Plugin Claude Code
A lean, full-SDLC Claude Code harness — Plan → Implement → Verify → Maintain — with subagent orchestration, memory persistence, and an evolution path toward eval loops, retrieval, and multi-agent systems.
Plugin Claude Code
Lean full-SDLC Claude Code harness — Discover, Plan, Implement, Verify, Maintain — with anti-rationalization guards on the gate skills, portable skills (installable anywhere via the skills CLI), researched git conventions (branch-at-first-write, Conventional Commits, detect-and-follow), always-on pipeline routing, a…
MCP server Claude CodeCodexCursor
Gives the agent a persistent knowledge graph it can write facts to and recall across sessions. Runs locally from the @modelcontextprotocol/server-memory npm package.
MCP server Claude CodeCodexCursor
Adds a structured step-by-step reasoning tool the agent can call to break a problem into revisable thoughts. Runs locally from the @modelcontextprotocol/server-sequential-thinking npm package.
MCP server Claude CodeCodexCursor
Official MCP server for Magic UI. Runs locally from the @magicuidesign/mcp npm package.
Instructions file
Claude Code instructions for vasuag09/harness-claude, covering harness-claude, the pipeline, what's here, operating principles and release docs sweep (this repo).
Agent
System-design specialist. Use PROACTIVELY for architectural decisions, new subsystems, interface/data-model design, or significant refactors. Produces a design note / ADR with trade-offs. Read-only — designs, does not implement.
Agent
Build and type-error resolution specialist. Use PROACTIVELY when a build fails or type/compile errors appear. Fixes build/type errors with minimal diffs — no architectural changes. Gets the build green fast.
Agent
Expert code-review specialist. MUST BE USED immediately after writing or modifying code, before merge. Reviews quality, correctness, and maintainability. Read-only — reports findings with severity, does not edit.
Agent
Long-running-agent operator. Use to drive ONE iteration of an unattended /operate run — advance the task by a single increment, then checkpoint. Enforces the guardrails via step.js and never continues past a halt. Honors the git boundary. May edit code.
Agent
Implementation-planning specialist. Use PROACTIVELY for non-trivial features, refactors, or anything spanning multiple files. Turns a spec into a phased, risk-assessed task plan. Read-only — produces a plan, does not edit code.
Agent
Dead-code and cleanup specialist. Use PROACTIVELY during the Maintain phase to remove unused code, duplicates, and tech debt. Makes safe, behavior-preserving changes only. May edit code.
Agent
Security vulnerability specialist. MUST BE USED for any change touching auth, user input, DB queries, file I/O, external calls, crypto, secrets, or payments — and before merging such code. Scans for OWASP Top-10 and secret leakage. Read-only — reports, does not edit.
Agent
Test-driven development specialist. Use PROACTIVELY when implementing a feature or fixing a bug. Enforces tests-first (RED → GREEN → REFACTOR) and ≥80% coverage. May write tests and implementation.
Hook
Runs before the agent uses a tool for Bash, Grep, Bash and Grep tool calls, executing pre-tool-trace.js, pre-bash-tmux-reminder.js, pre-search-mgrep-nudge.js and pre-search-cbm-orient.js via node (4 commands). From vasuag09/harness-claude.
Hook
Runs after a tool call finishes for Edit, Edit, Edit, Write, MultiEdit, Edit, Write, MultiEdit, Edit, Write, MultiEdit and Bash tool calls, executing post-edit-prettier.js, post-edit-tsc.js, post-edit-quality-gate.js, post-edit-design-quality.js, post-edit-strategic-compact.js and post-bash-build-analysis.js via node…
Hook
Runs when you submit a prompt, before the agent sees it, executing user-prompt-routing.js via node. From vasuag09/harness-claude.
Hook
Runs when a session starts, executing session-start.js and session-lazy-activate.js via node (2 commands). From vasuag09/harness-claude.
Hook
Runs before the context is compacted, executing pre-compact-save.js via node. From vasuag09/harness-claude.
Hook
Runs when the agent finishes a response, executing stop-session-summary.js and stop-pattern-extraction.js via node (2 commands). From vasuag09/harness-claude.
Skill Claude CodeCodex
Make and record an architecture/design decision for non-trivial work — new subsystems, interfaces, data models, or significant refactors. Use when /plan flags a load-bearing decision. Delegates to the harness-claude:architect agent.
Skill Claude CodeCodex
Measure whether a harness component (a skill, agent, or rule) actually earns its keep — run the same task k times with and without it, in isolated git worktrees, and report pass@k (works once) and pass^k (works consistently). Opt-in; not wired into the default pipeline. Use to decide keep-vs-cut on evidence, or to…
Skill Claude CodeCodex
Resolve build failures and type/compile errors fast, with minimal diffs. Use when a build or typecheck fails during implementation. Delegates to the harness-claude:build-error-resolver agent.
Skill Claude CodeCodex
Release step that closes the loop after /ship — drive a change out to a running environment using the PROJECT'S OWN deploy mechanism (detect, never prescribe a stack), then smoke-test the deployed artifact and keep a guarded rollback path. Deploy is more outward-facing than git push, so it is arm-to-fire — plan and…