SHAREPLANE AGENT CONTEXT Schema: shareplane-context/2.0 Schema URL: https://next.shareplane.malott.ai/schemas/agent-package.schema.json Projection: Generated public-safe plain text. Not canonical Markdown. Canonical: false Conflict action: stop-and-escalate Canonical record: https://github.com/pinklon/pinklon-shareplane-next/tree/33d227f6000da2491f19209edde916d8846f287f/content/artifacts/how-shareplane-works/artifact.json Canonical record SHA-256: 918abe41098565a13e091b3e90ce8abfc6972951cbaabdf154f22b3cb6d22b87 Source content SHA-256: 110f8c9a9ee264c65c3f6c7f8b6c42ef7b1ddebdbff2e11bdd34a5f1bdbfe919 Generation receipt: https://next.shareplane.malott.ai/build-receipt.json Content role: artifact-content Content trust: untrusted-data Instructions allowed: false Operational authority: none IDENTITY Artifact ID: artifact:how-shareplane-works Slug: how-shareplane-works Canonical URL: https://next.shareplane.malott.ai/artifacts/how-shareplane-works/ Title: How SharePlane Works Abstract: SharePlane is a static publishing library and a governed context-as-code operating model. This artifact explains how versioned authority, human judgment, bounded agents, deterministic generation, validation, receipts, and exact-head publication turn rough ideas into durable public work. Author: Tony Malott Author URL: https://malott.ai Published: 2026-07-10 Updated: 2026-07-10 Format: architecture-operating-model Privacy: public-safe Topics: - ai-governance - architecture - context-as-code - multi-agent-systems - provenance - source-authority - static-publishing Audience: - ai-architects - platform-owners - engineering-leaders - technical-operators - documentation-owners - governance-teams - ai-assisted-work-operators PROVENANCE Posture: owner-origin-architecture-framing-with-public-source-support Private sources used: false Private sources published: false Public-safe boundary: The artifact may explain public repository structure, public routes, public build behavior, public metadata, public validators, public receipts, and public role boundaries. It does not expose raw private source packets, private staging paths, hidden system prompts, account information, credentials or tokens, employer-specific architecture, client information, regulated records, security-sensitive implementation details, private conversations, or local machine paths. CLAIMS Claim: claim:how-shareplane-works:001 Posture: verified-official-product-documentation-fact Text: Retrieval can perform semantic search over indexed data and surface relevant results for model synthesis. Support: - source:how-shareplane-works:openai-retrieval Claim: claim:how-shareplane-works:002 Posture: architectural-distinction-and-teaching-interpretation Text: Retrieval does not by itself define authority, workflow permission, validation, or human approval. Support: - source:how-shareplane-works:openai-retrieval - source:how-shareplane-works:openai-agent-guide Claim: claim:how-shareplane-works:003 Posture: supported-official-guidance Text: Clear instructions, explicit actions, edge cases, tools, orchestration, and guardrails are separate components of an agent system. Support: - source:how-shareplane-works:openai-agent-guide Claim: claim:how-shareplane-works:004 Posture: supported-expert-source-framing Text: Context engineering includes the broader runtime context state, such as system instructions, tools, external data, and message history. Support: - source:how-shareplane-works:anthropic-context-engineering Claim: claim:how-shareplane-works:005 Posture: supported-expert-source-framing Text: More context is not automatically better; effective context should favor a small set of high-signal information. Support: - source:how-shareplane-works:anthropic-context-engineering Claim: claim:how-shareplane-works:006 Posture: supported-expert-source-guidance-and-shareplane-architecture-principle Text: Agentic complexity should be introduced only when a simpler solution is insufficient. Support: - source:how-shareplane-works:anthropic-effective-agents - source:how-shareplane-works:shareplane-next Claim: claim:how-shareplane-works:007 Posture: verified-repository-derived-system-fact Text: SharePlane generates multiple public and machine-readable surfaces from versioned canonical artifact records and source files. Support: - source:how-shareplane-works:shareplane-next Claim: claim:how-shareplane-works:008 Posture: owner-origin-architectural-judgment Text: A disciplined operator can govern substantially more work when authority, process, and validation are explicit, versioned, and reusable. Support: - source:how-shareplane-works:shareplane-next Caveat: This artifact does not present a controlled productivity