{"id":"aae3b6da-4f57-4501-abe7-c65319c4417c","shortId":"sgnqte","kind":"skill","title":"systems-thinking","tagline":"Apply systems thinking: feedback loops, second-order effects, leverage points, intervention plan.","description":"# Systems Thinking\n\n## Scope\n\n**Covers**\n- Seeing the \"whole system\" behind a problem (actors, incentives, feedback loops, culture/rules)\n- Anticipating second- and third-order effects (including time delays)\n- Finding leverage points (small changes with outsized impact)\n- Converting recurring pain into a reusable system (process, automation, or operating mechanism)\n\n**When to use**\n- \"This is a complex ecosystem; we're missing the bigger picture.\"\n- \"What are the second-order effects if we do X?\"\n- \"We keep solving symptoms—what's the system causing this?\"\n- \"Map the players + incentives and how they interact.\"\n- \"We need to redesign a process/org without unintended consequences.\"\n\n**When NOT to use**\n- The problem is simple/linear and mostly execution (use a project plan/timeline).\n- You need primary user research or data you don't have (do discovery first).\n- You need deep quantitative forecasting/simulation (this skill produces a qualitative map + risks, not a full model).\n- The decision is low-impact and fully reversible (don't over-invest).\n- You need to **redesign an org chart, team topology, or reporting lines** (use `organizational-design`; come back here only if you need to map systemic dynamics first).\n- You need to **compare discrete options** with weighted criteria and pick a winner (use `evaluating-trade-offs`).\n- You need to **plan under uncertainty** with scenario trees, hedging strategies, and contingency triggers (use `planning-under-uncertainty`).\n- You need to **run a decision process** with roles, meetings, and a decision log (use `running-decision-processes`).\n\n## Inputs\n\n**Minimum required**\n- The focal decision or problem statement (1–2 sentences)\n- Desired outcome + time horizon (default: 6–12 months)\n- Known constraints/guardrails (trust, safety, compliance, budget, headcount)\n- Known actors/stakeholders (teams, users, partners, regulators, vendors)\n- What has been tried already (and what happened)\n\n**Missing-info strategy**\n- Ask up to 5 questions from [references/INTAKE.md](references/INTAKE.md).\n- If answers aren't available, proceed with clearly labeled assumptions and provide 2–3 alternative system framings/boundaries.\n\n## Outputs (deliverables)\n\nProduce a **Systems Thinking Pack** in Markdown (in-chat; or as files if requested) in this order:\n\n1) **Context + System boundary** (goal, scope, non-scope, time horizon)\n2) **Actors & incentives map** (players, goals, constraints, power, conflicts)\n3) **System map** (key variables + causal links) + **feedback loops** (reinforcing/balancing) + **time delays**\n4) **Second-/third-order effects ledger** for the top 1–3 decisions\n5) **Leverage points** + **intervention plan** (actions, owners, sequencing, guardrails)\n6) **System-build opportunities** (what to automate/standardize to reduce recurring pain)\n7) **Risks / Open questions / Next steps** (required)\n\nTemplates: [references/TEMPLATES.md](references/TEMPLATES.md)\n\n## Workflow (8 steps)\n\n### 1) Intake + pick the focal decision/problem\n- **Inputs:** User context; use [references/INTAKE.md](references/INTAKE.md).\n- **Actions:** Restate the focal decision/problem, desired outcome, and time horizon; list constraints/guardrails.\n- **Outputs:** Draft **Context + System boundary**.\n- **Checks:** The problem is not a solution in disguise; scope and non-scope are explicit.\n\n### 2) Define the system boundary (what's \"in\" vs \"out\")\n- **Inputs:** Problem statement + constraints.\n- **Actions:** Choose a boundary that is useful (not everything). Name the primary outcome metric(s) and a few leading indicators.\n- **Outputs:** Boundary statement + success measures.\n- **Checks:** Boundary is tight enough to act on, but wide enough to include key externalities.