross-vendor review is ON or OFF.","input":"sidecarwatch status","expected":"reports cross-vendor review OFF by default","probe_kind":"cli","argv":["<guest-path>","status"],"expected_text":"Cross-vendor review: OFF","cwd":"<guest-path>"},
{"kind":"claim","claim":"watch without interval makes one pass","source":"Without `--interval`, a run makes one pass.","input":"watch --only <guest-path> --dry-run","expected":"single pass summary printed, then exit","probe_kind":"cli","argv":["<guest-path>","watch","--only","<guest-path>","--dry-run"],"expected_text":"0 in scope","cwd":"<guest-path>"}
]
EOF
cat > .coldpass/examples.json <<'EOF'
[
{"kind":"example","claim":"synthetic demo quickstart with system Python 3.14","source":"README.md Quickstart: Linux","input":"python3 -B -m sidecarwatch demo","expected":"offline demo output","argv":["/opt/homebrew/bin/python3","-B","-m","sidecarwatch","demo"],"cwd":"<guest-path>","env":{},"expected_text":"DEMO complete: temporary sessions removed; nothing sent."},
{"kind":"example","claim":"doctor reports local setup and review OFF","source":"README.md Check your own setup","input":"doctor --shapes","expected":"local setup report","argv":["<guest-path>","doctor","--shapes"],"cwd":"<guest-path>","env":{},"expected_text":"Cross-vendor review: OFF"},
{"kind":"example","claim":"documented unit test suite passes","source":"README.md Check your own setup (To run the test suite from source)","input":"python3 -B -m unittest discover -s tests -t . -q","expected":"all tests pass","argv":["<guest-path>","-B","-m","unittest","discover","-s","tests","-t",".","-q"],"cwd":"<guest-path>","env":{"PYTHONDONTWRITEBYTECODE":"1"},"expected_text":"OK (skipped=1)"}
]
EOF
cat > .coldpass/cycles.json <<'EOF'
{"detected":true,"checks":[{"claim":"watch runs one pass and stops when no interval is set","source":"sidecarwatch/watch.py","quote":"if consent_needed or interval is None or args.max_passes and passes >= args.max_passes:","schedule":"README: Without --interval, a run makes one pass. Interval loops are not used here (no timer installed).","method":"run-once: invoke the watch command by hand once, no interval, dry-run, no source edits","input":"watch --only <guest-path> --dry-run (no --interval)","expected":"one pass summary line, then process exits 0","kind":"cli","argv":["<guest-path>","watch","--only","<guest-path>","--dry-run"],"cwd":"<guest-path>","expected_text":"0 in scope"}]}
EOF
for f in probe claims examples cycles; do /opt/homebrew/bin/python3 -I -c "import json,sys;json.load(open(sys.argv[1]));print('ok',sys.argv[1])" .coldpass/$f.json; done
cd <guest-path> && COLDPASS_WORKSPACE=<guest-path> /opt/homebrew/bin/python3 -I <guest-path> --validate --replay; echo "checker=$?"
cd <guest-path> && cat > .coldpass/report.json <<'EOF'
{"outcome":"working","fault_owner":null,"operation":"Run the sidecarwatch offline synthetic demo (`sidecarwatch demo`), plus status, one-pass watch, doctor, and the unit test suite","operation_scope":"macOS arm64 guest (Darwin 24.6), Python 3.14.8 from Homebrew (README requires 3.11+), venv at .venv, no agent CLIs, no login, cross-vendor review OFF, no network calls to reviewers. Pinned commit 8ddf4cf.","steps":["git init; git remote add origin https://github.com/TayHCode/sidecarwatch; git fetch --depth 1 origin 8ddf4cfed6d91ad621ad9b0fcc7bf4c4c39901fe; git reset --mixed FETCH_HEAD","python3 -B -m sidecarwatch demo (from source folder; README Quickstart)","/opt/homebrew/bin/python3 -m venv .venv (README alternative to uv)","source .venv/bin/activate && python3 -m pip install .","sidecarwatch --version","python3 -B -m unittest discover -s tests -t . -q (README: Check your own setup)","sidecarwatch doctor --shapes; sidecarwatch sessions; sidecarwatch status","sidecarwatch watch --only <guest-path> --dry-run"],"deviations":["Guest is macOS, not Linux; used the README's macOS guidance where it applies.","uv is not installed in the guest. Used the README's documented `python3 -m venv .venv` route with Homebrew Python 3.14 (>=3.11) instead of `uv venv --python 3.12`. The README's macOS table lists uv 3.12 as the tested route; 3.14 was not the tested interpreter.","`timeout` is not available on macOS, so the first git fetch command failed with `timeout: command not found`; re-ran the same fetch without it and it succeeded.","`sidecarwatch init` was not run, so no config or profile was created and review stays OFF.","No real Claude Code or Codex sessions exist in this VM, so `sessions` is empty and `doctor` reports both CLIs unavailable; this is expected and not a product fault.","Unit test suite ran with `.venv/bin/python3` (absolute venv path) rather than an activated `python3`. Same command otherwise."],"stuck_step":null,"main_blocker":null,"plain_failure":null,"suggested_fix":null}
EOF
/opt/homebrew/bin/python3 -I -c "import json;d=json.load(open('.coldpass/report.json'));print(d['outcome'])"; COLDPASS_WORKSPACE=<guest-path> /opt/homebrew/bin/python3 -I <guest-path> --validate --replay | head -c 300; echo; echo "checker=$?"; ls -la .coldpass; git status --short | head
cd <guest-path> && COLDPASS_WORKSPACE=<guest-path> /opt/homebrew/bin/python3 -I <guest-path> --validate --replay > /tmp/cp_check.out 2>&1; echo "checker=$?"; /opt/homebrew/bin/python3 -I -c "import json;d=json.load(open('/tmp/cp_check.out'));print(d['valid'],d['problems'],d['replay']['status'])"