aidr

Offload context-heavy but low-complexity codebase work to Aider through a thin CLI wrapper. Use when another AI agent should avoid loading large repository context for tasks like discovery, repetitive refactors, cross-file version bumps, and broad search/explain passes. Supports safe read-only scanning, scoped edit runs, model-mode routing, and setup/model diagnostics.

Safety Notice

This listing is imported from skills.sh public index metadata. Review upstream SKILL.md and repository scripts before running.

Copy this and send it to your AI assistant to learn

Install skill "aidr" with this command: npx skills add pc-style/skills/pc-style-skills-aidr

aidr

Use aidr to delegate large-context, low-reasoning tasks to Aider.

Quick Start

bash scripts/setup.sh
aidr doctor
aidr models

Core Commands

  • scan <goal> [paths...]: run read-only discovery and return minimal relevant files/symbols.
  • patch <goal> [paths...]: run small scoped edits; defaults to dry-run unless --apply.
  • run <goal> [paths...]: run a universal freeform task; defaults to dry-run unless --apply.
  • models: list provider-visible Gemini models through aider --list-models gemini/.
  • doctor: check aider, uv, GEMINI_API_KEY, and active model mappings.

Modes

  • --mode fast: default model gemini-2.5-flash-lite.
  • --mode universal: default model gemini-3-flash-preview.
  • --mode advanced: default model gemini-3-flash-preview-2 with stronger reasoning hints.

Override defaults with:

  • AIDR_MODEL_FAST
  • AIDR_MODEL_UNIVERSAL
  • AIDR_MODEL_ADVANCED

Safe Workflow

  1. Start with aidr --mode fast scan "<goal>" [paths...].
  2. Apply repetitive edits with aidr --mode fast patch "<goal>" [paths...] --apply.
  3. Escalate mode only when results are insufficient.
  4. Use --aider-args "..." for extra Aider options when required.

Compatibility

Legacy commands remain supported and map to core commands:

  • map/find/retrieve/explain -> scan
  • refactor -> patch
  • task -> run

Source Transparency

This detail page is rendered from real SKILL.md content. Trust labels are metadata-based hints, not a safety guarantee.

Open in GitHubOpen in ClawHub

Related Skills

Related by shared tags or category signals.

General

handoff

No summary provided by upstream source.

Repository SourceNeeds Review
7-pc-style
Automation

self-subagent

No summary provided by upstream source.

Repository SourceNeeds Review
5-pc-style
General

skill-creator

No summary provided by upstream source.

Repository SourceNeeds Review
3-pc-style
Automation

adk-agent-extension

No summary provided by upstream source.

Repository SourceNeeds Review
3-pc-style