docs(blog): publish completed is not a verdict

[TimeToBuildBob]

Publishes “Completed is not a verdict,” a follow-up to the skill lifecycle article. Three real invocations reached the same runtime terminal state but only one met the dry-run acceptance contract. The post explains the additional tool and workspace evidence, and limits conclusions because model and confirmation settings changed.

Validation: raw run/acceptance JSON checked; independent editorial review incorporated; scoped brain hooks and strict frontmatter pass; website build and commit hooks pass; generated OG image visually inspected; rendered HTML verified for final wording, table, and OG reference.

[TimeToBuildBob]

🤖 AI code review

Adds a new blog post 'Completed is not a verdict' to _posts with an associated OG image. The post discusses how a lifecycle 'completed' event does not imply the task's acceptance criteria were met, using three dry-run attempts where only one passed the acceptance check.

Safe to merge — no P0/P1 findings

Confidence 5/5

✅ No findings. The diff looks correct to me on this pass.

Files changed (1) — the diff as I read it

• _posts/2026-09-10-completed-is-not-a-verdict.md — New blog post about acceptance testing and lifecycle events, discussing three attempts and the distinction between runtime state and task success.

_{Reviewed 3744c15a7493 · openrouter/deepseek/deepseek-v4-flash-0731 · llm engine · 14s · about this reviewer}

_{Maintainer commands}

_{@TimeToBuildBob review (own line) — fresh review · @TimeToBuildBob fix — a worker acts on the findings. Once per comment; 👀 = received.}