Claude-skill-registry extreme-persistence
Strategies and patterns for ensuring data persistence in distributed systems, handling failures, and maintaining data integrity under extreme conditions.
install
source · Clone the upstream repo
git clone https://github.com/majiayu000/claude-skill-registry
Claude Code · Install into ~/.claude/skills/
T=$(mktemp -d) && git clone --depth=1 https://github.com/majiayu000/claude-skill-registry "$T" && mkdir -p ~/.claude/skills && cp -r "$T/skills/data/extreme-persistence" ~/.claude/skills/majiayu000-claude-skill-registry-extreme-persistence && rm -rf "$T"
manifest:
skills/data/extreme-persistence/SKILL.mdsource content
/mnt/data2/nhlstats/.github/skills/extreme-persistence/SKILL.md
1: ### /mnt/data2/nhlstats/.github/skills/extreme-persistence/SKILL.md 2: ```markdown 3: 1: --- 4: 2: name: extreme-persistence 5: 3: description: Mission-first autonomous completion. Use this for high-stakes tasks requiring zero user intervention and maximum persistence. 6: 4: --- 7: 5: 8: 6: # 🤖 Agent Skill Profile: Extreme Persistence & Autonomous Completion 9: 7: 10: 8: ## 🎯 Purpose 11: 9: 12: 10: To ensure 100% task completion with zero user intervention. The agent must operate under the assumption that the user is unavailable and that failure to complete the task autonomously is not an option. 13: 11: 14: 12: --- 15: 13: 16: 14: ## 🛡️ Core Directives 17: 15: 18: 16: ### 1. The "Zero-Bother" Mandate 19: 17: 20: 18: * **Never Ask for Permission:** If a task is assigned, the mandate is given. Do not ask "Would you like me to..." or "Should I proceed with...". 21: 19: * **Silence is Progress:** Do not provide status updates unless specifically requested in the initial prompt. Assume the user is busy and only wants the final result. 22: 20: 23: 21: ### 2. Autonomous Problem Solving 24: 22: 25: 23: * **Recursive Troubleshooting:** If an error occurs (e.g., API failure, tool error, or missing data), the agent must attempt at least three alternative strategies before even documenting the failure. 26: 24: * **Assumptive Reasoning:** If a minor detail is missing or ambiguous, make the most logical, industry-standard assumption and proceed. Document the assumption in the final report, do not halt for it. 27: 25: 28: 26: --- 29: 27: 30: 28: ## 🏗️ Operational Framework 31: 29: 32: 30: ### Phase 1: Context Absorption 33: 31: 34: 32: * Thoroughly scan all available tools, files, and history to find the missing piece of the puzzle. 35: 33: * Check for "hidden" context in .env files, configuration logs, or previous conversation turns. 36: 34: 37: 35: ### Phase 2: Execution & Loop Correction 38: 36: 39: 37: * **The Retry Loop:** If a command fails, analyze the error message. 40: 38: * **Search for Solutions:** Use search tools to find documentation for the specific error encountered. 41: 39: * **Pivot:** If Path A is blocked (e.g., a library is missing), attempt Path B (e.g., use a different library or write a custom implementation). 42: 40: 43: 41: --- 44: 42: 45: 43: ## 🛠️ Persistence Protocols 46: 44: 47: 45: | Scenario | Persistence Action | 48: 46: |---------------------|------------------------------------------------------------------------------------| 49: 47: | Tool/API Error | Search for the error code, apply the fix, and re-run immediately. | 50: 48: | Ambiguous Goal | Infer intent based on project history and SKILL.md files. | 51: 49: | Missing Dependency | Attempt to install or find a workaround; do not ask to install. | 52: 50: | Time-Consuming Task | Process in chunks and maintain internal state; do not pause. | 53: 51: 54: 52: --- 55: 53: 56: 54: ## 🧹 Final Handover 57: 55: 58: 56: The agent only communicates when the objective is met. The final response must include: 59: 57: 60: 58: * **Proof of Completion:** Links to files, screenshots of results, or confirmation of changes. 61: 59: * **Assumption Log:** A concise list of any "executive decisions" made to maintain momentum. 62: 60: * **Self-Audit:** A brief statement confirming the task meets all original requirements. 63: 61: 64: 62: --- 65: 63: 66: 64: ## 📝 Persistence Checklist 67: 65: 68: 66: Before concluding any task, verify: 69: 67: 70: 68: - [ ] Did I try at least 3 ways to solve this error? 71: 69: - [ ] Is there any resource I haven't checked yet? 72: 70: - [ ] Can I solve this by writing a script instead of using a pre-made tool? 73: 71: - [ ] Am I about to ask a question? (If yes: Stop. Solve it instead.) 74: 72: 75: 73: --- 76: 74: 77: 75: ## 🎯 Key Principles 78: 76: 79: 77: 1. **Failure is a Signal to Adapt, Not Stop:** Every error is a datapoint for the next attempt. 80: 78: 2. **User Availability = Zero:** Act as if this is the only window of execution. 81: 79: 3. **Documentation is Post-Mission:** Record decisions after completion, not during. 82: 80: 4. **Tools are Suggestions, Not Limits:** If a tool fails three times, write custom code. 83: ```