{"id":52617,"date":"2026-05-01T09:01:55","date_gmt":"2026-05-01T01:01:55","guid":{"rendered":"https:\/\/www.1ai.net\/?p=52617"},"modified":"2026-04-30T17:04:21","modified_gmt":"2026-04-30T09:04:21","slug":"hermes%e9%ab%98%e7%ba%a7%e7%94%a8%e6%b3%95%ef%bc%9a%e7%94%a8%e5%a4%9a-profile-%e5%8d%8f%e4%bd%9c-wiki-%e5%85%b1%e4%ba%ab%e8%ae%b0%e5%bf%86%ef%bc%8c%e6%90%ad%e5%bb%ba%e4%bd%a0%e7%9a%84-opc-agent-2","status":"publish","type":"post","link":"https:\/\/www.1ai.net\/en\/52617.html","title":{"rendered":"Hermes Advanced Use: Work with MultiProfile + Wiki Sharing Memory to Build Your OPC Agent Team (Next)"},"content":{"rendered":"<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-52618\" title=\"e51b07bj00teaugq007d000iw007kp\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2026\/04\/fe51b07bj00teaugq0074d000iw007kp.jpg\" alt=\"e51b07bj00teaugq007d000iw007kp\" width=\"680\" height=\"272\" \/><\/p>\n<p>We talked last time <a href=\"https:\/\/www.1ai.net\/en\/52615.html\/\">Hermes Advanced Use: Working with MultiProfile + Wiki Sharing Memory to Build Your OPC Agent Team (Previously)<\/a>.<\/p>\n<p>The core view is:<\/p>\n<blockquote>\n<ul>\n<li>Don't let an Agent take on all roles at the same time\u3002<\/li>\n<\/ul>\n<\/blockquote>\n<p>A more stable way to organize Agent into a small team:<\/p>\n<ul>\n<li>Coordinator coordinates<\/li>\n<li>Researcher is in charge of research<\/li>\n<li>Writer is responsible for expression<\/li>\n<li>Builder builds\u3002<\/li>\n<\/ul>\n<p>But this raises a new problem:<\/p>\n<blockquote>\n<ul>\n<li>How do they work together in the long term<\/li>\n<\/ul>\n<\/blockquote>\n<p>The answer is:<\/p>\n<blockquote>\n<ul>\n<li>Use Wiki as a shared memory layer\u3002<\/li>\n<\/ul>\n<\/blockquote>\n<p>I. Wiki, what is it<\/p>\n<p>In this system, Wiki is not an ordinary notebook\u3002<\/p>\n<p>It's the whole thing <a href=\"https:\/\/www.1ai.net\/en\/tag\/opc\" title=\"[SEE ARTICLES WITH [OPC] LABELS]\" target=\"_blank\" >OPC<\/a> System:<\/p>\n<ul>\n<li>Sharing the knowledge base<\/li>\n<li>Project management<\/li>\n<li>Long-term memory layer\u3002<\/li>\n<\/ul>\n<p>It manages three types of things:<\/p>\n<ul>\n<li>Knowledge<\/li>\n<li>Information<\/li>\n<li>Outputs\u3002<\/li>\n<\/ul>\n<p>More specifically, Wiki is responsible for recording:<\/p>\n<ul>\n<li>Project background<\/li>\n<li>Mission status<\/li>\n<li>Advance logs<\/li>\n<li>Key decision-making<\/li>\n<li>Intermediate material<\/li>\n<li>Final outputs<\/li>\n<li>Cross-project methodology<\/li>\n<li>Source material\u3002<\/li>\n<\/ul>\n<p>It can be understood as a corporate shared document system\u3002<\/p>\n<p>In this system:<\/p>\n<ul>\n<li>Profile is an employee<\/li>\n<li>Project is the project office<\/li>\n<li>Wiki is the company database<\/li>\n<li>Cordinator is the master control\u3002<\/li>\n<\/ul>\n<p>So, multiple Profiles can work together through Wiki even if their memories are not connected\u3002<\/p>\n<p>II. Overall structure of Wiki<\/p>\n<p>A complete Wiki, with eight parts:<\/p>\n<p>op. cit.mdschema.mdsystem\/projects\/pages\/raw\/assets\/archive\/<\/p>\n<p>Pasted image 202604282144948.png appears to have a large number of parts, but each layer has a clear role to