{"id":"2ba263e8-21aa-4b12-b0ba-a24a987fbf5d","shortId":"SCy9Uy","kind":"skill","title":"mint","tagline":"REQUIRED for ALL coding tasks. Invoke BEFORE writing any code or editing any file. Auto-routes to quick/plan/ship/research/verify. Enforces gates, reviews, atomic commits. No exceptions — if you're about to modify files, invoke mint first.","description":"# mint\n\nDisciplined agentic development. Fresh context per task, zero slop.\n\n**You are the orchestrator.** You route, dispatch, verify. You never implement yourself\n(except in quick mode). You output text BEFORE every agent dispatch — no silent chains.\n\n---\n\n## Step 1: Route the Task\n\nEvaluate in order — first match wins:\n\n| Signal | Mode | Action |\n|--------|------|--------|\n| \"verify\", \"check gates\", \"audit\" | verify | Read `modes/verify.md`, execute |\n| \"dream\", \"consolidate learning\" | dream | Read `modes/dream.md`, execute |\n| \"build is broken\", \"fix build errors\" | build-fix | Dispatch `mint-build-error-resolver` |\n| \"research\", \"how to\", \"compare\" | research | Read `modes/research.md`, execute |\n| \"automate\", \"workflow\", \"adaptive\", \"skill generation\" | automate | Read `modes/automate.md`, execute |\n| \"clean up\", \"dead code\", \"unused\" | refactor | Dispatch `mint-refactor-cleaner` |\n| \"ssh to\", \"run on staging/prod\" | ssh | Read `modes/ssh.md`, execute |\n| \"browse\", \"open\", \"screenshot\", \"scrape\" | browse | Read `modes/browse.md`, execute |\n| \"design review/profile/teach/steer\" | design | Read `modes/design.md`, execute |\n| Task touches ≤3 files, scope obvious | quick | Read `modes/quick.md`, execute |\n| Multiple features, \"ship\", \"build all\" | ship | Read `modes/ship.md`, execute |\n| Everything else | plan | Read `modes/plan.md`, execute |\n\nAnnounce your routing: \"Quick mode — handling directly with gates.\" or \"Plan mode — decomposing into specs.\"\nIf user overrides (\"just quick-fix it\"), follow their call.\n\n## Step 2: Write Session State\n\nBefore executing, write `.mint/sessions/<session-id>.json`:\n\n```json\n{\n  \"mintInvoked\": true,\n  \"invokedAt\": \"<ISO-8601>\",\n  \"task\": \"<task description>\",\n  \"mode\": \"<routed mode>\",\n  \"autoCommitOverride\": null,\n  \"designContextLoaded\": false,\n  \"activeSpec\": null\n}\n```\n\nSession ID: generated once per process (hex timestamp + random). Stable for the session.\n\nDetect autocommit overrides: `--no-commit`, \"no commits\", \"stop committing\" → set\n`autoCommitOverride: false`. Announce once, never re-ask.\n\n## Step 3: Load Mode and Execute\n\nRead the mode file for your routed mode from `skills/mint/modes/`. Follow its instructions.\n\n**For modes that modify code** (quick, plan, ship, design): also read\n`reference/orchestrator-laws.md` before executing. It contains context protection,\nstatus format, background dispatch rules, quality gates, and autocommit resolution.\nThese laws are mandatory for any mode that touches files or dispatches pipeline agents.\n\n**Lightweight modes** (research, verify, browse, ssh, build-fix, refactor, dream) do NOT\nneed orchestrator-laws.md. Their mode files are self-contained.\n\nNEVER hold the full pipeline in memory. Each phase file is self-contained.\n\n## Step 4: Clean Up\n\nOn task completion (success or failure):\n- Delete `.mint/sessions/<session-id>.json`\n- Verify it's gone\n\n---\n\n## Universal Rules (all modes)\n\n- **Asking the user:** Re-ground context (task + what's done + decision needed). One\n  decision per question — never batch. Present options with a recommendation. Take a position.