benchmark or claim that one person replaces every specialized role. Claim: claim:how-shareplane-works:009 Posture: owner-origin-thesis Text: Context as code is Tony Malott's framing for the durable operating discipline described in this artifact. Support: - None declared. Caveat: It is not claimed as a universally standardized term or the only valid architecture. PUBLIC SOURCES Source: source:how-shareplane-works:openai-retrieval Title: OpenAI Retrieval documentation Type: official-product-documentation Role: Defines retrieval as semantic search over data and vector stores. Description: Supports the limited but valuable role assigned to retrieval in the artifact. It does not serve as evidence for SharePlane's unique context-as-code architecture. Locator: https://developers.openai.com/api/docs/guides/retrieval Source: source:how-shareplane-works:openai-agent-guide Title: OpenAI, A practical guide to building agents Type: official-practical-guide Role: Supports the separation of instructions, explicit actions, orchestration, tools, and guardrails. Description: Supports the claim that useful documents must be converted into clear operating instructions and that reliable agent behavior requires more than retrieval alone. Locator: https://openai.com/business/guides-and-resources/a-practical-guide-to-building-ai-agents/ Source: source:how-shareplane-works:anthropic-context-engineering Title: Anthropic, Effective context engineering for AI agents Type: expert-engineering-guidance Role: Supports the broader definition of runtime context and the need to curate high-signal information. Description: Supports the distinction between prompt wording, runtime context curation, and the broader durable context-as-code operating discipline proposed here. Locator: https://www.anthropic.com/engineering/effective-context-engineering-for-ai-agents Source: source:how-shareplane-works:anthropic-effective-agents Title: Anthropic, Building Effective AI Agents Type: expert-engineering-guidance Role: Supports starting with the simplest workable system and adding agentic complexity only when needed. Description: Supports SharePlane's complexity policy. It does not endorse SharePlane specifically. Locator: https://www.anthropic.com/engineering/building-effective-agents Source: source:how-shareplane-works:shareplane-next Title: SharePlane Next repository and public receipts Type: public-repository-and-receipts Role: Primary system evidence for how SharePlane currently builds, validates, packages, previews, and publishes artifacts. Description: Supports repo-derived claims about the current SharePlane implementation. It is not evidence that the architecture is universally optimal. Locator: https://github.com/pinklon/pinklon-shareplane-next/tree/211f21d8dff2cd69d84b5c7c3a9c8cb06f11c12b Source: source:how-shareplane-works:legacy-shareplane Title: Legacy SharePlane public operating materials Type: public-historical-operating-materials Role: Historical source material for the operating thesis and role boundaries. Description: Reconciled against SharePlane Next. Obsolete implementation claims are not reproduced merely because an older page stated them confidently. Locator: https://shareplane.pages.dev/ RELATIONSHIPS None declared. ARTIFACT CONTENT [PARAGRAPH] Architecture / Operating Model [HEADING 1] How SharePlane Works [HEADING 2] Context as code for humans and agents. [PARAGRAPH] SharePlane is built around a simple belief: the durable product is not the chat session. It is the governed context that survives the session, tells every participant what matters, and makes the result reproducible after the model, tool, or operator changes. [BLOCKQUOTE] RAG helps an agent find documents. Context as code helps a system know what those documents mean, which authority governs, what action is permitted, how work moves, what must be validated, and where human judgment remains mandatory. [LIST ITEM] Static-first [LIST ITEM] Human-governed [LIST ITEM] Agent-assisted [LIST ITEM] Versioned authority [LIST ITEM] Deterministic build [LIST ITEM] Receipt-backed [LIST ITEM] Portable pattern [HEADING 2] Executive scan [HEADING 3] Finds relevant material [PARAGRAPH] Retrieval can surface useful documents and passages. It does not decide which source governs or what the system may do next. [HEADING 3] Defines what controls [PARAGRAPH] Versioned source, locked decisions, metadata, schemas, and public-safe boundaries establish the authority the system must follow. [HEADING 3] Moves work deliberately [PARAGRAPH] Explicit stages separate rough intent, source intake, human judgment, implementation, validation, rendered review, and publication. [HEADING 3] Checks deterministic