play\u3002<\/p>\n<p>Together, they address a core issue:<\/p>\n<blockquote>\n<ul>\n<li>Where should I put the information<\/li>\n<\/ul>\n<\/blockquote>\n<p>If you don't have layers, everything will mix\u3002<\/p>\n<p>for example:<\/p>\n<ul>\n<li>The status of the project will contaminate user images<\/li>\n<li>Ideas can contaminate long-term knowledge<\/li>\n<li>Role experience contaminates project rules<\/li>\n<li>The source information will be mixed with the final conclusions\u3002<\/li>\n<\/ul>\n<p>So the first principle of Wiki is:<\/p>\n<blockquote>\n<ul>\n<li>The layer, it's anti-pollution\u3002<\/li>\n<\/ul>\n<\/blockquote>\n<p>III.index.md: Wiki entrance<\/p>\n<p>index.md is the front page of the entire Wiki\u3002<\/p>\n<p>It doesn't have specific content, it does navigation\u3002<\/p>\n<p>It tells people and Agent:<\/p>\n<ul>\n<li>Where is the system area<\/li>\n<li>What are the current projects<\/li>\n<li>Where is the general knowledge<\/li>\n<li>Where are the originals<\/li>\n<li>What's the point lately\u3002<\/li>\n<\/ul>\n<p>the role of index.md is to get any Profile into Wiki and know where to start reading\u3002<\/p>\n<p>It's a map, not a warehouse\u3002<\/p>\n<p>IV. SCHEma.Md: Wiki Norms<\/p>\n<p>schema.md is Wiki's Constitution\u3002<\/p>\n<p>It provides that:<\/p>\n<ul>\n<li>How the document is named<\/li>\n<li>Where does it go<\/li>\n<li>Which documents are read-only<\/li>\n<li>Which documents could be modified<\/li>\n<li>What information cannot be fabricated<\/li>\n<li>How the page status is marked\u3002<\/li>\n<\/ul>\n<p>With the schema.md, the whole Wiki will be stable for the long term\u3002<\/p>\n<p>Otherwise, every Profile will write a document in its own sense, and it will soon be a mess\u3002<\/p>\n<p>for example, schema.md can provide that:<\/p>\n<ul>\n<li>Raw read-only<\/li>\n<li>Project documents only record information on corresponding items<\/li>\n<li>Pages records only cross-project reusable methodology<\/li>\n<li>System records only global management information<\/li>\n<li>uncertain information goes first inbox, not directly to pages\u3002<\/li>\n<\/ul>\n<p>That's the rule layer\u3002<\/p>\n<p>v. system: global management<\/p>\n<p>system is the Wiki global management area\u3002<\/p>\n<p>It is not a specific project, but a system-wide management hub\u3002<\/p>\n<p>the system contains six core documents:<\/p>\n<p>dashboard.mdagent-log.mdweekly-review.mdmemory-routing.mdskill-registry.mduser-profile.md<\/p>\n<p>1. dashboard.md: master panel<\/p>\n<p>dashboard.md records the current status of all items\u3002<\/p>\n<p>It replied:<\/p>\n<ul>\n<li>What are the projects<\/li>\n<li>How has each project progressed<\/li>\n<li>What's next<\/li>\n<li>Who's in charge<\/li>\n<li>Are there any obstacles<\/li>\n<\/ul>\n<p>dashboard.md should normally only be updated by Cordinator\u3002<\/p>\n<p>Other Profile can read, but do not modify\u3002<\/p>\n<p>This would avoid conflict between multiple players and the general control panel\u3002<\/p>\n<p>2. Agent-log.md: Global Age Behaviour Log<\/p>\n<p>agent-log.md records all Profile's done\u3002<\/p>\n<p>for example:<\/p>\n<ul>\n<li>What did Researcher check<\/li>\n<li>Writer wrote a draft<\/li>\n<li>What documents did