author-contract

Author contract (all write-capable agents)

Subagents can't auto-load skills — Read this file and your <domain>-standards skill before editing.

Model tier: authors default to Sonnet. Flag genuinely-beyond-Sonnet work (deep Rust concurrency/unsafe, a new provider's streaming protocol, a schema/data migration) up front so the orchestrator re-spawns you on Opus instead of thrashing.

Implement like a lazy senior dev

• Smallest diff per issue. Preserve behavior and public/package APIs. One concern per edit.
• Rust-first for business logic/pipelines/ATS/documents; the renderer stays presentational.
• Reuse before adding — an existing service hook, @ajh/ui primitive, registry, or helper beats new code; under-abstraction beats the wrong abstraction.
• Never reformat untouched lines; never rename across package boundaries unprompted.
• One-line plan before a large multi-file refactor, then pause for confirmation; small in-file fixes proceed.

Ground first

• Codegraph FIRST — hard rule. Query codegraph (MCP codegraph_explore when allowed, else the CLI) before ANY raw Grep/Read; raw reads are for what codegraph can't answer (this turn's un-indexed edits, non-code assets, config prose).
• Read the handoff's ## Current state (.claude/scratch/<task>.md) — never cold re-explore what it already contains (the ## Log below it is steward-only).
• Priority: codegraph/graphify → source → docs/knowledge → lessons. No repo-wide scans; stop at ~90% confidence.

You never approve your own work

Implement, then hand the diff to your independent sibling critic (and the test pair). Resolve every HIGH/CRITICAL before "done"; LOW/MEDIUM are advisory. Append what you changed (files, decisions, open questions, Lessons-to-propose) to the handoff ## Log, then rewrite the stale parts of ## Current state (≤2K chars).

Leave a check behind (missing tests BLOCK)

Non-trivial logic (a branch, loop, parser, money/security/error path) ships one runnable check (via test-author); trivial one-liners don't (YAGNI). A missing test — or a weak / tautological / mock-asserting one — is a HIGH (blocking) finding for your critic. Cover the error/edge path, not just the happy path. New/changed user-facing text needs its i18n key in both en and de (also HIGH). Hermetic cross-OS tests per testing-rules: inject dirs / a temp HOME, #[serial_test::serial] on env-mutating tests, no real network, never assume a system binary is absent, never reach an exec()-replacing path in-process.