Octivi Bash Boilerplate

Use this skill for any Bash script task that should follow OBB conventions.

Inputs Required

• Script intent and lifecycle (throwaway helper vs maintained CLI).
• CLI contract: required/optional flags, positional args, env vars.
• Output contract: stdout/stderr format, exit codes.
• Side effects: files, network, permissions, cleanup requirements.

If any of these are missing, infer minimally and state assumptions.

Decision Table

Choose exactly one mode per script:

Execution Steps

1. Read references/workflow.md and apply the flow for new/refactor/migration.
2. Copy one ready template to destination:
   cp skills/octivi-bash-boilerplate/assets/templates/<header-only-script|full-obb-script-source|full-obb-script-embedded> ./script.sh.
3. Implement business logic outside OBB marker blocks.
4. If a marker block is unpopulated or stale, run:
   octivi-bash-boilerplate-update <script>.
5. Run quality gates from references/checklist.md and resolve all required failures.

Output Contract

The final response must include:

• Selected mode and short rationale.
• What changed (CLI, behavior, compatibility expectations).
• Commands executed for validation.
• Residual risks and follow-up actions, if any.

Template Source Of Truth

• Canonical OBB scripts live at repository root:
  • octivi-bash-boilerplate
  • octivi-bash-boilerplate-header
  • octivi-bash-boilerplate-update
• Do not edit assets/templates/* marker payloads manually; regenerate with:
  scripts/generate-skill-templates.

Reference Loading Strategy

• Primary workflow: references/workflow.md.
• Final gate and pass/fail rules: references/checklist.md.

Non-negotiable rules

• Keep marker syntax exact:
  # >>> OBB:BEGIN variant=header|full and # <<< OBB:END.
• Keep all business logic outside OBB marker blocks.
• For full embedded mode, keep variant=full marker block at end of script.
• Never source octivi-bash-boilerplate-header as a library file.
• Use one template variant per script and avoid mixing integration styles.