bug(skills): content soft-wrapped mid-sentence

[KemingHe]

What Happened?

Expected: Each output line is complete - no sentence or bullet wraps mid-sentence to a continuation line. Bullets use one line per bullet; if too long or multi-thought, split into separate distinct bullets. Prose uses one sentence per line.

Actual: Agents soft-wrap all output content mid-sentence to fit an arbitrary column width, affecting both bullets and prose. Example from github-issue skill output:

- Verify each `SKILL.md` has valid YAML frontmatter with `name` and `description` fields
  (all current skills already satisfy this requirement)

The parenthetical is a continuation of the first line, not a sub-bullet. Correct output is one of:

- Verify each `SKILL.md` has valid YAML frontmatter (name + description); all current skills satisfy this

or split into two distinct bullets:

- Verify each `SKILL.md` has valid YAML frontmatter with `name` and `description` fields
- Note: all current skills already satisfy this requirement

Steps to Reproduce

1. Use the github-issue skill to generate a feature request with a multi-point proposed solution section
2. Observe output - long bullets are soft-wrapped mid-sentence with two-space indented continuation lines

Environment

• Affected skills and versions (all lack a no-mid-sentence-wrap constraint):
  • github-issue v3.0.0
  • github-pull-request v3.0.0
  • meeting-memo v3.0.0
  • meeting-agenda v1.0.0
  • commit-message v2.0.0
  • readme v2.1.0
• Root cause: none of the six skill SKILL.md files include a rule against breaking bullet content across lines; agents default to prose wrapping behavior
• Fix: add a constraint to each affected skill's Constraints section: "Never wrap any output content mid-sentence. Each bullet is one line; if too long or multi-thought, split into separate distinct bullets. Each prose sentence is one line."