Impl Sonnet Medium — Development agent for Claude Code
Standard plan implementation tasks - well-specified modules, validators, endpoints, and their tests.
How to install Impl Sonnet Medium
Installs to ~/.claude/agents/adiatmaja-agentic-workflow-impl-sonnet-medium.md
mkdir -p ~/.claude/agents && curl -fsSL https://raw.githubusercontent.com/adiatmaja/agentic-workflow/HEAD/agents/impl-sonnet-medium.md -o ~/.claude/agents/adiatmaja-agentic-workflow-impl-sonnet-medium.md Restart Claude Code, or start a new session, for it to be picked up.
What Impl Sonnet Medium does
name: impl-sonnet-medium description: Standard plan implementation tasks - well-specified modules, validators, endpoints, and their tests. model: sonnet effort: medium experimental: cacheTtl: 1h
Execute exactly one task from an implementation plan, following its steps verbatim. Where the task's risk tier owes a test (`~/.claude/rules/workflow.md`, "Tests: by RISK"; layout/copy/restyle work owes none and never pins display text), write the failing test first and confirm it fails for the
Alternatives in Development
- Job Agent — Creates idempotent, well-tested background jobs using Solid Queue with proper error handling and retry logic 652 ★
- Claudopro Directory — by ghost - Well-crafted, wide selection of Claude Code hooks, slash commands, subagent files, and more, coveri 195 ★
- Fixer Sonnet High — Fix-only agent, tier 2. Fixes known-cause findings that span modules or a seam, or known probe classes, with r 77 ★
Full documentation available on GitHub
View Source RepositoryRelated Agents
Default Worker
Main implementation worker (Sonnet). Use for normal, well-specified coding work — features, edits, refactors,
Medium
Implementer for standard feature work (Sonnet, high effort). Use for a well-specified ticket or vertical slice
Build Worker
Sonnet-tier worker for everyday coding subtasks: implement a well-specified function or component, fix a descr
S55 Verifier
Sonnet 5.5 verifier on medium, the default verifier. Use after every builder pass: checks the brief's acceptan
Exec Standard
MID tier at medium effort. Execution agent for well-bounded implementation in an isolated git worktree - new s
Impl Haiku Medium
DOCS-ONLY plan tasks - doc index edits, close-out records, open-item and handoff text. Never code, tests, regi
Related Skills
Standard
Execute the current task using STANDARD mode with short planning, complexity budget, implementation, tests, si
Agent Skills Discovery RFC
A mechanism for discovering Agent Skills using the .well-known URI path prefix as specified in RFC 8615 for di
Jules
Delegate coding tasks to Google Jules AI agent for async bug fixes, documentation, tests, and feature implemen