[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7EMDTW5K6NZDVJASQJETA44":3,"skill-same:01M7EMDTW5K6NZDVJASQJETA44":47},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":25,"commitSha":26,"skillDirectory":27,"forksCount":28,"repositoryHtmlUrl":29,"repositoryHomepageUrl":6,"ownerHtmlUrl":30,"ownerBlogUrl":31,"ownerDescription":6,"updatedAt":32,"install":33,"resources":46},"01M7EMDTW5K6NZDVJASQJETA44","brain-computer-interfaces",null,"BCI design and evaluation — signal decoding, calibration, closed-loop paradigms, and performance metrics.","TanNhatCMS\u002Fjarvis","TanNhatCMS","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F113796420?v=4","01M753MV17KG5CD3152G3WQEQY","KietNT","01M7EKEDYVYQ6PWVJJMZ911VMX",0,"MIT","https:\u002F\u002Fgithub.com\u002FTanNhatCMS\u002Fjarvis\u002Fblob\u002Fmain\u002Fskills\u002Fams-brain-computer-interfaces\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: brain-computer-interfaces\ndescription: BCI design and evaluation — signal decoding, calibration, closed-loop paradigms, and performance metrics.\ncategory: scientific\n---\n\n## Overview\n\nbrain-computer-interfaces covers the engineering and neuroscience of systems that translate neural\nactivity into control signals: invasive ( Utah arrays, Neuropixels, ECoG) and non-invasive (EEG,\nfNIRS) BCIs for communication, motor restoration, and control. It focuses on the closed loop —\ndecoding must work in real time, adapt to non-stationary brains, and be evaluated on metrics that\nreflect actual usability, not offline accuracy.\n\nA BCI is not a decoder; it is a system comprising signal acquisition, preprocessing, decoding,\nfeedback, and a user learning to use it. Failures usually live in the loop, not the classifier.\n\n## When to use\n\n- Choosing a BCI paradigm: motor imagery, P300 spellers, SSVEP, ECoG high-gamma, intracortical\n  kinematic decoding.\n- Designing calibration: open-loop training data collection, session length, avoiding user fatigue.\n- Building the real-time pipeline: latency budgets, feature extraction, decoder updates.\n- Closed-loop decoder adaptation: CLDA, recalibration, handling non-stationarity.\n- Evaluation: information transfer rate, bit rate, task success, user-centered metrics.\n- Safety and ethics: informed consent, data privacy, managing user expectations.\n\n## Core concepts\n\n- **Paradigms.** Motor imagery (EEG mu\u002Fbeta desynchronization — slow, needs training); P300\n  (oddball responses — no training, slow spelling); SSVEP (flicker-driven — fast, fatiguing);\n  ECoG high-gamma (excellent spatial specificity, requires implants); intracortical (single-unit\n  kinematics — highest performance, surgical risk). Match paradigm to user capability and goal.\n- **The non-stationarity problem.** Neural signals drift within and across sessions (electrode\n  shifts, learning, fatigue). Decoders trained once decay. Solutions: daily recalibration,\n  adaptive decoders (closed-loop decoder adaptation), and features robust to drift.\n- **Latency budget.** Total loop delay (acquisition → processing → feedback) must stay low —\n  \u003C100-300 ms for motor BCIs, or the user cannot close the loop. Offline accuracy with 2-second\n  windows is irrelevant if the real-time system lags.\n- **Calibration design.** Collect labeled data with the actual task the user will perform;\n  keep sessions short (fatigue degrades both signal and motivation); interleave rest. More\n  calibration data is not always better — stale data hurts adaptive decoders.\n- **Closed-loop adaptation (CLDA).** Update decoder parameters during use, guided by the user's\n  intended vs decoded output. This co-adaptation (brain + algorithm learning together) is what\n  makes chronic BCIs work — but unstable updates can diverge, so bound the learning rate.\n- **Evaluation metrics.** Information transfer rate (bits\u002Fmin) for communication; target\n  acquisition time and success rate for motor tasks; and crucially, metrics measured closed-loop\n  with the user in the loop — offline cross-validation overestimates real performance.\n- **User learning.** The brain adapts to the decoder. Provide consistent, immediate feedback;\n  keep the mapping stable enough to learn but adaptive enough to track drift. Report learning\n  curves, not just final performance.