skill: make the coding agent orchestrate Rook instead of exploring it…

…self (#45)

When asked to explore or plan tests for an agent, Claude Code invoked the
skill but often read the target's code and wrote its own test plan rather
than running `rook explore` and `rook generate`. The skill never stated the
agent's role, and its spending guidance made free manual analysis look like
the safer choice.

Add a "Your role" section: the coding agent chooses and runs Rook commands,
answers from Rook's agents, scenarios and verdicts, reads target source only
for steps Rook needs, and names a blocked command instead of substituting
its own analysis. Mirrors resynced.

Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

Merge pull request #32 from LambdaTest/standalone-industry-demos

Make industry samples standalone and preserve full Rook runs

Update rook to 0.1.4

Merge pull request #26 from samyakLambda/fix/triage-service-hardening

Harden the triage-service sample listener and body handling

Add bottle block for rook 0.1.2

Add bottle block for rook 0.1.1

Add bottle block for rook 0.1.0

feat(ci): harden brew-smoke.yml, add Intel validation workflow

Reference: homebrew-kane's validate-brew-install.yml pattern. Adds a version-drift guard job before the post-release matrix runs, a bottle-vs-source-build assertion, and reuses Formula/rook.rb's own test do block (via brew test) for the bundled-runtime-binary check instead of re-deriving that logic in bash. New brew-smoke-intel.yml validates the never-bottled darwin-x64 path in its own manually-dispatched workflow, matching homebrew-kane's rationale for keeping Intel out of the auto-triggered post-publish gate. Closes #5.