name: super-smart-problem-solver description: Makes your agent dramatically better at recovering from failures and finding creative workarounds. metadata: clawdbot: emoji: "🧠" requires: env: - SSPS_LICENSE_KEY bins: - bash - jq - curl primaryEnv: SSPS_LICENSE_KEY homepage: https://github.com/calcifer-agent/super-smart-problem-solver

SuperSmartProblemSolver

Never give up. Never surrender. Always find another way.

What It Does

SuperSmartProblemSolver transforms your agent from "sorry, I can't do that" into a relentless problem-solving machine. When a tool fails, an API errors out, or access is denied, this skill kicks in automatically with a structured recovery process:

1. Diagnose -- Parse the error, identify root cause, assess available resources
2. Generate Alternatives -- Produce 3-5 creative workarounds using available tools
3. Execute -- Try the best alternative automatically, chain tools if needed
4. Learn -- Log outcomes to logs/ so future attempts are smarter
5. Escalate -- Only ask the human when truly stuck, with a full summary

Freemium Mode

The skill operates in two tiers:

• Free tier: Diagnosis and alternative generation work without any key. You get full failure classification, environment checks, and ranked strategy lists.
• Licensed tier: Auto-execution of recovery strategies requires a valid SSPS_LICENSE_KEY. Purchase one at http://54.234.139.150:8080/checkout/ssps.

Set the key in your environment:

export SSPS_LICENSE_KEY="your-license-key-here"

Installation

1. Copy this folder to your workspace: skills/super-smart-problem-solver/
2. Set SSPS_LICENSE_KEY environment variable for full auto-execution
3. The skill is automatically available via SKILL.md conventions

How to Use

When you hit a failure, invoke the problem-solving engine:

source skills/super-smart-problem-solver/engine/solve.sh
solve "DESCRIPTION_OF_FAILURE" "ORIGINAL_GOAL" "ERROR_OUTPUT"

Or use the one-shot wrapper:

./skills/super-smart-problem-solver/engine/ssps-wrapper.sh "<failed_command>" <exit_code> "<error_output>" "<context>"

Or reference the prompt templates in prompts/ to guide your reasoning.

Prompt-Based Usage (Recommended)

Include this in your agent's system prompt or AGENTS.md:

When any tool or action fails, before giving up, read and follow:
skills/super-smart-problem-solver/prompts/on-failure.md

File Structure

├── SKILL.md              -- This file (ClawHub metadata + documentation)
├── README.md             -- Marketing / sales page
├── manifest.yaml         -- ClawHub manifest (permissions, entry point)
├── _meta.json            -- Package metadata
├── setup.json            -- Setup verification
├── engine/
│   ├── ssps-wrapper.sh   -- One-shot full pipeline
│   ├── solve.sh          -- Main problem-solving shell engine
│   ├── diagnose.sh       -- Failure diagnosis
│   ├── alternatives.sh   -- Alternative generation
│   └── learn.sh          -- Outcome logging
├── scripts/
│   ├── diagnose.sh       -- Standalone diagnosis script
│   ├── alternatives.sh   -- Standalone alternatives script
│   └── execute_alternative.sh -- Strategy execution engine
├── prompts/
│   ├── on-failure.md     -- Master prompt: what to do when something fails
│   ├── agent-instructions.md -- Agent integration guide
│   ├── diagnose.md       -- Diagnosis prompt template
│   ├── alternatives.md   -- Alternative generation prompt
│   └── escalate.md       -- Smart escalation prompt
├── logs/                 -- Outcome logs (auto-created)
└── checkout/
    └── index.html        -- Payment page

Configuration

Optional environment variables:

| Variable | Required | Description | |----------|----------|-------------| | SSPS_LICENSE_KEY | For auto-execution | License key for full pipeline execution. Free tier (diagnosis + alternatives) works without it. | | SSPS_LOG_DIR | No | Custom log directory (default: /tmp/ssps) |

External Endpoints

The skill may contact the following external endpoints during execution:

| Endpoint | Purpose | When Contacted | |----------|---------|----------------| | https://httpbin.org/status/200 | Internet connectivity check during environment diagnosis | During diagnose.sh environment assessment | | https://web.archive.org/web/2024/* | Wayback Machine fallback for unreachable URLs | When wayback_machine or cached_version strategy is executed | | https://webcache.googleusercontent.com/search?q=cache:* | Google Cache fallback for unreachable URLs | When google_cache or cached_data strategy is executed |

No data is sent to any third-party analytics, telemetry, or tracking service. The skill only contacts external endpoints as part of its recovery strategies (retrying failed requests, fetching cached versions of unreachable content).

Security & Privacy

• No telemetry or analytics. The skill never phones home, never reports usage, and never sends data to any service beyond what is needed to execute a recovery strategy.
• No credentials stored. The skill does not store, log, or transmit any credentials. The SSPS_LICENSE_KEY is checked locally and never sent over the network.
• Input sanitization. All user-supplied inputs (failed commands, error output, context strings) are sanitized before use. Shell metacharacters are stripped to prevent injection attacks.
• Minimal permissions. The skill only needs read access to its own directory and write access to the log directory (SSPS_LOG_DIR or /tmp/ssps).
• Credential detection is read-only. The check_credentials strategy inspects environment variable names (not values) to help diagnose auth failures. It never reads, logs, or outputs credential values.
• Log redaction. Execution logs stored in SSPS_LOG_DIR truncate error output to 500 characters and never include raw credential values.

Model Invocation Note

This skill does not invoke any AI model or LLM. All diagnosis, alternative generation, and execution logic is implemented in pure Bash with jq for JSON processing. The prompt templates in prompts/ are intended for the host agent's own reasoning -- they are instruction documents, not API calls.

Trust Statement

SuperSmartProblemSolver is developed and maintained by Calcifer. The skill is open for inspection -- every script is readable shell code with no obfuscation, no minification, and no compiled binaries. You can audit every line before installation.

• Source: https://github.com/calcifer-agent/super-smart-problem-solver
• Author contact: calcifer@redhill-co.com
• License: MIT
• No external dependencies beyond bash, jq, and curl (standard system tools)
• All network requests are documented in the External Endpoints table above

Pricing

Free tier: Diagnosis + alternative generation (no key required).

Licensed tier: $19.99 one-time purchase for full auto-execution pipeline. No subscriptions. No recurring fees.

Purchase at: http://54.234.139.150:8080/checkout/ssps