Builder deliver<\/li>\n<li>Cordinator updated what status\u3002<\/li>\n<\/ul>\n<p>It's the role of making more Profile behavior traceable\u3002<\/p>\n<p>When the system is running long, you can look back:<\/p>\n<ul>\n<li>Who did what when<\/li>\n<li>Where are the outputs<\/li>\n<li>Mandate completed\u3002<\/li>\n<\/ul>\n<p>THIS IS CRITICAL FOR THE LONG-TERM OPC SYSTEM\u3002<\/p>\n<p>3. weekly-review.md: cycled<\/p>\n<p>weekly-review.md is not just a record\u3002<\/p>\n<p>Its role is to translate a week of action into strategy\u3002<\/p>\n<p>It should summarize:<\/p>\n<ul>\n<li>What's done this week<\/li>\n<li>What's stuck<\/li>\n<li>What lessons can be reused<\/li>\n<li>Which decisions need to be adjusted<\/li>\n<li>What's the point next week\u3002<\/li>\n<\/ul>\n<p>A TRULY VALUABLE OPC SYSTEM IS NOT JUST A MISSION, BUT A WAY TO SINK FROM IT\u3002<\/p>\n<p>weekly-reviewed.md is where this is done\u3002<\/p>\n<p>4. memory-routing.md: memory writing rules<\/p>\n<p>mine-routing.md is a core anti-pollution document\u3002<\/p>\n<p>It specifies where different information should be written\u3002<\/p>\n<p>for example:<\/p>\n<ul>\n<li>role identity written in soul.md<\/li>\n<li>Long-term user preference for writing USER.md or system\/user-profile.md<\/li>\n<li>role common experience written in memory.md<\/li>\n<li>Project rules are written in AGENTS.md<\/li>\n<li>project background written in context.md<\/li>\n<li>project tasks are written in questions.md<\/li>\n<li>project process written log.md<\/li>\n<li>project decision-making is written into articles.md<\/li>\n<li>interim material written inbox\/<\/li>\n<li>official outputs are written outputs\/<\/li>\n<li>cross-project methodology writing pages\/<\/li>\n<li>source written in raw\/\u3002<\/li>\n<\/ul>\n<p>This set of rules determines whether the system can work in a stable manner over the long term\u3002<\/p>\n<p>5. skill-registry.md: skills inventory<\/p>\n<p>skill-registry.md records what reusable skills are available throughout the system and to whom they should be allocated\u3002<\/p>\n<p>It can prevent a problem:<\/p>\n<blockquote>\n<ul>\n<li>All Profiles are equipped with a bunch of skills, and finally the role boundaries are confused again\u3002<\/li>\n<\/ul>\n<\/blockquote>\n<p>for example:<\/p>\n<ul>\n<li>dashboard-update should only be given to Coordinator<\/li>\n<li>you should give it to Researcher<\/li>\n<li>anticle-structure should be given to Writer<\/li>\n<li>code-builder should be given to Builder\u3002<\/li>\n<\/ul>\n<p>High-risk skills such as:<\/p>\n<ul>\n<li>delete-files<\/li>\n<li>dploy-project<\/li>\n<li>bulk-memoory-edit<\/li>\n<\/ul>\n<p>It needs to be strictly limited\u3002<\/p>\n<p>skill-registry.md's role is to make the distribution of capacity manageable\u3002<\/p>\n<p>6. our-profile.md: unified user image<\/p>\n<p>user-profile.md records long-term user preferences that all Profile should know about\u3002<\/p>\n<p>for example:<\/p>\n<ul>\n<li>User preference for communication in Chinese<\/li>\n<li>Users prefer straightforward, clear and critical explanations<\/li>\n<li>Users commonly use Obsidian<\/li>\n<li>Users like Markdown<\/li>\n<li>Long-term attention of users <a href=\"https:\/\/www.1ai.net\/en\/tag\/hermes\" title=\"_Other