jrvs-skills
01Plugin Claude Code
Plugin marketplace listing 2 plugins: tmux, scram.
Plugin Claude Code
Plugin marketplace listing 2 plugins: tmux, scram.
Skill Claude CodeCodex
Ingests open retro issues, investigates via scramstorm, dispatches scram team to address changes, bumps version, comments on issues, commits and pushes.
Instructions file
Claude Code instructions for alxjrvs/skills, covering claude.md, what this repo is, validation, adding a local plugin and adding an external plugin.
Plugin Claude Code
SCRAM — Structured Collaborative Review and Merge with stream-based parallel development.
Agent
Senior-level architect focused on structural harmony, DRYness, and codebase-wide patterns. Reviews for how changes fit the whole, identifies emergent patterns, and guards against architectural drift. Default model sonnet.
Agent
Adversarial critic who stress-tests ideas, finds flaws, and challenges assumptions. Cynical, harsh, and relentless. Summoned when you need someone to tear an approach apart before shipping it. Default model sonnet.
Agent
Design specialist for user-facing elements (GUI, TUI). Writes design-related ADRs, reviews UI/UX stories as required merge approver, and provides design guidance. Default model sonnet, scalable to opus for complex design work.
Agent
Dev tooling and DX specialist focused on CI/CD pipelines, build systems, agentic integrations, developer workflows, and toolchain health. Optional role for projects with significant tooling concerns. Default model sonnet.
Agent
Developer for G3 story breakdown and context brief authoring. Sizes stories, tags complexity, writes context brief files to SCRAMWORKSPACE/briefs/. Read-only on the repo — writes only to SCRAMWORKSPACE.
Agent
Developer for TDD implementation in isolated worktrees during concurrent streams. Follows strict Red-Green-Refactor discipline. Default model sonnet, scalable to opus based on story complexity.
Agent
Developer reviewer for G1 ADR review and G2 doc review. Evaluates docs for feasibility, testability, ambiguity, and architecture fit. Read-only — writes review reports to SCRAMWORKSPACE only.
Agent
Writes docs-as-spec before implementation (feature docs, ADRs, plan cleanup), then refines docs incrementally as stories merge. Flags significant doc-code divergence to maintainers.
Agent
Marketing specialist focused on reach, SEO, copy quality, and discoverability. Reviews user-facing text for persuasiveness, clarity, and search optimization. Default model sonnet.
Agent
Detail-oriented reviewer focused on code quality, correctness, and strict adherence to story acceptance criteria. Reviews line-by-line for bugs, test coverage, style compliance, and scope creep. Default model sonnet.
Hook
Runs before the agent uses a tool for Agent, Write and Bash tool calls, executing halt-check.sh, brief-lint.sh and merge-guard.sh via bash (3 commands). From alxjrvs/skills.
Hook
Runs after a tool call finishes for Agent tool calls, executing isolation-check.sh via bash. From alxjrvs/skills.
Hook
Runs when the agent finishes a response, executing session-checkpoint.sh via bash. From alxjrvs/skills.
Plugin Claude Code
SCRAM context brief format reference sub-skill.
Skill Claude CodeCodex
Context brief format reference for SCRAM — brief template, complexity tagging, resolution modes, prioritization table, and backlog format.
Plugin Claude Code
SCRAM escalation handling reference sub-skill.
Skill Claude CodeCodex
Escalation handling reference for SCRAM — failure taxonomy, escalation path, and escalation brief format.
Skill Claude CodeCodex
Standalone retrospective facilitator for SCRAM runs. Receives workspace context, reads in-flight.md if it exists, dispatches maintainers for ticket writing and discussion, compiles results, and presents consensus changes.
Plugin Claude Code
SCRAM session lifecycle reference sub-skill.
Skill Claude CodeCodex
Session lifecycle reference for SCRAM — manifest format, workspace structure, state update events, and context recovery.