{"id":"5fe4ab5e-8745-4a77-9381-08d97fe46ec5","shortId":"knWm2w","kind":"skill","title":"review-implementation","tagline":"Review implementation against provided markdown requirements, specs, PRDs, or phase docs; optionally apply focused fixes. Use when a user supplies .md file paths and asks to verify correctness/completeness, check PRD phase completion, find missing or overbuilt work, assess critic","description":"# Review Implementation\n\n## Overview\n\nReview code against markdown requirements and produce evidence-backed findings.\nWhen the user asks for fixes, apply targeted changes that bring the\nimplementation back to the docs without expanding scope.\n\n## Modes\n\n- Use **review-only mode** when the user asks to review, assess, audit, verify,\n  or find issues.\n- Use **fix mode** when the user asks to fix, update, refactor, complete,\n  apply changes, or make the implementation match the docs.\n- In fix mode, keep changes minimal and reversible. Do not add features that are\n  not justified by the docs.\n\n## Workflow\n\n### 1) Intake and Discovery Gate\n\n- Read all provided .md files and extract explicit requirements, constraints, and acceptance criteria.\n- Read repo guidance such as AGENTS.md, README, architecture notes, and existing\n  task or PRD files when relevant.\n- Inspect affected code, tests, fixtures, routes, schemas, migrations, services,\n  components, permissions, config, and observability surfaces as needed.\n- Identify the current behavior, expected behavior, data flow, integration\n  points, validation options, and likely blast radius.\n- Ask only when a critical decision cannot be made safely after discovery.\n\n### 2) Map Requirements to Evidence\n\n- Build an internal traceability map: requirement -> status -> code evidence ->\n  gap -> action.\n- Classify each requirement as satisfied, partial, missing, conflicting,\n  overbuilt, or deferred.\n- Use file/line evidence for important claims. Do not rely on general\n  impressions when code can be inspected.\n\n### 3) Run Multi-Pass Review\n\nReview in this order:\n\n1. Requirements coverage: every requirement is satisfied or explicitly\n   unresolved.\n2. Correctness: happy paths, edge cases, errors, empty states, permissions,\n   state transitions, and rollback behavior are handled.\n3. Integration: changed modules fit together without contract breaks,\n   duplicated ownership, or hidden assumptions.\n4. Simplicity: the solution is no more complex than necessary.\n5. Cleanup: repeated logic, dead code, temporary code, noisy logs, unused files,\n   and unused dependencies are removed.\n6. Security/privacy: auth, access control, secrets, sensitive data, injection\n   risks, and audit needs are safe.\n7. Performance: expensive queries, N+1 patterns, unnecessary renders, redundant\n   network calls, and blocking work are addressed.\n8. Validation: chosen checks are appropriate for the risk.\n9. Documentation/operability: docs, release notes, migrations, rollback,\n   monitoring, or support notes are updated when required.\n\nTreat overbuilt code as a review finding when it adds features, abstractions,\nstate, dependencies, or workflow paths not required by the docs.\n\n### 4) Apply Focused Fixes\n\n- Fix issues directly in code with minimal, targeted changes.\n- Split overly long files/functions only when the split reduces real review or\n  maintenance risk.\n- Remove or simplify overbuilt code that is not justified by the docs.\n- Preserve existing patterns unless the docs or code evidence justify a change.\n- If PRD or phase files are part of the request, update checkboxes, validation\n  notes, discoveries, and change logs only when implementation evidence supports\n  the update.\n\n### 5) Validate with Evidence\n\n- Choose the smallest sufficient validation for the risk: static checks, unit\n  tests, integration tests, API-level E2E, browser/UI checks, simulator checks,\n  screenshots, manual smoke checks, or observability checks.\n- Run relevant checks when available and appropriate. If checks are unavailable,\n  too costly, or not allowed, state the gap and use the best available evidence.