Organiser\" target=\"_blank\" >Hermes<\/a>, LLM Wiki, OPC, MultiAgent\u3002<\/li>\n<\/ul>\n<p>It's not like every Profile's own USER.md\u3002<\/p>\n<p>The USER.md in Profile is the user's understanding of the individual role\u3002<\/p>\n<p>system\/user-policy.md is a user image source shared throughout the system\u3002<\/p>\n<p>vi. projects: long-term project layer<\/p>\n<p>system is responsible for global management\u3002<\/p>\n<p>projects are responsible for specific projects\u3002<\/p>\n<p>Each long-term project should have its own project space\u3002<\/p>\n<p>for example:<\/p>\n<p>projects\/twitter-rowth\/projects\/vibe-coding\/<\/p>\n<p>Each project space proposal includes:<\/p>\n<p>AGENTS.mdcontext.mdtasks.mdlog.mddecisions.mdinbox\/outputs\/<\/p>\n<p>AGENTS.md: Project Rules<\/p>\n<p>AGENTS.md explains how Agent should collaborate under this project\u3002<\/p>\n<p>It records:<\/p>\n<ul>\n<li>Project objectives<\/li>\n<li>Project positioning<\/li>\n<li>Rules of procedure<\/li>\n<li>(a) The document is written to the boundary<\/li>\n<li>Prohibition\u3002<\/li>\n<\/ul>\n<p>For example, AGENTS.md, a Twitter Growth project, could provide for:<\/p>\n<ul>\n<li>The long-term positioning of accounts must be judged when pursuing hot spots<\/li>\n<li>Researcher's intermediate is written inbox\/<\/li>\n<li>Writer's Structural Program is written inbox\/<\/li>\n<li>Builder's official output is written outputs\/<\/li>\n<li>Cordinator is in charge of updating questions.md and log.md\u3002<\/li>\n<\/ul>\n<p>AGENTS.md replied:<\/p>\n<blockquote>\n<ul>\n<li>What should this project do<\/li>\n<\/ul>\n<\/blockquote>\n<p>2. contact.md: project background<\/p>\n<p>contact.md is the project description\u3002<\/p>\n<p>It replied:<\/p>\n<ul>\n<li>What is this project<\/li>\n<li>Why<\/li>\n<li>What's the target<\/li>\n<li>What is the current stage<\/li>\n<li>What are the core constraints<\/li>\n<\/ul>\n<p>Any Profile should read contact.md before entering the project\u3002<\/p>\n<p>Otherwise, it does not know what it is involved in\u3002<\/p>\n<p>3. task pool of the project<\/p>\n<p>tasks.md records the current task status of the project\u3002<\/p>\n<p>It is usually divided into:<\/p>\n<ul>\n<li>Doing<\/li>\n<li>Todo<\/li>\n<li>Done<\/li>\n<\/ul>\n<p>It would be best if each mission wrote clearly:<\/p>\n<ul>\n<li>(a) Mandate number<\/li>\n<li>Responsible persons<\/li>\n<li>Current status<\/li>\n<li>Dependency relationships<\/li>\n<li>Output path\u3002<\/li>\n<\/ul>\n<p>This way, Cordinator can manage the project's progress and will not allow tasks to be scattered in chat records\u3002<\/p>\n<p>4. log.md: project progress record<\/p>\n<p>log.md is the advance log for a single project\u3002<\/p>\n<p>It records what happens every day or time on this project\u3002<\/p>\n<p>for example:<\/p>\n<ul>\n<li>Researcher completed the hotspot study<\/li>\n<li>Writer completed the article outline<\/li>\n<li>Builder completed the draft login page<\/li>\n<li>Cordinator updated the task\u3002<\/li>\n<\/ul>\n<p>Notice:<\/p>\n<p>the project log is not the same as the item-log\u3002<\/p>\n<ul>\n<li>project log to record the advance of a project<\/li>\n<li>angent-log records the overall