Loading...
Loading...
Found 21 Skills
Monitors context window health throughout a session and rides peak context quality for maximum output fidelity. Activates automatically after plan-interview and intent-framed-agent. Stays active through execution and hands off cleanly to simplify-and-harden and self-improvement when the wave completes naturally or exits via handoff. Use this skill whenever a multi-step agent task is underway and session continuity or context drift is a concern. Especially important for long-running tasks, complex refactors, or any work where degraded context would silently corrupt the output. Trigger even if the user doesn't say "context surfing" — if an agent task is running across multiple steps with intent and a plan already established, this skill is live.
Start a repo-local OptimizeSpec self-improvement change. Use when the user wants to create evals, optimize an agent with GEPA, define an agent self-improvement loop, or begin an ASI-first evaluation workflow.
Use when the user asks to "improve my agent", "self-improving agent", "auto-tune my agent", "iterate on my agent prompt", "fix my agent based on test results", "close the loop on agent quality", "auto-improve agent prompt", "use eval results to improve agent", "optimize my prompt based on failures", "rewrite my prompt", or describes agent self-improvement, prompt iteration from run results, or automated agent quality loops. Covers the full diagnose → propose → apply → re-validate loop for VAPI agents (squads + tool definitions) and for self-hosted agents (custom websocket servers, including the offline / pasted-prompt degenerate variant).
Anthropic's method for training harmless AI through self-improvement. Two-phase approach - supervised learning with self-critique/revision, then RLAIF (RL from AI Feedback). Use for safety alignment, reducing harmful outputs without human labels. Powers Claude's safety system.
Add persistent learning and self-improvement to AI agents using ACE framework
HOWL v2 — Hunt, Optimize, Win, Learn. Nightly self-improvement loop for the WOLF autonomous trading strategy. Runs once per day (via cron) to review all trades from the last 24 hours, compute win rates, analyze signal quality correlation, evaluate DSL tier performance, identify missed opportunities, and produce concrete improvement suggestions for the wolf-strategy skill. v2 adds fee drag ratio (FDR) analysis, holding period bucketing, LONG vs SHORT regime detection, rotation cost tracking, cumulative drift detection, and gross vs net profit factor separation. Use when setting up daily trade review automation, analyzing trading performance, or improving an autonomous trading strategy through data-driven feedback loops. Requires Senpi MCP connection, mcporter CLI, and OpenClaw cron system.
Workflow orchestration for complex coding tasks. Use for ANY non-trivial task (3+ steps or architectural decisions) to enforce planning, subagent strategy, self-improvement, verification, elegance, and autonomous bug fixing. Triggers: multi-step implementation, bug fixes, refactoring, architectural changes, or any task requiring structured execution.
Load this skill immediately when the user expresses any intent. System capabilities (tools/knowledge/scripts) live inside the plugin and are maintained through plugin updates. User data must live at project-level `.claude/pensieve/` and is never overwritten by the plugin. When the user asks to improve Pensieve system behavior (plugin content), you must use the Self-Improve tool (`tools/self-improve/_self-improve.md`).
Gain wisdom from setbacks — Go through the 5-step interactive reflection (Setback → Automatic Output → Old Weights → New Parameters → Alternative Action), move from "emotional review" to "behavioral training", and update the L3 weights of your first reactions. Use when Wang Jianshuo reflects on a personal setback, mistake, or recurring pattern (reflection, post-mortem review, review, draw lessons, learn from a setback, gain wisdom, "I messed up again", "Why does this keep happening?", "Why do I always…?", "I can't just let it go", "I know the principles but can't put them into practice"). For the user as a human, not for Claude's task post-mortems.