\n- Do not run broad or expensive validation by reflex when a narrower check proves\n  the changed behavior.\n\n## Report Format\n\nLead with the result that matters most:\n\n- In review-only mode, list findings first in severity order with file/line\n  references, then summarize coverage, validation, and residual risk.\n- In fix mode, summarize fixes applied, validation performed, and any remaining\n  findings or risks.\n- If no issues are found, say so clearly and name any validation gaps.\n- For critical blockers, stop and ask. For non-critical ambiguity, make the best\n  reasonable decision, record the assumption, and continue.","tags":["review","implementation","agent","skills","golbin","agent-skills","code-review","codex","implementation-review","prd","review-implementation","skills-sh"],"capabilities":["skill","source-golbin","skill-review-implementation","topic-agent-skills","topic-code-review","topic-codex","topic-implementation-review","topic-prd","topic-review-implementation","topic-skills-sh"],"categories":["agent-skills"],"synonyms":[],"warnings":[],"endpointUrl":"https://skills.sh/golbin/agent-skills/review-implementation","protocol":"skill","transport":"skills-sh","auth":{"type":"none","details":{"cli":"npx skills add golbin/agent-skills","source_repo":"https://github.com/golbin/agent-skills","install_from":"skills.sh"}},"qualityScore":"0.508","qualityRationale":"deterministic score 0.51 from registry signals: · indexed on github topic:agent-skills · 117 github stars · SKILL.md body (4,275 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":"2026-04-22T00:55:20.923Z","embedding":null,"createdAt":"2026-04-20T19:37:20.181Z","updatedAt":"2026-04-22T00:55:20.923Z","lastSeenAt":"2026-04-22T00:55:20.923Z","tsv":"'+1':358 '1':136,270 '2':216,280 '3':260,297 '4':311,416 '5':321,492 '6':338 '7':353 '8':370 '9':379 'abstract':405 'accept':152 'access':341 'action':231 'add':126,403 'address':369 'affect':172 'agents.md':159 'allow':540 'ambigu':634 'api':511 'api-level':510 'appli':16,63,107,417,602 'appropri':375,531 'architectur':161 'ask':28,60,86,101,204,629 'assess':41,89 'assumpt':310,642 'audit':90,349 'auth':340 'avail':529,548 'back':55,70 'behavior':191,193,294,566 'best':547,637 'blast':202 'block':366 'blocker':626 'break':305 'bring':67 'broad':553 'browser/ui':514 'build':221 'call':364 'cannot':210 'case':285 'chang':65,108,120,299,428,466,483,565 'check':32,373,505,515,517,521,524,527,533,562 'checkbox':478 'choos':496 'chosen':372 'claim':248 'classifi':232 'cleanup':322 'clear':618 'code':47,173,228,256,326,328,396,424,447,462 'complet':35,106 'complex':318 'compon':180 'config':182 'conflict':239 'constraint':150 'continu':644 'contract':304 'control':342 'correct':281 'correctness/completeness':31 'cost':537 'coverag':272,592 'criteria':153 'critic':42,208,625,633 'current':190 'data':194,345 'dead':325 'decis':209,639 'defer':242 'depend':335,407 'direct':422 'discoveri':139,215,481 'doc':14,73,115,134,381,415,454,460 'documentation/operability':380 'duplic':306 'e2e':513 'edg':284 'empti':287 'error':286 'everi':273 'evid':54,220,229,245,463,488,495,549 'evidence-back':53 'exist':164,456 'expand':75 'expect':192 'expens':355,555 'explicit':148,278 'extract':147 'featur':127,404 'file':25,145,168,332,471 'file/line':244,588 'files/functions':432 'find':36,56,93,400,582,608 'first':583 'fit':301 'fix':18,62,96,103,117,419,420,598,601 'fixtur':175 'flow':195 'focus':17,418 'format':568 'found':615 'gap':230,543,623 'gate':140 'general':253 'guidanc':156 'handl':296 'happi':282 'hidden':309 'identifi':188 'implement':3,5,44,69,112,487 'import':247 'impress':254 'inject':346 'inspect':171,259 'intak':137 'integr':196,298,508 'intern':223 'issu':94,421,613 'justifi':131,451,464 'keep':119 'lead':569 'level':512 'like':201 'list':581 'log':330,484 'logic':324 'long':431 'made':212 'mainten':441 'make':110,635 'manual':519 'map':217,225 'markdown':8,49 'match':113 'matter':574 'md':24,144 'migrat':178,384 'minim':121,426 'miss':37,238 'mode':77,82,97,118,580,599 'modul':300 'monitor':386 'multi':263 'multi-pass':262 'n':357 'name':620 'narrow':561 