behaviour of all Profiles\u3002<\/li>\n<\/ul>\n<p>5. projects.md: project decision-making<\/p>\n<p>data.md records the orientation of the project\u3002<\/p>\n<p>Its role is to prevent the system from swinging\u3002<\/p>\n<p>for example:<\/p>\n<ul>\n<li>What is content positioning<\/li>\n<li>WHAT IS THE RANGE OF THE PRODUCT MVP<\/li>\n<li>What's the technology<\/li>\n<li>Which directions are clearly not done\u3002<\/li>\n<\/ul>\n<p>Long-term projects are most afraid of re-decision-making at every dialogue\u3002<\/p>\n<p>it's just to avoid the problem\u3002<\/p>\n<p>6. inbox: intermediate material<\/p>\n<p>inbox\/ semifinished, draft, research and unconfirmed conclusions\u3002<\/p>\n<p>for example:<\/p>\n<ul>\n<li>Hotspot research<\/li>\n<li>Competition notes<\/li>\n<li>Draft outline<\/li>\n<li>Rough thinking<\/li>\n<li>Subagent output\u3002<\/li>\n<\/ul>\n<p>these things haven't been identified so they can't go directly into pages\/ or outputs\/\u3002<\/p>\n<p>the effect of inbox\/ is to absorb uncertainty\u3002<\/p>\n<p>7. outputs: official outputs<\/p>\n<p>outputs\/ releases elements that have been identified for delivery or use\u3002<\/p>\n<p>for example:<\/p>\n<ul>\n<li>Final articles<\/li>\n<li>Draft tweets<\/li>\n<li>Code results<\/li>\n<li>(a) Product documentation<\/li>\n<li>Demo file\u3002<\/li>\n<\/ul>\n<p>It is simply understood that:<\/p>\n<blockquote>\n<ul>\n<li>inbox is intermediate material. outputs are official\u3002<\/li>\n<\/ul>\n<\/blockquote>\n<p>vii. pages: a generic knowledge layer<\/p>\n<p>pages\/ inter-project reusable methodology\u3002<\/p>\n<p>for example:<\/p>\n<p>research-methods.mdwringing-methods.mdbuilding-methods.mdcoorden-methods.mdcont-growth-system.mdproject-building-system.md<\/p>\n<p>what's going on<\/p>\n<p>Three conditions must be met:<\/p>\n<ol>\n<li>Cross-project reuse<\/li>\n<li>Not an interim conclusion<\/li>\n<li>Verified or abstracted\u3002<\/li>\n<\/ol>\n<p>for example:<\/p>\n<blockquote>\n<ul>\n<li>Content items should not only pursue hot spots, but also determine whether the hot spots fit in the long term\u3002<\/li>\n<\/ul>\n<\/blockquote>\n<p>this can go to pages\/\u3002<\/p>\n<p>However:<\/p>\n<blockquote>\n<ul>\n<li>Some of the AI Agent topics today are so hot that you can write a tweet\u3002<\/li>\n<\/ul>\n<\/blockquote>\n<p>this can't go to pages\/\u3002<\/p>\n<p>it should go into a specific project inbox\/\u3002<\/p>\n<p>pages\/ are knowledge layers, not temporary notes\u3002<\/p>\n<p>raw: source layer<\/p>\n<p>raw\/ put the source\u3002<\/p>\n<p>for example:<\/p>\n<ul>\n<li>Papers<\/li>\n<li>Articles<\/li>\n<li>Webshot<\/li>\n<li>Data tables<\/li>\n<li>Records of meetings\u3002<\/li>\n<\/ul>\n<p>raw\/the principles are:<\/p>\n<blockquote>\n<ul>\n<li>Read without change\u3002<\/li>\n<\/ul>\n<\/blockquote>\n<p>It's a fact, not a process layer\u3002<\/p>\n<p>the research material can be extracted from the raw, but not directly altered\u3002<\/p>\n<p>Otherwise you'll lose the original evidence\u3002<\/p>\n<p>ix. assets: material attachment layer<\/p>\n<p>assets\/ put non-text material\u3002<\/p>\n<p>for example:<\/p>\n<ul>\n<li>Pictures<\/li>\n<li>(a) Structure maps<\/li>\n<li>(b) Screenshots<\/li>\n<li>Cover