\n- **Ethics and expectations.** Neural data is uniquely sensitive (medical, potentially\n  identifiable). Consent must cover data reuse; never promise restored function a system cannot\n  deliver — hype harms patients.\n\n## Practical workflow\n\n1. **Define the task.** What will the user actually do (spell, move a cursor, control a robotic\n   arm)? Define success in user terms first, then derive engineering specs.\n2. **Pick modality + paradigm.** Non-invasive for accessibility, invasive for performance; choose\n   the paradigm that fits the user's residual abilities.\n3. **Build the real-time chain.** Acquisition → preprocessing (causal filters only — no filtfilt\n   in real time) → features → decoder → feedback. Measure end-to-end latency before any user\n   testing.\n4. **Calibrate.** Short open-loop session with the real task; train an initial decoder; verify\n   above-chance closed-loop control in the same session.\n5. **Close the loop with adaptation.** Enable CLDA with conservative updates; monitor for\n   divergence; schedule recalibration.\n6. **Evaluate properly.** Closed-loop metrics over multiple sessions: ITR, success rate, learning\n   curves, user workload\u002Ffatigue questionnaires. Compare against sensible baselines (e.g. existing\n   assistive tech), not just chance.\n7. **Document.** Decoder architecture, update rules, latency, failure modes, and ethical\n   safeguards. Plan long-term support — abandoning a BCI user is a harm.\n\nExample pipeline sketch:\n```python\n# per incoming chunk (causal, low-latency):\nx = bandpass_causal(chunk, 8, 30)          # no zero-phase filtering live\nfeat = log_bandpower(x, window=250ms)\ncmd = decoder.predict(feat)                # previously calibrated\nclda_update(decoder, cmd, intended)        # bounded adaptation\nrender_feedback(cmd)                       # \u003C200 ms total budget\n```\n\n## Common pitfalls\n\n- Reporting offline cross-validated accuracy as \"BCI performance.\"\n- Non-causal filtering in the real-time path (filtfilt looks into the future).\n- Ignoring latency — a 90%-accurate decoder with 2 s lag is unusable.\n- One-shot calibration with no adaptation plan for drift.\n- Unstable CLDA updates that diverge mid-session.\n- Paradigm-user mismatch (e.g. SSVEP for photosensitive users).\n- Overpromising clinical outcomes; underplanning long-term maintenance and support.\n","## Overview\n\nbrain-computer-interfaces covers the engineering and neuroscience of systems that translate neural\nactivity into control signals: invasive ( Utah arrays, Neuropixels, ECoG) and non-invasive (EEG,\nfNIRS) BCIs for communication, motor restoration, and control. It focuses on the closed loop —\ndecoding must work in real time, adapt to non-stationary brains, and be evaluated on metrics that\nreflect actual usability, not offline accuracy.\n\nA BCI is not a decoder; it is a system comprising signal acquisition, preprocessing, decoding,\nfeedback, and a user learning to use it. Failures usually live in the loop, not the classifier.\n\n## When to use\n\n- Choosing a BCI paradigm: motor imagery, P300 spellers, SSVEP, ECoG high-gamma, intracortical\n  kinematic decoding.\n- Designing calibration: open-loop training data collection, session length, avoiding user fatigue.\n- Building the real-time pipeline: latency budgets, feature extraction, decoder updates.\n- Closed-loop decoder adaptation: CLDA, recalibration, handling non-stationarity.\n- Evaluation: information transfer rate, bit rate, task success, user-centered metrics.\n- Safety and ethics: informed consent, data privacy, managing user expectations.\n\n## Core concepts\n\n- **Paradigms.** Motor imagery (EEG mu\u002Fbeta desynchronization — slow, needs training); P300\n  (oddball responses — no training, slow spelling); SSVEP (flicker-driven — fast, fatiguing);\n  ECoG high-gamma (excellent spatial specificity, requires implants); intracortical (single-unit\n  kinematics — highest performance, surgical risk). Match paradigm to user capability and goal.\n- **The non-stationarity problem.