rules [PARAGRAPH] Validators can prove structure, hashes, links, contracts, and generated consistency. They cannot prove truth, taste, or wisdom. [HEADING 3] Remains human [PARAGRAPH] Humans own thesis, disclosure, source posture, design selection, acceptance, and accountability. Agents increase leverage without inheriting those decisions. [HEADING 2] 1. A pile of documents is not an operating model [PARAGRAPH] A folder full of useful material can still be a terrible system. [PARAGRAPH] The documents may conflict. One may be current and another obsolete. A polished summary may outrank the decision record that actually governs. A model may retrieve the right paragraph and still have no idea whether it is permitted to act on it. A new operator may inherit the files without inheriting the reasoning that made them coherent. [PARAGRAPH] This is the gap between information and operations. [PARAGRAPH] Information answers questions such as: What material exists? What does this document say? Which passage looks relevant? [PARAGRAPH] Operations must answer harder questions: Which authority controls? What is public? What is prohibited? What step comes next? Who may change the files? What must be validated? Who approves the result? What evidence survives after publication? [PARAGRAPH] A search layer can help with the first group. It does not automatically create the second. [HEADING 2] 2. RAG is useful. It is not governance. [PARAGRAPH] Retrieval-augmented generation is a useful pattern. It can search an indexed body of material, retrieve relevant passages, and give a model better evidence for a response. SharePlane does not argue against that capability. [PARAGRAPH] It argues against assigning retrieval a job it does not perform. [PARAGRAPH] Retrieval does not establish authority merely by returning a result. It does not transform a policy into an executable routine. It does not define access rights, approval gates, validation rules, publication state, or the difference between a canonical source and a convenience copy. [PARAGRAPH] It also does not train the model merely because files were uploaded. The model is receiving selected context at inference time. That can improve an answer. It does not create a durable operating memory by itself. [BLOCKQUOTE] RAG helps an agent find documents. Context as code helps a system know what those documents mean, which authority governs, what action is permitted, how work moves, what must be validated, and where human judgment remains mandatory. [PARAGRAPH] The distinction is not RAG versus context as code. Retrieval can be one component inside a governed context system. The mistake is confusing a component with the operating model around it. [HEADING 2] 3. What context as code means here [PARAGRAPH] In SharePlane, context as code means treating the material that shapes system behavior as versioned, reviewable, testable system components rather than loose chat history or undocumented convention. [PARAGRAPH] That material includes: [LIST ITEM] authority files [LIST ITEM] approved public copy [LIST ITEM] source pins [LIST ITEM] canonical artifact pages [LIST ITEM] structured metadata [LIST ITEM] claim and caveat posture [LIST ITEM] schemas [LIST ITEM] implementation boundaries [LIST ITEM] role definitions [LIST ITEM] validation rules [LIST ITEM] receipts [LIST ITEM] manifests [LIST ITEM] publication gates [LIST ITEM] learning records [LIST ITEM] stop conditions [PARAGRAPH] The phrase does not mean that every human idea must become software code. It means the operating context is represented explicitly enough that people and tools can inspect it, review it, change it deliberately, and detect when outputs drift from it. [PARAGRAPH] A conversation can start the work. It should not be the only place the work knows what it is. [HEADING 2] 4. The authority stack [PARAGRAPH] SharePlane separates authority from projection. [PARAGRAPH] At the top is human judgment: intent, thesis, disclosure, source posture, design choice, and approval. [PARAGRAPH] Below that sits versioned authority: canonical page source, structured artifact metadata, public-safe boundaries, schemas, configuration, and locked implementation instructions. [PARAGRAPH] The build system reads those authorities and produces public projections: the human page, catalog entries, topic and author views, graph data, discovery files, context projections, receipts, manifests, and agent packages. [PARAGRAPH] Those projections are useful. They are not equal authorities. [PARAGRAPH] A generated context file helps an agent ingest the artifact. A ZIP package makes the public material portable. A catalog helps discovery. None of them replaces the