09.03.06 — Author skills#

Skills that produce a deliverable. Tests, probes, docs, reports.

Listing (potentially non-exhaustive)#

empiricism#

input  : user intent ("write a test that asserts X")
output : before encoding X:
            1. empirically verify that X is true
                - probe
                - gh api on the source code
                - curl -v to inspect HTTP responses
                - ...
            2. if X confirmed → encode
            3. if X false → correct intent
            4. if X ambiguous → ask the human

“Fair point, I’m assuming. Let me verify empirically.”

write-a-probe#

input  : question to investigate ("does the confirmation arrive in < 5s?")
output : throwaway Python script that:
            - lives in a gitignored directory (typically: `probes/`)
            - bypasses the Ocarina workflow (no Test, no Suite)
            - drives the browser or makes HTTP calls directly
            - prints state (URL, DOM, cookies, timings) to stdout
            - is never committed
         after use: finding goes to IDENTIFIED_GAPS.md / FRD / comment
         then the probe is deleted

CLAUDE.md:

A probe is a one-off script that drives the browser (or raw HTTP) through a suspect flow and prints concrete runtime state. It bypasses the Ocarina workflow entirely — no create_selenium_test, no suites, no campaigns, no assertions. Probes live in a gitignored directory, are never committed or pushed, and are deleted once the answer lands in a durable artifact.

And:

Reach for one when the framework’s error surface doesn’t show enough — a bare TimeoutException or AssertionError you can’t act on. The trigger is the visibility gap, not where you are in the test lifecycle.

write-test-strategy#

input  : test heritage, test base
output : a test-strategy .md containing:
            - scope
            - test types
            - coverage tables (REQ-X-N × test_X)
            - suite/campaign tree
            - expected pass/fail breakdown
            - gaps
            - CI matrix

Output: CURA_TEST_STRATEGY.md

extend-coverage#

input  : a gap / an attack idea found by a black-hat skill
output : a new test that exercises this gap:
            - via the normal UI (no payload)
            - asserts rejection if expected (gap test)
            - asserts current behavior if exploratory
            - hooks into an FRD update

update-frd-and-tests#

input  : spec change ("CURA fixed G-DATA-1")
output : atomic diff:
            - FRD §9.7: adds revision ("2026-XX-XX: fixed")
            - IDENTIFIED_GAPS.md: adds revision
            - test: invert assertion (FAIL → PASS), rename, move if needed
            - analyze every inline comment referencing the old gap

The only skill that modifies the spec it’s given as reference.

Don’t rewrite the spec; only update-frd-and-tests does, with a revision line.

manual-reproduction-guide#

input  : a failing test
output : a step-by-step guide for a human to reproduce the failure manually

1. Open https://katalon-demo-cura.herokuapp.com/profile.php#login
2. Type "John Doe" in username field
3. Type "ThisIsNotAPassword" in password field
4. Click Login
5. Navigate to /appointment.php
6. ...
7. Observe: confirmation page rendered with past date (expected: rejection)

manage-backlog#

input  : work sessions / discussions
output : maintains BACKLOG.md:
            - findings to investigate later
            - suggested improvements
            - tests to write
            - open questions on the SUT
         prioritization

pr-report#

input  : an update to push as a PR
output : a PR report adapted to the change type:
            - feature PR: tests added, coverage diff
            - bugfix PR: regression test, link to the gap
            - refactor PR: checks behavior hasn't changed
            - doc PR: no tests
            - mixed PR: multiple sections

Produces a coherent report in gh pr create --body ... format.

Cross-cutting discipline#

All Author skills produce something persistent, deliverables: test, doc, file.