poly-gents OpenAnt
Purpose
Execute OpenAnt CLI tasks safely, capture stdout/stderr, and return concise actionable findings and suggested follow-up commands.
When to Use
- User explicitly asks to run OpenAnt or refers to OpenAnt analysis.
- Interpreting OpenAnt output after a local or CI run.
- Exploring repo/analysis workflows documented for OpenAnt.
When NOT to Use
- Tasks better served by other CLIs or skills without OpenAnt context.
- Destructive operations without explicit user confirmation.
- Skill repo authoring → create-skill / publish-skill.
Expected Outcome
- Commands run (or exact commands provided if sandboxed) with captured output.
- Summary of findings, errors, and recommended next commands.
- Clear stop before destructive flags.
Inputs to Gather
- Requested OpenAnt subcommand or goal.
- Working directory and repo context.
- Flags that may be destructive (
--force, delete, overwrite). - Whether network or credentials are required.
Workflow
- Confirm
openanttool or shell access. - Construct command from user intent; avoid destructive flags unless approved.
- Run and capture stdout/stderr with exit code.
- Summarize actionable findings and anomalies.
- Suggest next commands; do not chain destructive steps without confirmation.
Domain guidance
Run OpenAnt for the task the user requested; capture stdout/stderr; summarize actionable findings and next commands. Do not run destructive commands without explicit confirmation.
- Prefer dry-run or read-only subcommands when available.
- Quote paths with spaces; run from repo root unless specified.
- Never fabricate CLI output.
Tool Availability Rules
| Access | Behavior |
|---|---|
| Full tool access | Execute OpenAnt and report results. |
| Read-only | Output exact command for user to run. |
| No shell | Provide command template only; no fabricated output. |
Related tool sets
openant
Review / Decision / Execution Criteria
- Include exit code and key log lines in summary.
- Flag destructive operations before execution.
Output Format
- Command(s) executed.
- Exit status.
- Findings (bullets).
- Suggested next commands.
- Warnings or blockers.
Quality Bar
- Concise operator summary, not full log dump unless requested.
Safety and Boundaries
- No destructive commands without explicit user approval.
- Do not fabricate OpenAnt output or success.
Escalation / Dispatch Rules
- Implementation follow-up → implement-features, refactor-code.
References
skills/old_skills.json(openant).skills/skill.instruction.md,skills/meta.instructions.md