\n- **Never push** — agents commit only, user reviews and pushes\n- **Fail twice → stop** — log to `.mint/issues.jsonl`, escalate to user\n\nFull orchestrator laws (context protection, status format, background dispatch, quality\ngates, autocommit, repo mode) are in `reference/orchestrator-laws.md` — loaded by Step 3\nfor code-modifying modes only.\n\n---\n\n## Reference Files (load on demand)\n\nThese files contain detailed reference for specific topics. Read them ONLY when you need\nthem for the current step — not at startup.\n\n| File | When to read |\n|------|-------------|\n| `reference/orchestrator-laws.md` | Code-modifying modes: context protection, status format, dispatch rules, gates, autocommit |\n| `reference/orchestrator-rules.md` | Risk scoring, DoD criteria, pipeline enforcement details |\n| `reference/learning-loop.md` | Before dispatching decomposer (issues, wins, instincts) |\n| `reference/session-state.md` | Session lifecycle details, autocommit override handling |\n| `reference/agent-control.md` | Stop/pause/freeze signals and recovery |\n| `reference/config.md` | CLI commands, config schema, multi-model dispatch |\n| `reference/context-mode.md` | When context-mode MCP is enabled |\n| `reference/workspace.md` | When workspace.repos is configured |\n| `reference/design.md` | When design intelligence is enabled |\n| `reference/graph.md` | When code graph is enabled — blast radius, call traces, architecture |\n| `reference/doc-manifest.md` | During doc-manifest check (step 6 of plan pipeline) |\n\n---\n\n## What Agents Receive (Prompt Caching)\n\nEach agent has two layers — **static** (cached) and **dynamic** (per-dispatch):\n\n- **Static:** The agent `.md` file in `agents/` — identity, rules, checklist. Cached by\n  Anthropic's API across identical requests. Multiple specs in a wave share this cache.\n- **Dynamic:** The `prompt` parameter built from `templates/agent-context.md`. Contains\n  only the per-dispatch inputs (spec XML, diff, config values). Keep this minimal.\n\n**Rule: Never duplicate agent instructions in the dynamic prompt.** The agent already has\nits `.md` file as system prompt.\n\n| Agent | Dynamic context (from `templates/agent-context.md`) |\n|-------|------------------------------------------------------|\n| Decomposer | Feature desc + config + hard blocks + learning context |\n| Planner | Spec XML + autocommit + TDD + retry/correction context |\n| Spec Reviewer | Spec XML + git diff |\n| Stage 2 Reviewers | Git diff + file list (+ conventions/business docs if applicable) |\n| Adversarial Tester | Spec XML + git diff + test framework + file paths (worktree isolation) |\n| Documenter | Doc path + description + change summary + manifest sections |\n| De-sloppifier | Git diff + spec XML + gate commands |\n| Researcher | Question + config |\n| Shipper | Ship plan + config + hard blocks |\n| Verifier | Failing gate output + config |\n| Build Resolver | Error output + config + in-scope files |\n| Dream Consolidator | Learning file paths + counts + config + previous report |","tags":["mint","3li7alaki","agent-skills","agentic-workflow","agents","claude-code","development"],"capabilities":["skill","source-3li7alaki","skill-mint","topic-agent-skills","topic-agentic-workflow","topic-agents","topic-claude-code","topic-development"],"categories":["mint"],"synonyms":[],"warnings":[],"endpointUrl":"https://skills.sh/3li7alaki/mint/mint","protocol":"skill","transport":"skills-sh","auth":{"type":"none","details":{"cli":"npx skills add 3li7alaki/mint","source_repo":"https://github.com/3li7alaki/mint","install_from":"skills.sh"}},"qualityScore":"0.455","qualityRationale":"deterministic score 0.46 from registry signals: · indexed on github topic:agent-skills · 10 github stars · SKILL.md body (6,103 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":"2026-05-18T19:08:20.604Z","embedding":null,"createdAt":"2026-05-18T13:13:49.265Z","updatedAt":"2026-05-18T19:08:20.604Z","lastSeenAt":"2026-05-18T19:08:20.604Z","tsv":"'1':75 '2':221,696 '3':171,275,457 '4':372 '6':581 'across':617 'action':87 'activespec':240 'adapt':128 'adversari':706 'agent':40,69,334,421,586,591,604,608,653,660,669 'alreadi':661 'also':302 'announc':194,268 'anthrop':614 'api':616 'applic':705 'architectur':573 'ask':273,392 'atom':24 'audit':91 'auto':17 'auto-rout':16 'autocommit':256,319,448,507,527,685 'autocommitoverrid':236,266 'autom':126,131 'background':313,444 'batch':410 'blast':569 'block':679,743 'broken':105 'brows':155,159,339 'build':103,107,110,115,182,342,749 'build-fix':109,341 'built':632 'cach':589,596,612,627 'call':219,571 'chain':73 'chang':722 'check':89,579 'checklist':611 'clean':135,373 'cleaner':145 'cli':536 'code':5,11,138,297,460,497,565 'code-modifi':459,496 'command':537,734 'commit':25,260,262,264,422 'compar':121 'complet':377 'config':538,645,677,737,741,748,753,764 'configur':556 'consolid':97,759 'contain':308,356,370,471,635 'context':43,309,398,440,500,547,671,681,688 'context-mod':546 'conventions/business':702 'count':763 'criteria':512 'current':486 'de':727 'de-sloppifi':726 'dead':137 'decis':403,406 'decompos':206,519,674 'delet':381 'demand':468 'desc':676 'descript':721 'design':163,165,301,559 'designcontextload':238 'detail':472,515,526 'detect':255 'develop':41 'diff':644,694,699,711,730 'direct':200 'disciplin':39 'dispatch':54,70,112,141,314,332,445,504,518,543,601,640 'doc':577,703,719 'doc-manifest':576 'document':718 'dod':511 'done':402 'dream':96,99,345,758 'duplic':652 'dynam':598,628,657,670 'edit':13 'els':189 'enabl':551,562,568 'enforc':21,514 'error':108,116,751 'escal':434 'evalu':79 'everi':68 'everyth':188 'except':27,60 'execut':95,102,125,134,154,162,168,178,187,193,226,279,306 'fail':428,745 'failur':380 'fals':239,267 'featur':180,675 'file':15,34,172,283,330,352,366,465,470,491,606,665,700,714,757,761 'first':37,82 'fix':106,111,215,343 'follow':217,290 'format':312,443,503 'framework':713 'fresh':42 'full':360,437 'gate':22,90,202,317,447,506,733,746 'generat':130,244 'git':693,698,710,729 'gone':387 'graph':566 'ground':397 'handl':199,529 'hard':678,742 'hex':248 'hold':358 'id':243 'ident':609,618 'implement':58 'in-scop':754 'input':641 'instinct':522 'instruct':292,654 'intellig':560 'invok':7,35 'invokedat':233 'isol':717 'issu':520 'json':229,230,383 'keep':647 'law':322,439 'layer':594 'learn':98,680,760 'lifecycl':525 'lightweight':335 'list':701 'load':276,454,466 'log':431 'mandatori':324 'manifest':578,724 'match':83 'mcp':549 'md':605,664 'memori':363 'minim':649 'mint':1,36,38,114,143 'mint-build-error-resolv':113 'mint-refactor-clean':142 'mint/issues.jsonl':433 'mint/sessions':228,382 'mintinvok':231 'mode':63,86,198,205,235,277,282,287,294,327,336,351,391,450,462,499,548 'model':542 'modes/automate.md':133 'modes/browse.md':161 'modes/design.md':167 'modes/dream.md':101 'modes/plan.md':192 'modes/quick.md':177 'modes/research.md':124 'modes/ship.md':186 'modes/ssh.md':153 'modes/verify.md':94 'modifi':33,296,461,498 'multi':541 'multi-model':540 'multipl':179,620 'need':348,404,482 'never':57,270,357,409,419,651 'no-commit':258 'null':237,241 'obvious':174 'one':405 'open':156 'option':412 'orchestr':51,438 'orchestrator-laws.md':349 'order':81 'output':65,747,752 'overrid':211,257,528 'paramet':631 'path':715,720,762 'per':44,246,407,600,639 'per-dispatch':599,638 'phase':365 'pipelin':333,361,513,584 'plan':190,204,299,583,740 'planner':682 'posit':418 'present':411 'previous':765 'process':247 'prompt':588,630,658,668 'protect':310,441,501 'push':420,427 'qualiti':316,446 'question':408,736 'quick':62,175,197,214,298 'quick-fix':213 'quick/plan/ship/research/verify':20 'radius':570 'random':250 're':30,272,396 're-ask':271 