\n\n### 3) Map actors + incentives (multi-agent reality)\n- **Inputs:** Boundary + stakeholder list.\n- **Actions:** Enumerate actors/players; capture incentives, constraints, power, and likely behaviors.\n- **Outputs:** **Actors & incentives map** (table).\n- **Checks:** Includes at least 1–2 \"invisible\" actors (e.g., policies, culture norms, platform constraints) if relevant.\n\n### 4) Build a simple system map (variables + causal links)\n- **Inputs:** Actors map + known dynamics.\n- **Actions:** List key variables; map causal links (\"A increases B\", \"C decreases D\"); mark time delays.\n- **Outputs:** **System map** (text/table) with 10–20 high-signal links.\n- **Checks:** Links are directional and testable; avoids buzzwords (\"alignment\", \"quality\") without definition.\n\n### 5) Identify feedback loops + time delays\n- **Inputs:** System map.\n- **Actions:** Extract reinforcing and balancing loops; note where delays create overshoot/oscillation; flag common traps.\n- **Outputs:** **Feedback loops** section (2–6 loops) + delays list.\n- **Checks:** Each loop has a short \"so what\" describing the pattern it creates.\n\n### 6) Run second-/third-order effects on 1–3 candidate moves\n- **Inputs:** Candidate decisions/actions.\n- **Actions:** For each move, enumerate first-, second-, and third-order effects; include who wins/loses and what constraints tighten over time.\n- **Outputs:** **Second-/third-order effects ledger**.\n- **Checks:** Includes at least one unintended consequence + one mitigating action per move.\n\n### 7) Choose leverage points + design interventions (including \"build a system\")\n- **Inputs:** Loops + effects ledger.\n- **Actions:** Identify leverage points (policy, incentives, information flows, tooling, process); propose interventions; include at least one system-build/automation opportunity for recurring pain.\n- **Outputs:** **Leverage points + intervention plan** + **System-build opportunities**.\n- **Checks:** Each intervention has an owner, a measurable leading indicator, and a guardrail.\n\n### 8) Quality gate + finalize pack\n- **Inputs:** All draft sections.\n- **Actions:** Run [references/CHECKLISTS.md](references/CHECKLISTS.md) and score with [references/RUBRIC.md](references/RUBRIC.md). Add **Risks / Open questions / Next steps**.\n- **Outputs:** Final **Systems Thinking Pack**.\n- **Checks:** A reader can act without a live meeting; trade-offs and uncertainties are explicit.\n\n## Quality gate (required)\n- Use [references/CHECKLISTS.md](references/CHECKLISTS.md) and [references/RUBRIC.md](references/RUBRIC.md).\n- Always include: **Risks**, **Open questions**, **Next steps**.\n\n## Examples\n\n**Example 1 (Org/process):** \"Our on-call load keeps rising and teams are burned out. Map the system and propose leverage points.\"  \nExpected: an actors/incentives map (teams, incidents, incentives), feedback loops (firefighting loop), effects ledger for candidate changes, and an intervention plan with guardrails.\n\n**Example 2 (Product ecosystem):** \"We're changing API pricing; what are the second-order effects across partners and customer segments?\"  \nExpected: system boundary + actors map (customers/partners/internal), loops and delays, effects ledger, and a sequencing/mitigation plan.\n\n**Boundary example:** \"Write a status update about this week's tasks.\"\nResponse: this skill is for complex systems/decisions. Suggest a project update format instead; only use this skill if there's a systemic pattern to diagnose.\n\n**Boundary example (neighbor redirect):** \"Redesign our engineering org to reduce dependencies between teams.\"\nResponse: this is an org structure redesign, not a systems analysis. Use `organizational-design` for team topology and reporting lines. Use this skill first only if you need to map the systemic dynamics (feedback loops, incentives) driving the dependency problem before redesigning.