** Neural signals drift within and across sessions (electrode\n  shifts, learning, fatigue). Decoders trained once decay. Solutions: daily recalibration,\n  adaptive decoders (closed-loop decoder adaptation), and features robust to drift.\n- **Latency budget.** Total loop delay (acquisition → processing → feedback) must stay low —\n  \u003C100-300 ms for motor BCIs, or the user cannot close the loop. Offline accuracy with 2-second\n  windows is irrelevant if the real-time system lags.\n- **Calibration design.** Collect labeled data with the actual task the user will perform;\n  keep sessions short (fatigue degrades both signal and motivation); interleave rest. More\n  calibration data is not always better — stale data hurts adaptive decoders.\n- **Closed-loop adaptation (CLDA).** Update decoder parameters during use, guided by the user's\n  intended vs decoded output. This co-adaptation (brain + algorithm learning together) is what\n  makes chronic BCIs work — but unstable updates can diverge, so bound the learning rate.\n- **Evaluation metrics.** Information transfer rate (bits\u002Fmin) for communication; target\n  acquisition time and success rate for motor tasks; and crucially, metrics measured closed-loop\n  with the user in the loop — offline cross-validation overestimates real performance.\n- **User learning.** The brain adapts to the decoder. Provide consistent, immediate feedback;\n  keep the mapping stable enough to learn but adaptive enough to track drift. Report learning\n  curves, not just final performance.\n- **Ethics and expectations.** Neural data is uniquely sensitive (medical, potentially\n  identifiable). Consent must cover data reuse; never promise restored function a system cannot\n  deliver — hype harms patients.\n\n## Practical workflow\n\n1. **Define the task.** What will the user actually do (spell, move a cursor, control a robotic\n   arm)? Define success in user terms first, then derive engineering specs.\n2. **Pick modality + paradigm.** Non-invasive for accessibility, invasive for performance; choose\n   the paradigm that fits the user's residual abilities.\n3. **Build the real-time chain.** Acquisition → preprocessing (causal filters only — no filtfilt\n   in real time) → features → decoder → feedback. Measure end-to-end latency before any user\n   testing.\n4. **Calibrate.** Short open-loop session with the real task; train an initial decoder; verify\n   above-chance closed-loop control in the same session.\n5. **Close the loop with adaptation.** Enable CLDA with conservative updates; monitor for\n   divergence; schedule recalibration.\n6. **Evaluate properly.** Closed-loop metrics over multiple sessions: ITR, success rate, learning\n   curves, user workload\u002Ffatigue questionnaires. Compare against sensible baselines (e.g. existing\n   assistive tech), not just chance.\n7. **Document.** Decoder architecture, update rules, latency, failure modes, and ethical\n   safeguards. Plan long-term support — abandoning a BCI user is a harm.\n\nExample pipeline sketch:\n```python\n# per incoming chunk (causal, low-latency):\nx = bandpass_causal(chunk, 8, 30)          # no zero-phase filtering live\nfeat = log_bandpower(x, window=250ms)\ncmd = decoder.predict(feat)                # previously calibrated\nclda_update(decoder, cmd, intended)        # bounded adaptation\nrender_feedback(cmd)                       # \u003C200 ms total budget\n```\n\n## Common pitfalls\n\n- Reporting offline cross-validated accuracy as \"BCI performance.\"\n- Non-causal filtering in the real-time path (filtfilt looks into the future).\n- Ignoring latency — a 90%-accurate decoder with 2 s lag is unusable.\n- One-shot calibration with no adaptation plan for drift.\n- Unstable CLDA updates that diverge mid-session.\n- Paradigm-user mismatch (e.g. SSVEP for photosensitive users).\n- Overpromising clinical outcomes; underplanning long-term maintenance and support.