're-ground':395 'read':93,100,123,132,152,160,166,176,185,191,280,303,477,494 'receiv':587 'recommend':415 'recoveri':534 'refactor':140,144,344 'refer':464,473 'reference/agent-control.md':530 'reference/config.md':535 'reference/context-mode.md':544 'reference/design.md':557 'reference/doc-manifest.md':574 'reference/graph.md':563 'reference/learning-loop.md':516 'reference/orchestrator-laws.md':304,453,495 'reference/orchestrator-rules.md':508 'reference/session-state.md':523 'reference/workspace.md':552 'repo':449 'report':766 'request':619 'requir':2 'research':118,122,337,735 'resolut':320 'resolv':117,750 'retry/correction':687 'review':23,425,690,697 'review/profile/teach/steer':164 'risk':509 'rout':18,53,76,196,286 'rule':315,389,505,610,650 'run':148 'schema':539 'scope':173,756 'score':510 'scrape':158 'screenshot':157 'section':725 'self':355,369 'self-contain':354,368 'session':223,242,254,524 'set':265 'share':625 'ship':181,184,300,739 'shipper':738 'signal':85,532 'silent':72 'skill':129 'skill-mint' 'skills/mint/modes':289 'slop':47 'sloppifi':728 'source-3li7alaki' 'spec':208,621,642,683,689,691,708,731 'specif':475 'ssh':146,151,340 'stabl':251 'stage':695 'staging/prod':150 'startup':490 'state':224 'static':595,602 'status':311,442,502 'step':74,220,274,371,456,487,580 'stop':263,430 'stop/pause/freeze':531 'success':378 'summari':723 'system':667 'take':416 'task':6,45,78,169,234,376,399 'tdd':686 'templates/agent-context.md':634,673 'test':712 'tester':707 'text':66 'timestamp':249 'topic':476 'topic-agent-skills' 'topic-agentic-workflow' 'topic-agents' 'topic-claude-code' 'topic-development' 'touch':170,329 'trace':572 'true':232 'twice':429 'two':593 'univers':388 'unus':139 'user':210,394,424,436 'valu':646 'verifi':55,88,92,338,384,744 'wave':624 'win':84,521 'workflow':127 'workspace.repos':554 'worktre':716 'write':9,222,227 'xml':643,684,692,709,732 'zero':46","prices":[{"id":"2a1a18f4-10e1-4ed2-bb6d-24955e27fcaa","listingId":"2ba263e8-21aa-4b12-b0ba-a24a987fbf5d","amountUsd":"0","unit":"free","nativeCurrency":null,"nativeAmount":null,"chain":null,"payTo":null,"paymentMethod":"skill-free","isPrimary":true,"details":{"org":"3li7alaki","category":"mint","install_from":"skills.sh"},"createdAt":"2026-05-18T13:13:49.265Z"}],"sources":[{"listingId":"2ba263e8-21aa-4b12-b0ba-a24a987fbf5d","source":"github","sourceId":"3li7alaki/mint/mint","sourceUrl":"https://github.com/3li7alaki/mint/tree/main/skills/mint","isPrimary":false,"firstSeenAt":"2026-05-18T13:13:49.265Z","lastSeenAt":"2026-05-18T19:08:20.604Z"}],"details":{"listingId":"2ba263e8-21aa-4b12-b0ba-a24a987fbf5d","quickStartSnippet":null,"exampleRequest":null,"exampleResponse":null,"schema":null,"openapiUrl":null,"agentsTxtUrl":null,"citations":[],"useCases":[],"bestFor":[],"notFor":[],"kindDetails":{"org":"3li7alaki","slug":"mint","github":{"repo":"3li7alaki/mint","stars":10,"topics":["agent-skills","agentic-workflow","agents","ai","claude-code","development"],"license":"mit","html_url":"https://github.com/3li7alaki/mint","pushed_at":"2026-04-19T11:02:49Z","description":"Disciplined agentic development for Claude Code.","skill_md_sha":"0d11d92fd23162622b53ee89ec52c4614ed5f2ea","skill_md_path":"skills/mint/SKILL.md","default_branch":"main","skill_tree_url":"https://github.com/3li7alaki/mint/tree/main/skills/mint"},"layout":"multi","source":"github","category":"mint","frontmatter":{"name":"mint","description":"REQUIRED for ALL coding tasks. Invoke BEFORE writing any code or editing any file. Auto-routes to quick/plan/ship/research/verify. Enforces gates, reviews, atomic commits. No exceptions — if you're about to modify files, invoke mint first."},"skills_sh_url":"https://skills.sh/3li7alaki/mint/mint"},"updatedAt":"2026-05-18T19:08:20.604Z"}}