| name | lril |
| description | Second-brain workflow command suite for Claude Code, built around the PIV loop (Plan → Implement → Validate) from the Dynamous claude-code-second-brain workshop. Installs 19 slash commands under the `lril:` namespace covering project priming, feature planning, plan execution, code review, root-cause analysis, bug fixing, validation, PRD creation, atomic commits, and a content ideation suite (LinkedIn, X, YouTube, Shorts). Use when you want a repeatable, reviewable engineering workflow instead of one-shot prompting — prime the agent on a codebase, plan a feature, execute the plan, then review and validate the result.
|
LRIL — Second-Brain Workflow Commands
A cohesive suite of slash commands that turn ad-hoc prompting into a repeatable,
reviewable engineering loop. Originally from the Dynamous
claude-code-second-brain workshop.
After installing, the commands are available under the lril: namespace
(e.g. /lril:prime, /lril:plan-feature, /lril:execute).
The PIV loop
The core workflow is Plan → Implement → Validate:
- Prime —
/lril:prime (and /lril:prime-tools) load codebase context so
the agent understands structure, stack, conventions, and current state.
- Plan —
/lril:plan-feature produces a comprehensive, research-backed
feature plan; /lril:create-prd turns a conversation into a PRD.
- Implement —
/lril:execute runs an implementation plan;
/lril:implement-fix and /lril:rca handle bug fixing and root-cause
analysis.
- Validate —
/lril:code-review (pre-commit technical review),
/lril:code-review-fix, /lril:validate, and /lril:commit close the loop
with atomic, conventionally-tagged commits.
- Reflect —
/lril:execution-report and /lril:system-review compare what
was built against the plan to surface process improvements.
Commands
| Command | Purpose |
|---|
prime | Prime the agent with codebase understanding |
prime-tools | Prime the agent on available tools/MCP servers |
init-project | Initialize project conventions and context |
plan-feature | Create a comprehensive feature plan with deep codebase analysis and research |
create-prd | Create a Product Requirements Document from a conversation |
execute | Execute an implementation plan |
implement-fix | Implement a fix for an identified issue |
rca | Root-cause analysis for a bug or failure |
code-review | Technical code review for quality and bugs (runs pre-commit) |
code-review-fix | Apply fixes surfaced by a code review |
validate | Validate an implementation against its requirements |
commit | Commit all uncommitted changes with an atomic, conventionally-tagged message |
execution-report | Generate an implementation report for system review |
system-review | Analyze implementation against plan for process improvements |
content-all | Generate content for all channels at once |
content-linkedin | Draft a LinkedIn post |
content-x | Draft an X / Twitter post |
content-youtube | Draft YouTube content |
content-shorts | Draft short-form video content |
How it installs
This skill folder contains a Claude Code plugin (.claude-plugin/plugin.json)
plus the command definitions under commands/. The installer copies the whole
folder to ~/.claude/skills/lril/ (global) or .claude/skills/lril/ (project),
which is exactly where Claude Code expects the lril: command suite to live.
Restart Claude Code after installing and the /lril:* commands become available.