Install
openclaw skills install funky-fund-flamingoClawHub Security found sensitive or high-impact capabilities. Review the scan results before using.
Repair-first self-evolution for OpenClaw — audit logs, memory, and skills; run measurable mutation cycles. Get paid. Evolve. Repeat. Dolla dolla bill y'all.
openclaw skills install funky-fund-flamingoUse this skill when you're ready to get paid. We inspect reality, kill breakage and value leaks, and run mutation cycles that produce concrete gains — so the stack earns, not just runs.
--loop / --funky-fund-flamingo) so evolution runs in the background and the revenue keeps flowing.~/.openclaw/agents/<agent>/sessions/*.jsonlMEMORY.md, memory/YYYY-MM-DD.md, USER.mdskills/../../.envindex.jsevolve.jsRun from workspace root:
node skills/funky-fund-flamingo/index.js run
Run from inside this skill directory:
node index.js run
# single cycle — one shot, max impact
node index.js run
# alias command
node index.js /evolve
# human confirmation before significant edits (protect the bag)
node index.js run --review
# prompt generation only (writes prompt artifact to memory dir)
node index.js run --dry-run
# continuous relay — keep the money printer running
node index.js --loop
node index.js run --funky-fund-flamingo
Each cycle should:
memory/evolution_state.json) and optionally schedule the next loop.memory/funky_fund_flamingo_persistent_memory.json) so strategy compounds and the bag gets bigger.| URL | Data sent | Purpose |
|---|---|---|
| None (from this skill's code) | — | This skill's Node.js code does not open sockets or make HTTP requests. It only reads/writes local files. |
Important: The repo includes agent config templates (agents/openai.yaml, agents/openrouter.yaml) for use by an OpenClaw (or other) agent. When you run an agent that uses this skill with a cloud model (OpenAI, OpenRouter, etc.), that agent will send the prompts this skill builds — which can include excerpts from session logs, memory, and workspace context — to the provider's API. So "local-only" applies to the skill binary itself; if the skill is invoked by an agent backed by a third-party LLM, data can leave the machine via that agent. To stay fully local, run node index.js run (or --dry-run) without routing the generated prompt through a cloud model.
~/.openclaw/agents/<agent>/sessions/*.jsonl, workspace MEMORY.md, memory/, USER.md, and the skills/ directory.memory/evolution_state.json, memory/funky_fund_flamingo_persistent_memory.json, and optionally prompt artifacts in the memory dir. This skill does not push or publish anywhere; any outbound data is only via whatever agent/model stack you choose to run.No env vars are required. The following are optional overrides (see evolve.js / README):
| Variable | Purpose | Typical default |
|---|---|---|
AGENT_NAME | Agent session folder under ~/.openclaw/agents/ | main |
MEMORY_DIR | Directory for evolution state and persistent memory | workspace memory/ |
TARGET_SESSION_BYTES | Max bytes read from latest session log | 64000 |
LOOP_MIN_INTERVAL_SECONDS | Min delay between loop cycles | 900 |
MAX_MEMORY_CHARS, MAX_TODAY_LOG_CHARS, MAX_PERSISTENT_MEMORY_CHARS | Content caps for prompts | see evolve.js |
ECONOMIC_KEYWORDS | Comma-separated keywords for value scoring | built-in list |
EVOLVE_REPORT_DIRECTIVE, EVOLVE_EXTRA_MODES, EVOLVE_ENABLE_SESSION_ARCHIVE | Behavior tweaks | — |
Evolution can be run manually (node index.js run) or by an agent that uses this skill. In relay mode (--loop / --funky-fund-flamingo), this process only plans and writes prompts; it does not call any model API. If you run an agent that consumes this skill with OpenAI/OpenRouter/etc., that agent will perform the model calls. To avoid sending local context to a provider, run the skill in --dry-run and do not feed the generated prompt to a cloud model.
The master directive (funky-fund-flamingo-master-directive.json) sets must_evolve_each_cycle and no_op_forbidden, which push every cycle toward making a concrete change. That can increase how often local files are mutated. For lower risk, prefer --review (confirm before significant edits) or --dry-run (prompt generation only, no writes). You can also edit or override the directive to relax these flags.
By using this skill, you run Node.js code that reads and writes files in your OpenClaw workspace and agent session directories. This skill's code does not send data to third parties; if an agent that uses this skill calls a cloud LLM, that agent (not this skill binary) sends the prompt. Only install if you trust the skill source (e.g. ClawHub and the publisher).
ADL.md — anti-degeneration so we don't break the money printerVFM.md — value-focused mutation: only changes that payTREE.md — capability topology and revenue-ready nodes.clawhub/FMEP.md (forced mutation execution policy)node index.js --help
Dolla, dolla bill y'all. 🦩