'necessari':320 'need':187,350 'network':363 'noisi':329 'non':632 'non-crit':631 'note':162,383,389,480 'observ':184,523 'option':15,199 'order':269,586 'over':430 'overbuilt':39,240,395,446 'overview':45 'ownership':307 'part':473 'partial':237 'pass':264 'path':26,283,410 'pattern':359,457 'perform':354,604 'permiss':181,289 'phase':13,34,470 'point':197 'prd':33,167,468 'prds':11 'preserv':455 'produc':52 'prove':563 'provid':7,143 'queri':356 'radius':203 'read':141,154 'readm':160 'real':438 'reason':638 'record':640 'reduc':437 'redund':362 'refactor':105 'refer':589 'reflex':558 'releas':382 'relev':170,526 'reli':251 'remain':607 'remov':337,443 'render':361 'repeat':323 'repo':155 'report':567 'request':476 'requir':9,50,149,218,226,234,271,274,393,412 'residu':595 'result':572 'revers':123 'review':2,4,43,46,80,88,265,266,399,439,578 'review-implement':1 'review-on':79,577 'risk':347,378,442,503,596,610 'rollback':293,385 'rout':176 'run':261,525,552 'safe':213,352 'satisfi':236,276 'say':616 'schema':177 'scope':76 'screenshot':518 'secret':343 'security/privacy':339 'sensit':344 'servic':179 'sever':585 'simplic':312 'simplifi':445 'simul':516 'skill' 'skill-review-implementation' 'smallest':498 'smoke':520 'solut':314 'source-golbin' 'spec':10 'split':429,436 'state':288,290,406,541 'static':504 'status':227 'stop':627 'suffici':499 'summar':591,600 'suppli':23 'support':388,489 'surfac':185 'target':64,427 'task':165 'temporari':327 'test':174,507,509 'togeth':302 'topic-agent-skills' 'topic-code-review' 'topic-codex' 'topic-implementation-review' 'topic-prd' 'topic-review-implementation' 'topic-skills-sh' 'traceabl':224 'transit':291 'treat':394 'unavail':535 'unit':506 'unless':458 'unnecessari':360 'unresolv':279 'unus':331,334 'updat':104,391,477,491 'use':19,78,95,243,545 'user':22,59,85,100 'valid':198,371,479,493,500,556,593,603,622 'verifi':30,91 'without':74,303 'work':40,367 'workflow':135,409","prices":[{"id":"c0e7c7a3-000d-42aa-bed2-e3762990b697","listingId":"5fe4ab5e-8745-4a77-9381-08d97fe46ec5","amountUsd":"0","unit":"free","nativeCurrency":null,"nativeAmount":null,"chain":null,"payTo":null,"paymentMethod":"skill-free","isPrimary":true,"details":{"org":"golbin","category":"agent-skills","install_from":"skills.sh"},"createdAt":"2026-04-20T19:37:20.181Z"}],"sources":[{"listingId":"5fe4ab5e-8745-4a77-9381-08d97fe46ec5","source":"github","sourceId":"golbin/agent-skills/review-implementation","sourceUrl":"https://github.com/golbin/agent-skills/tree/main/skills/review-implementation","isPrimary":false,"firstSeenAt":"2026-04-20T19:37:20.181Z","lastSeenAt":"2026-04-22T00:55:20.923Z"}],"details":{"listingId":"5fe4ab5e-8745-4a77-9381-08d97fe46ec5","quickStartSnippet":null,"exampleRequest":null,"exampleResponse":null,"schema":null,"openapiUrl":null,"agentsTxtUrl":null,"citations":[],"useCases":[],"bestFor":[],"notFor":[],"kindDetails":{"org":"golbin","slug":"review-implementation","github":{"repo":"golbin/agent-skills","stars":117,"topics":["agent-skills","code-review","codex","implementation-review","prd","product-requirements-document","review-implementation","skills-sh"],"license":null,"html_url":"https://github.com/golbin/agent-skills","pushed_at":"2026-04-20T05:10:10Z","description":"Reusable agent skills for Codex and compatible tools","skill_md_sha":"37085c80d702b3363502da2c8bc35d9810a2c571","skill_md_path":"skills/review-implementation/SKILL.md","default_branch":"main","skill_tree_url":"https://github.com/golbin/agent-skills/tree/main/skills/review-implementation"},"layout":"multi","source":"github","category":"agent-skills","frontmatter":{"name":"review-implementation","description":"Review implementation against provided markdown requirements, specs, PRDs, or phase docs; optionally apply focused fixes. Use when a user supplies .md file paths and asks to verify correctness/completeness, check PRD phase completion, find missing or overbuilt work, assess critical issues, refactor long files/functions, or update code to match the docs."},"skills_sh_url":"https://skills.sh/golbin/agent-skills/review-implementation"},"updatedAt":"2026-04-22T00:55:20.923Z"}}