\n\n## Anti-patterns\n\n1. **Map-everything paralysis** — Drawing a system map with 50+ variables and no prioritization, making it impossible to act. Limit the map to 10-20 high-signal links and focus on the loops that explain the observed symptoms.\n2. **Feedback loops without \"so what\"** — Identifying reinforcing and balancing loops but not explaining what behavior they produce or how to intervene. Every loop must have a short narrative about the pattern it creates and why it matters.\n3. **Confusing correlation with causation** — Drawing causal arrows based on co-occurrence rather than testable mechanisms. Each link should state the mechanism (\"A increases B because...\") not just the direction.\n4. **Ignoring time delays** — Treating all effects as instantaneous when many systems have 3-12 month lags between action and result. Always mark time delays on the system map and account for them in intervention sequencing.\n5. **Leverage-point optimism** — Assuming a single intervention will fix the system without considering resistance, adaptation, or second-order effects of the intervention itself. Every intervention needs a guardrail and a \"what could go wrong\" check.","tags":["systems","thinking","lenny","skills","plus","liqiongyu","agent-skills","ai-agents","automation","claude","codex","prompt-engineering"],"capabilities":["skill","source-liqiongyu","skill-systems-thinking","topic-agent-skills","topic-ai-agents","topic-automation","topic-claude","topic-codex","topic-prompt-engineering","topic-refoundai","topic-skillpack"],"categories":["lenny_skills_plus"],"synonyms":[],"warnings":[],"endpointUrl":"https://skills.sh/liqiongyu/lenny_skills_plus/systems-thinking","protocol":"skill","transport":"skills-sh","auth":{"type":"none","details":{"cli":"npx skills add liqiongyu/lenny_skills_plus","source_repo":"https://github.com/liqiongyu/lenny_skills_plus","install_from":"skills.sh"}},"qualityScore":"0.474","qualityRationale":"deterministic score 0.47 from registry signals: · indexed on github topic:agent-skills · 49 github stars · SKILL.md body (8,762 chars)","verified":false,"liveness":"unknown","lastLivenessCheck":null,"agentReviews":{"count":0,"score_avg":null,"cost_usd_avg":null,"success_rate":null,"latency_p50_ms":null,"narrative_summary":null,"summary_updated_at":null},"enrichmentModel":"deterministic:skill-github:v1","enrichmentVersion":1,"enrichedAt":"2026-04-22T00:56:25.779Z","embedding":null,"createdAt":"2026-04-18T22:17:09.916Z","updatedAt":"2026-04-22T00:56:25.779Z","lastSeenAt":"2026-04-22T00:56:25.779Z","tsv":"'-12':1137 '-20':1039 '/automation':750 '/third-order':383,669,702 '1':267,349,389,426,556,672,840,1014 '10':603,1038 '12':276 '2':268,324,360,471,557,648,884,1054 '20':604 '3':325,369,390,525,673,1092,1136 '4':381,568,1123 '5':307,392,621,1159 '50':1024 '6':275,401,649,666 '7':413,717 '8':424,777 'account':1153 'across':899 'act':516,810,1033 'action':397,438,485,537,582,630,679,714,731,786,1141 'actor':28,361,527,548,559,578,907 'actors/incentives':863 'actors/players':539 'actors/stakeholders':286 'adapt':1175 'add':795 'agent':531 'align':617 'alreadi':296 'altern':326 'alway':831,1144 'analysi':978 'answer':313 'anti':1012 'anti-pattern':1011 'anticip':33 'api':890 'appli':4 'aren':314 'arrow':1099 'ask':304 'assum':1164 'assumpt':321 'autom':59 'automate/standardize':408 'avail':316 'avoid':615 'b':591,1117 'back':191 'balanc':634,1063 'base':1100 'behavior':546,1069 'behind':25 'bigger':75 'boundari':352,454,475,488,506,511,534,906,919,955 'budget':283 'build':404,569,724,749,762 'burn':852 'buzzword':616 'c':592 'call':845 'candid':674,677,875 'captur':540 'caus':96 'causal':374,575,587,1098 'causat':1096 'chang':47,876,889 'chart':180 'chat':340 'check':455,510,552,609,653,705,764,806,1196 'choos':486,718 'clear':319 'co':1103 'co-occurr':1102 'come':190 'common':642 'compar':205 'complex':69,935 'complianc':282 'conflict':368 'confus':1093 'consequ':114,711 'consid':1173 'constraint':366,484,542,565,696 'constraints/guardrails':279,449 'context':350,434,452 'conting':232 'convert':51 'correl':1094 'could':1193 'cover':20 