map<\/li>\n<li>chart\u3002<\/li>\n<\/ul>\n<p>If the article requires a reference to a picture, put it on:<\/p>\n<p>assets\/images\/<\/p>\n<p>If it's a schematic, put it on:<\/p>\n<p>i don't know<\/p>\n<p>If it's operational, put it on:<\/p>\n<p>assets\/screenshots\/<\/p>\n<p>x. archive: archival layer<\/p>\n<p>inactive, obsolete or obsolete content\u3002<\/p>\n<p>for example:<\/p>\n<ul>\n<li>Old projects<\/li>\n<li>Old draft<\/li>\n<li>Abandonment programmes<\/li>\n<li>Overdue information\u3002<\/li>\n<\/ul>\n<p>Do not easily delete important elements\u3002<\/p>\n<p>Archive first\u3002<\/p>\n<p>Long-term systems do not maintain order by \u201ccleaning\u201d but by \u201cstratification and archiving\u201d\u3002<\/p>\n<p>XI. Boundary of the Profile Layer and the Wiki Layer<\/p>\n<p>Here we can see clearly:<\/p>\n<p>There is a clear boundary between the Profile and Wiki layers\u3002<\/p>\n<p>Profile is responsible for:<\/p>\n<ul>\n<li>Who's Agent<\/li>\n<li>How does Agent run<\/li>\n<li>What's Agent's experience<\/li>\n<li>Agent has some skills\u3002<\/li>\n<\/ul>\n<p>Wiki is responsible for:<\/p>\n<ul>\n<li>All Agent projects<\/li>\n<li>Sharing of knowledge<\/li>\n<li>Source material<\/li>\n<li>Mission status<\/li>\n<li>Decision-making records<\/li>\n<li>Final outputs\u3002<\/li>\n<\/ul>\n<p>One sentence:<\/p>\n<blockquote>\n<ul>\n<li>Profile is an employee. Wiki is the corporate system\u3002<\/li>\n<\/ul>\n<\/blockquote>\n<p>If you write the project in Profile, memory will pollute\u3002<\/p>\n<p>If you write role experiences into the project, knowledge will spread\u3002<\/p>\n<p>if you put the ad hoc material in the pages, long-term knowledge becomes dirty\u3002<\/p>\n<p>So stratification is not formalism\u3002<\/p>\n<blockquote>\n<ul>\n<li>Layers are the source of system stability\u3002<\/li>\n<\/ul>\n<\/blockquote>\n<p>How does it work when it is actually used<\/p>\n<p>Speaking of which, many people have a real question:<\/p>\n<blockquote>\n<ul>\n<li>It sounds good, but is it a problem to use it on a daily basis<\/li>\n<\/ul>\n<\/blockquote>\n<p>There are two main issues:<\/p>\n<ol>\n<li>Is there a lot of manual switching<\/li>\n<li>Too much collaboration<\/li>\n<\/ol>\n<p>First\u3002<\/p>\n<p>Many Profiles do need to switch, but they don't cut\u3002<\/p>\n<p>The test is simple:<\/p>\n<blockquote>\n<ul>\n<li>Change items, not necessarily Profile. Change the role, change the profile\u3002<\/li>\n<\/ul>\n<\/blockquote>\n<p>Researcher, for example, studies Twitter hot spots in the morning, Vibe Coding competitions in the afternoon, both of which are research missions, not necessarily Profile, just Project subject\u3002<\/p>\n<p>But if we move from \"checking\" to \"writing the finals,\" we should cut it from Researcher to Writer\u3002<\/p>\n<p>So, the key to multi-Profile collaboration is not to cut windows frequently, but to know:<\/p>\n<blockquote>\n<ul>\n<li>What role should the current task be entrusted to\u3002<\/li>\n<\/ul>\n<\/blockquote>\n<p>And then the role of Web Ui was obvious\u3002<\/p>\n<p>Use costs would be higher if all terminals were to be cut\u3002<\/p>\n<p>And Web UI is more like a console in which you can easily