Skip to content

Development

m4bwav edited this page Nov 21, 2026 · 1 revision

Layout

Path What
.claude-plugin/plugin.json, marketplace.json plugin and marketplace manifests (the version lives in plugin.json)
skills/techwright/SKILL.md the skill: rules, commands, routing
skills/techwright/references/ troubleshoot.md, network.md, backup.md, buying.md, homelab.md
scripts/tech.py the read-only checks, standard library only
tests/test_tech.py unit tests on fixture text
evals/evals.json, evals/fixtures/ skill eval cases with no-skill baselines
evergreen.json, RESEARCH.md, LEARNINGS.md, TESTS.md, CHANGELOG.md evergreen upkeep: refresh schedule, research, lessons, test runs

Tests

python -m pytest

or python -m unittest discover -s tests. The parsers (netsh, ipconfig, ping on Windows and POSIX), the network verdicts, the disk and health findings, the backup table and the restore test run on fixture text, so the tests pass on any system. 0.1.1 has 12 tests. There is no CI; run them before a pull request.

How the script is built

Each command collects raw output with run(), which never raises (a missing program or a timeout comes back as code -1 with the error text), and hands it to a pure function: parse_wlan_interfaces, parse_ipconfig_gateways, parse_ping, net_verdict, disk_findings, backup_table, restore_test, health_findings. Those functions are what the tests call. PowerShell queries are kept in the DISK_PS, BACKUP_PS and HEALTH_PS strings and return JSON.

To add a known event, add its "Provider, Id" key and a plain meaning to KNOWN_EVENTS, and a test case in tests/test_tech.py.

Evals

evals/evals.json holds five cases (restore, 3-2-1, battery, router, prices), each run with the skill and without it. Results are recorded in TESTS.md.

Releasing

Bump version in .claude-plugin/plugin.json, add a CHANGELOG.md entry, merge, tag vX.Y.Z on main, and create the GitHub release. Then update this wiki's Versions and upgrading page and the footer.

Clone this wiki locally