Browse Skills

Auto Memorize extracts lessons, errors, fixes, and user preferences from a session at the end and stores them as structured local memory. It asks for consent on first run and stays local unless an API model is configured.

Context Optimize helps when a conversation gets long, repetitive, or unstable. It identifies stale context and suggests compaction and session rules to keep turns coherent.

Critique challenges decisions, surfaces assumptions, and offers concrete counter-arguments when someone asks for a second opinion or devil’s advocate. It also reframes non-trivial refactors toward the elegant solution before choosing a pragmatic fix.

Deep Dive helps when requirements are incomplete or conflicting by running a structured interview, recording assumptions, and producing a launch brief. Use it before planning or implementation when scope or acceptance criteria are unclear.

Engineering Prompts creates and optimizes prompts for LLMs and AI systems. Use it when you need a new prompt, better prompt structure, or a fit-for-purpose prompting technique.

Focus Coach helps users choose one main direction, identify the biggest constraint, and decide whether to continue, pause, or pivot. It is used for strategic reviews of goals, projects, or learning paths before handing execution to learning-coach.

IJFW AGENTS.md Manager maintains a project's canonical AGENTS.md file and its marker blocks. Use it when updating agent instructions or regenerating the file from IJFW-generated agent data.

IJFW Compute runs sandboxed scripts for large or repetitive data tasks like log analysis, aggregation, deduping, and counting. It also supports indexing and searching findings so recurring questions can reuse prior results.

IJFW Core applies always-on output rules, mode routing, and context discipline for Claude Code interactions. Use it when you want terse, task-routed responses with optional smart, fast, deep, manual, or brutal modes.

IJFW Handoff generates a concise session handoff file and loads it on the next session start. Use it when ending work, hitting context limits, or resuming a paused task.

IJFW Metrics tracks session-level usage at stop time and summarizes tokens, agent activity, routing savings, and continuity. Use it when you want to review Claude Code performance with /ijfw-status.

IJFW Plan turns a brief or spec into a falsifiable PLAN.md with tasks, dependency waves, and success criteria. Use it when someone asks to plan work or wants the next execution steps after discovery or research.