\n",{"name":5,"category":24,"description":7},"scientific","main","af388a77c2fe5e0b4fd879cca4300a0ffa9a6579","skills\u002Fams-brain-computer-interfaces",1,"https:\u002F\u002Fgithub.com\u002FTanNhatCMS\u002Fjarvis","https:\u002F\u002Fgithub.com\u002FTanNhatCMS","https:\u002F\u002Fmrkiet.name.vn","2026-11-07 14:23:08",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":45},[35,38,41],{"tool":36,"command":37},"npx","npx skills add TanNhatCMS\u002Fjarvis --skill brain-computer-interfaces",{"tool":39,"command":40},"bunx","bunx skills add TanNhatCMS\u002Fjarvis --skill brain-computer-interfaces",{"tool":42,"command":43},"pnpm","pnpm dlx skills add TanNhatCMS\u002Fjarvis --skill brain-computer-interfaces","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FTanNhatCMS\u002Fjarvis\u002Fblob\u002Fmain\u002Fskills\u002Fams-brain-computer-interfaces\u002FSKILL.md",true,[],[48,55,63,70,77,84],{"skillNo":49,"name":50,"nameZh":6,"description":51,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":52,"categories":53,"keywords":54,"publishedAt":19},"01M7EWYJR0J8DARY7E59Z6Y9MV","web-scrape","Turn web pages into LLM-ready markdown via the Firecrawl API (renders JS, anti-bot handling). Use when browser_open\u002Ffetch fails on JS-heavy pages, or when you need clean markdown\u002Fstructured extraction from a site. Requires FIRECRAWL_API_KEY.","https:\u002F\u002Fgithub.com\u002FTanNhatCMS\u002Fjarvis\u002Fblob\u002Fmain\u002Fskills\u002Fweb-scrape\u002FSKILL.md",[],[],{"skillNo":56,"name":57,"nameZh":6,"description":58,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":59,"sourceHtmlUrl":60,"categories":61,"keywords":62,"publishedAt":19},"01M7EWYBN5P6F2AJG3FR24T9BP","design-taste-frontend","tasteskill v2, the anti-slop frontend skill. Art direction for websites, distilled from a study of 247 curated gallery sites (One Page Love, Awwwards Sites of the Day). Deals researched directions (idea, composition, type, palette, photographs, section blueprints) from a library, enforces a strict no-slop house style with a real-browser inspector, and runs a look-and-fix loop on screenshots. Use this whenever you build, design, redesign, restyle, polish or review a landing page, marketing site, portfolio, product page, one-pager, hero or any public-facing web page, in any stack (HTML\u002FCSS, React, Next.js, Vue, Svelte, Astro, Tailwind) — also when the user says a page looks generic, templated, \"AI-made\", sloppy, or asks to make it look designed, premium, award-level or less like every other site. Not for dashboards or app screens.","https:\u002F\u002Fgithub.com\u002FTanNhatCMS\u002Fjarvis\u002Fblob\u002Fmain\u002Fskills\u002Ftaste-design\u002FLICENSE","https:\u002F\u002Fgithub.com\u002FTanNhatCMS\u002Fjarvis\u002Fblob\u002Fmain\u002Fskills\u002Ftaste-design\u002FSKILL.md",[],[],{"skillNo":64,"name":65,"nameZh":6,"description":66,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":67,"categories":68,"keywords":69,"publishedAt":19},"01M7EWY8FQFNQ2GB7NZZWGSSDX","test-driven-development","Use when implementing any feature or bugfix, before writing implementation code","https:\u002F\u002Fgithub.com\u002FTanNhatCMS\u002Fjarvis\u002Fblob\u002Fmain\u002Fskills\u002Fsuperpowers-tdd\u002FSKILL.md",[],[],{"skillNo":71,"name":72,"nameZh":6,"description":73,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":74,"categories":75,"keywords":76,"publishedAt":19},"01M7EWY5SHBFRH1HDHVWM9K7TV","systematic-debugging","Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes","https:\u002F\u002Fgithub.com\u002FTanNhatCMS\u002Fjarvis\u002Fblob\u002Fmain\u002Fskills\u002Fsuperpowers-debugging\u002FSKILL.md",[],[],{"skillNo":78,"name":79,"nameZh":6,"description":80,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":81,"categories":82,"keywords":83,"publishedAt":19},"01M7EWY3NM0Z0JAKS7HDZ24K66","requesting-code-review","Use when completing tasks, implementing major features, or before merging to verify work meets requirements","https:\u002F\u002Fgithub.com\u002FTanNhatCMS\u002Fjarvis\u002Fblob\u002Fmain\u002Fskills\u002Fsuperpowers-code-review\u002FSKILL.md",[],[],{"skillNo":85,"name":86,"nameZh":6,"description":87,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":88,"categories":89,"keywords":90,"publishedAt":19},"01M7EWY0WRMWX4M9ZFY0WKC593","brainstorming","You MUST use this before any creative work - creating features, building components, adding functionality, modifying behavior, or planning anything new, in software or out of it (a talk, a business, a renovation).","https:\u002F\u002Fgithub.com\u002FTanNhatCMS\u002Fjarvis\u002Fblob\u002Fmain\u002Fskills\u002Fsuperpowers-brainstorming\u002FSKILL.md",[],[]]