DD-02 — Aider: The Git-Native Pair Programmer

Aider: The Git-Native Pair Programmer

~8 tools, ~2k prompt, ~25k LOC, git-gated. The thin-medium reference — and the field's benchmark creator. Three years of iteration refined around one bet: git-as-substrate.

minutes

artifacts

32/60

rubric score

Aider's git-first philosophy is the differentiator. One infrastructure choice — using git as the substrate for everything — yields five rubric modules: state, memory, verification, permission, observability. No other harness in the roster gets this much from a single decision. This is the +7 over Pi. But git buys capability, not safety: git-gating catches wrong changes, not malicious ones.

Load-bearing claims

Git-as-substrate is the multiplier

Aider inherits state (M8), memory (M4), verification (M9), permission (M6), and observability (M11) from git — five modules from one infrastructure choice. The entire +7 over Pi comes from this single bet. Module 8 (State) alone gains +3, the largest single-module delta in the deep-dive roster.

The benchmark creator

Aider created the Aider Polyglot Benchmark — the de facto standard for evaluating coding agents across languages. Most published coding-agent benchmark numbers trace to Aider's methodology. Aider is the reference because it defined how the field measures "best," not because it has the highest raw rubric score.

Git-gating catches wrong, not malicious

Aider's permission model reviews change quality, not change safety. A prompt-injected model produces a benign-looking "refactor" that buries an exfiltration payload. The human diff-review approves a valid-looking change — the exfil executes at tool-execution time, before the commit is reviewed. Git-gating is downstream of the damage.

After this module

Apply the 6-phase methodology to Aider and produce a scored card with file:line evidence.

Explain Aider's git-first philosophy as a distinct architectural bet — git-as-substrate yielding five modules from one choice.

Compare Aider (thin-medium) against Pi (thin) module-by-module: the +7 comes from git.

State why Aider's benchmark creation makes it the field's reference for coding-agent performance.

Articulate the critical limitation: git-gating catches wrong changes, not malicious ones.

Identify the three changes that move Aider from "mature" to "production-grade" without compromising git-first value.

Artifacts

Teaching Document

6-phase deep-dive + 12-module scoring + anti-patterns + key terms

READ

Diagrams

Git-as-substrate hub-and-spoke, git-gated loop, security nuance two-track, thickness spectrum

READ

Slide Deck

12 reveal.js slides — thesis, git-as-substrate, 12-module audit, security nuance, anti-patterns

VIEW

Teaching Script

~2,800-word verbatim transcript with [SLIDE N] cues, ~140 wpm

READ

Flashcards

24 cards — recall, application, analysis (TSV with tags)

TEST

Exam

15 questions, Bloom 20/40/40, 70% pass, rationale per question

TEST

Lab Spec

4-phase Python lab — git-gated loop trace, +7 scoring, injection attack construction, three-fix defense matrix

BUILD

Module Web Page

This page — single-file HTML hub

HERE