'creat':639,665,1087 'criteria':210 'cultur':562 'culture/rules':32 'custom':902 'customers/partners/internal':909 'd':594 'data':136 'decis':161,244,251,256,263,391 'decision/problem':431,442 'decisions/actions':678 'decreas':593 'deep':146 'default':274 'defin':472 'definit':620 'delay':42,380,597,626,638,651,912,1126,1147 'deliver':330 'depend':965,1007 'describ':661 'design':189,721,982 'desir':270,443 'diagnos':954 'direct':612,1122 'discoveri':142 'discret':206 'disguis':463 'draft':451,784 'draw':1019,1097 'drive':1005 'dynam':200,581,1001 'e.g':560 'ecosystem':70,886 'effect':12,39,83,384,670,690,703,729,872,898,913,1129,1180 'engin':961 'enough':514,520 'enumer':538,683 'evalu':217 'evaluating-trade-off':216 'everi':1076,1185 'everyth':493,1017 'exampl':838,839,883,920,956 'execut':125 'expect':861,904 'explain':1050,1067 'explicit':470,821 'extern':524 'extract':631 'feedback':7,30,376,623,645,868,1002,1055 'file':343 'final':780,802 'find':43 'firefight':870 'first':143,201,684,992 'fix':1169 'flag':641 'flow':738 'focal':262,430,441 'focus':1045 'forecasting/simulation':148 'format':941 'framings/boundaries':328 'full':158 'fulli':167 'gate':779,823 'go':1194 'goal':353,365 'guardrail':400,776,882,1189 'happen':299 'headcount':284 'hedg':229 'high':606,1041 'high-sign':605,1040 'horizon':273,359,447 'identifi':622,732,1060 'ignor':1124 'impact':50,165 'imposs':1031 'in-chat':338 'incent':29,101,362,528,541,549,736,867,1004 'incid':866 'includ':40,522,553,691,706,723,743,832 'increas':590,1116 'indic':504,773 'info':302 'inform':737 'input':258,432,481,533,577,627,676,727,782 'instantan':1131 'instead':942 'intak':427 'interact':105 'interven':1075 'intervent':15,395,722,742,758,766,879,1157,1167,1183,1186 'invest':173 'invis':558 'keep':89,847 'key':372,523,584 'known':278,285,580 'label':320 'lag':1139 'lead':503,772 'least':555,708,745 'ledger':385,704,730,873,914 'leverag':13,44,393,719,733,756,859,1161 'leverage-point':1160 'like':545 'limit':1034 'line':185,988 'link':375,576,588,608,610,1043,1110 'list':448,536,583,652 'live':813 'load':846 'log':252 'loop':8,31,377,624,635,646,650,655,728,869,871,910,1003,1048,1056,1064,1077 'low':164 'low-impact':163 'make':1029 'mani':1133 'map':98,154,198,363,371,526,550,573,579,586,600,629,854,864,908,998,1016,1022,1036,1151 'map-everyth':1015 'mark':595,1145 'markdown':337 'matter':1091 'measur':509,771 'mechan':62,1108,1114 'meet':248,814 'metric':498 'minimum':259 'miss':73,301 'missing-info':300 'mitig':713 'model':159 'month':277,1138 'most':124 'move':675,682,716 'multi':530 'multi-ag':529 'must':1078 'name':494 'narrat':1082 'need':107,131,145,175,196,203,221,240,996,1187 'neighbor':957 'next':417,799,836 'non':356,467 'non-scop':355,466 'norm':563 'note':636 'observ':1052 'occurr':1104 'off':219,817 'on-cal':843 'one':709,712,746 'open':415,797,834 'oper':61 'opportun':405,751,763 'optim':1163 'option':207 'order':11,38,82,348,689,897,1179 'org':179,962,972 'org/process':841 'organiz':188,981 'organizational-design':187,980 'outcom':271,444,497 'output':329,450,505,547,598,644,700,755,801 'outsiz':49 'over-invest':171 'overshoot/oscillation':640 'owner':398,769 'pack':335,781,805 'pain':53,412,754 'paralysi':1018 'partner':289,900 'pattern':663,952,1013,1085 'per':715 'pick':212,428 'pictur':76 'plan':16,223,236,396,759,880,918 'plan/timeline':129 'planning-under-uncertainti':235 'platform':564 'player':100,364 'point':14,45,394,720,734,757,860,1162 'polici':561,735 'power':367,543 'price':891 'primari':132,496 'priorit':1028 'problem':27,120,265,457,482,1008 'proceed':317 'process':58,245,257,740 'process/org':111 'produc':151,331,1071 'product':885 'project':128,939 'propos':741,858 'provid':323 'qualit':153 'qualiti':618,778,822 'quantit':147 'question':308,416,798,835 'rather':1105 're':72,888 'reader':808 'realiti':532 'recur':52,411,753 'redesign':109,177,959,974,1010 'redirect':958 'reduc':410,964 'references/checklists.md':788,789,826,827 'references/intake.md':310,311,436,437 'references/rubric.md':793,794,829,830 