temper

Quick Start: /plugin marketplace add galando/temper then /temper "add password reset" — one command for the full pipeline.

The Problem

AI writes code fast. But "fast" without "right" creates bugs, technical debt, and features that miss the point. AI-generated code has structural failure patterns:

Pattern	What Goes Wrong
Missing behaviors	Happy path works, edge cases never implemented
Wrong problem solved	Feature works perfectly, but nobody asked for it
Over-engineering	Factories and strategies for something used once
Hallucinated APIs	Methods called that don't exist
Missing wiring	Code correct, integration missing

Most AI tools check if code compiles. Temper checks if it solves the right problem, handles edge cases, and is safe to ship.

The Catch

The rate-limiting bug that vanilla AI always misses:

Scenario: Rate limiting on reset requests
  Given a user has requested 3 resets in 10 minutes
  When they request another reset
  Then the request is rejected with 429

AI built password reset. All tests pass. But Temper's scenario coverage gate caught the gap: no test for rate limiting. Build wrote the test. Test failed. Build implemented rate limiting. Test passed. Without the coverage gate, rate limiting would never have been implemented.

More: Evidence Gallery

How It Works

Three methodologies, one contract file (intent.md):

intent.md
|
+-- Intent (IDD)           WHY are we building this?
|   Problem, success criteria, constraints
|
+-- Scenarios (BDD)        WHAT should it do?
|   Gherkin Given/When/Then, derived BEFORE architecture
|
+-- /temper:build (TDD)    HOW do we build it?
    Tests from scenarios, RED → GREEN → REFACTOR

Key insight: Scenarios are derived before architecture. The file plan follows from what the system must do, not the other way around. This prevents over-engineering structurally.

Full methodology: docs/methodology.md

Autonomous Continuation (opt-in) (v6.0.0)

After you approve the plan, /temper can run the remaining stages (design → build → review → check → eval) unattended and leave a report. It never pushes or merges, never re-plans on its own, and parks before commit and on anything needing a human. Turn it off and Temper behaves exactly as before.

First run: pre-allow your build/test commands in settings.json, or the run parks on the first unpermitted command. No config yet? /temper:init seeds one. Details: docs/proposals/autonomous-mode.md

Token Efficiency (v5.9.0)

Temper cuts a run's token cost with three optimizations — all ON by default, all revert to v5.8.0 when their flag is off:

Optimization	Default	Quality tradeoff
Cache static methodology reads	on	None — same reads, ~90% off re-read cost
Adaptive depth — size the pipeline to the change	on (floor: simple)	Only one with a tradeoff — see below
Incremental loops — inline auto-fix, no subprocess	on (inline-threshold: 3)	None for auto-fixable findings

The one decision you make: adaptive-depth runs a lighter pipeline on small changes. It's safe when the change is genuinely isolated — but if it touches a shared interface, auth, money, or has an unclear blast radius, click "Escalate to full pipeline" at the plan gate (one click, that change only — no config editing). For a whole high-stakes codebase, set floor: medium in .claude/temper.config once and never think about it again.

Full explanation (mechanism, tiers, decision rule, when quality drops): docs/token-efficiency.md

Commands