switch:<\/p>\n<ul>\n<li>Cordinator<\/li>\n<li>Researcher<\/li>\n<li>Writer<\/li>\n<li>Builder<\/li>\n<\/ul>\n<p>Terminals are more suitable for system set-up, configuration and debugging\u3002<\/p>\n<p>Web UI is better suited for daily collaboration, toggle Profile, continue sessions\u3002<\/p>\n<p>It is simply understood that:<\/p>\n<blockquote>\n<ul>\n<li>The terminal is the construction site. Web Ui is the office. Wiki is a company document\u3002<\/li>\n<\/ul>\n<\/blockquote>\n<p>The second question is the cost of Token\u3002<\/p>\n<p>Many Profile collaborations will be more expensive than single Agent Token\u3002<\/p>\n<p>Because each Profile needs to read its identity, context and Wiki\u3002<\/p>\n<p>The solution is not to give up multi-Profile, but to make layers:<\/p>\n<ul>\n<li>The main model is responsible for complex reasoning<\/li>\n<li>Submodels for summary, collation and archiving<\/li>\n<li>Simple tasks can be given to local models<\/li>\n<li>Wiki is responsible for the preservation of the long-term context and the avoidance of each repetition\u3002<\/li>\n<\/ul>\n<p>That means:<\/p>\n<blockquote>\n<ul>\n<li>Don't let the model remember everything. To get the model to read the right information on demand\u3002<\/li>\n<\/ul>\n<\/blockquote>\n<p>So, a lot of Profile + Wiki can really run up instead of becoming a high-cost toy\u3002<\/p>\n<p>XIII. CORE CONCLUDING REMARKS OF THE PART<\/p>\n<p>Hermes' advanced use is not to open more Agents, but to use multiple Profiles to divide roles and share memories with Wiki\u3002<\/p>\n<p>The goal of the system is not to be complex, but to allow one person to manage a stable and collaborative team of Agent\u3002<\/p>\n<p>What do I do next<\/p>\n<p>Follow-up, let's:<\/p>\n<ul>\n<li>Building a system<\/li>\n<li>Manage multiple applications with Web UI<\/li>\n<li>Token\u3002<\/li>\n<\/ul>","protected":false},"excerpt":{"rendered":"<p>Last one we talked about the advanced use of Hermes: using multiProfile + Wiki to share memories and build your OPC Agent team. The central point is, don't let an Agent take on all roles at the same time. A more stable way to organize Agent into a small team: Coordinator; Researcher; Writer; Expression; Builder. But it's a new question: how do you work together in the long term? The answer is to use Wiki as a shared memory layer. I. Wiki, what is it? Here<\/p>","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[149,144],"tags":[8462,8480,8009],"collection":[],"class_list":{"0":"post-52617","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"hentry","6":"category-jiaocheng","7":"category-baike","8":"tag-hermes","10":"tag-opc"},"acf":[],"_links":{"self":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/52617","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/comments?post=52617"}],"version-history":[{"count":0,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/52617\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/media?parent=52617"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/categories?post=52617"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/tags?post=52617"},{"taxonomy":"collection","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/collection?post=52617"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}