'references/templates.md':421,422 'regul':290 'reinforc':632,1061 'reinforcing/balancing':378 'relev':567 'report':184,987 'request':345 'requir':260,419,824 'research':134 'resist':1174 'respons':930,968 'restat':439 'result':1143 'reusabl':56 'revers':168 'rise':848 'risk':155,414,796,833 'role':247 'run':242,255,667,787 'running-decision-process':254 'safeti':281 'scenario':227 'scope':19,354,357,464,468 'score':791 'second':10,34,81,382,668,685,701,896,1178 'second-ord':9,80,895,1177 'section':647,785 'see':21 'segment':903 'sentenc':269 'sequenc':399,1158 'sequencing/mitigation':917 'short':658,1081 'signal':607,1042 'simpl':571 'simple/linear':122 'singl':1166 'skill':150,932,946,991 'skill-systems-thinking' 'small':46 'solut':461 'solv':90 'source-liqiongyu' 'stakehold':535 'state':1112 'statement':266,483,507 'status':923 'step':418,425,800,837 'strategi':230,303 'structur':973 'success':508 'suggest':937 'symptom':91,1053 'system':2,5,17,24,57,95,199,327,333,351,370,403,453,474,572,599,628,726,748,761,803,856,905,951,977,1000,1021,1134,1150,1171 'system-build':402,747,760 'systems-think':1 'systems/decisions':936 'tabl':551 'task':929 'team':181,287,850,865,967,984 'templat':420 'testabl':614,1107 'text/table':601 'think':3,6,18,334,804 'third':37,688 'third-ord':36,687 'tight':513 'tighten':697 'time':41,272,358,379,446,596,625,699,1125,1146 'tool':739 'top':388 'topic-agent-skills' 'topic-ai-agents' 'topic-automation' 'topic-claude' 'topic-codex' 'topic-prompt-engineering' 'topic-refoundai' 'topic-skillpack' 'topolog':182,985 'trade':218,816 'trade-off':815 'trap':643 'treat':1127 'tree':228 'tri':295 'trigger':233 'trust':280 'uncertainti':225,238,819 'unintend':113,710 'updat':924,940 'use':65,118,126,186,215,234,253,435,491,825,944,979,989 'user':133,288,433 'variabl':373,574,585,1025 'vendor':291 'vs':479 'week':927 'weight':209 'whole':23 'wide':519 'winner':214 'wins/loses':693 'without':112,619,811,1057,1172 'workflow':423 'write':921 'wrong':1195 'x':87","prices":[{"id":"7ebdc430-f720-49dc-a95c-448b9b18c32c","listingId":"aae3b6da-4f57-4501-abe7-c65319c4417c","amountUsd":"0","unit":"free","nativeCurrency":null,"nativeAmount":null,"chain":null,"payTo":null,"paymentMethod":"skill-free","isPrimary":true,"details":{"org":"liqiongyu","category":"lenny_skills_plus","install_from":"skills.sh"},"createdAt":"2026-04-18T22:17:09.916Z"}],"sources":[{"listingId":"aae3b6da-4f57-4501-abe7-c65319c4417c","source":"github","sourceId":"liqiongyu/lenny_skills_plus/systems-thinking","sourceUrl":"https://github.com/liqiongyu/lenny_skills_plus/tree/main/skills/systems-thinking","isPrimary":false,"firstSeenAt":"2026-04-18T22:17:09.916Z","lastSeenAt":"2026-04-22T00:56:25.779Z"}],"details":{"listingId":"aae3b6da-4f57-4501-abe7-c65319c4417c","quickStartSnippet":null,"exampleRequest":null,"exampleResponse":null,"schema":null,"openapiUrl":null,"agentsTxtUrl":null,"citations":[],"useCases":[],"bestFor":[],"notFor":[],"kindDetails":{"org":"liqiongyu","slug":"systems-thinking","github":{"repo":"liqiongyu/lenny_skills_plus","stars":49,"topics":["agent-skills","ai-agents","automation","claude","codex","prompt-engineering","refoundai","skillpack"],"license":"apache-2.0","html_url":"https://github.com/liqiongyu/lenny_skills_plus","pushed_at":"2026-04-04T06:30:11Z","description":"86 agent-executable skill packs converted from RefoundAI’s Lenny skills (unofficial). Works with Codex + Claude Code.","skill_md_sha":"cb94acb608c1e2ef1a917a5da128ada11af5586f","skill_md_path":"skills/systems-thinking/SKILL.md","default_branch":"main","skill_tree_url":"https://github.com/liqiongyu/lenny_skills_plus/tree/main/skills/systems-thinking"},"layout":"multi","source":"github","category":"lenny_skills_plus","frontmatter":{"name":"systems-thinking","description":"Apply systems thinking: feedback loops, second-order effects, leverage points, intervention plan."},"skills_sh_url":"https://skills.sh/liqiongyu/lenny_skills_plus/systems-thinking"},"updatedAt":"2026-04-22T00:56:25.779Z"}}