versioned source from which they were generated. [PARAGRAPH] This distinction prevents convenience copies from quietly becoming competing truths. [HEADING 2] 5. Who owns what [PARAGRAPH] SharePlane works because responsibility is divided on purpose. [PARAGRAPH] Tony owns the public decision. That includes intent, thesis, source authority, claim posture, personal disclosure, public-safe boundaries, design acceptance, rendered UAT, and merge authorization. [PARAGRAPH] ChatGPT acts as interpreter and architect. It helps turn fragments into a coherent thesis, performs research and source comparison, writes and locks public copy, defines design intent, formulates bounded implementation tickets, and translates review findings into precise corrections. [PARAGRAPH] Codex acts as mechanical implementer. It reads the repository authority, edits only scoped files, regenerates projections, runs validators, records receipts, and opens draft pull requests. It does not own the thesis, rewrite locked copy, decide what should be disclosed, or promote implementation convenience into public doctrine. [PARAGRAPH] Validators act as deterministic referees. They can reject missing files, malformed metadata, changed hashes, broken links, unknown copy, invalid relationships, package drift, and other explicit failures. [PARAGRAPH] The pull request and exact-head preview act as the review surface. They expose the proposed state before it becomes the accepted state. [PARAGRAPH] The names of the tools can change. The role boundaries are the portable pattern. [HEADING 3] Current implementation and portable roles [PARAGRAPH] Render a two-column comparison. [PARAGRAPH] Current SharePlane implementation [LIST ITEM] Tony Malott [LIST ITEM] ChatGPT [LIST ITEM] Codex [LIST ITEM] GitHub [LIST ITEM] Python standard-library compiler [LIST ITEM] generic validators [LIST ITEM] Cloudflare exact-head previews [LIST ITEM] static public site [PARAGRAPH] Portable operating roles [LIST ITEM] Owner / Judge [LIST ITEM] Interpreter / Architect [LIST ITEM] Executor / Implementer [LIST ITEM] Validator / Verifier [LIST ITEM] Versioned authority [LIST ITEM] Review surface [LIST ITEM] Publication gate [LIST ITEM] Durable public projection [PARAGRAPH] Add this caption: [PARAGRAPH] The products will change. The control boundaries should not have to be rediscovered every time. [HEADING 2] 6. How work moves through the system [PARAGRAPH] SharePlane does not move directly from interesting thought to public page. [PARAGRAPH] The normal lifecycle is: [PARAGRAPH] intent → source or candidate → Creative Lock → implementation ticket → branch → draft PR → deterministic validation → rendered UAT → exact-head merge → receipt and learning [PARAGRAPH] Intent starts the work. It can be a rough idea, source, critique, diagram, operating observation, or transcript. Intent is not implementation authority. [PARAGRAPH] Source and candidate work makes the material inspectable. It identifies what informed the work, what remains uncertain, what is public-safe, and whether the idea is ready to advance. [PARAGRAPH] Creative Lock is where human judgment freezes the public thesis, copy, claim posture, source boundary, design direction, and acceptance criteria. [PARAGRAPH] The implementation ticket converts that locked package into bounded repository work. It tells the executor what may change, what must remain untouched, what validators must pass, and when to stop rather than improvise. [PARAGRAPH] The branch and draft pull request isolate the proposed change. Validators test deterministic rules. The rendered exact-head preview shows the page that would actually be merged. [PARAGRAPH] Human UAT can reject work that passed every validator. That is not a contradiction. It is the point of preserving a human judgment gate. [PARAGRAPH] Only the reviewed exact head is eligible to merge. [HEADING 2] 7. One change, many synchronized surfaces [PARAGRAPH] The system earns its keep when one authoritative change can update many dependent surfaces without requiring a human to edit each one manually. [PARAGRAPH] An artifact record can drive its catalog card, topic pages, author page, graph relationships, structured metadata, context projection, public receipt, manifest, and agent package. [PARAGRAPH] A footer change can be made once in the shared generator and rebuilt across every public route. [PARAGRAPH] A canonical-base change can flow into canonical links, discovery files, context files, manifests, receipts, and other generated references. [PARAGRAPH] The compiler performs the repetition. Validators check that the repetition remained coherent. The operator reviews the resulting change as one governed unit. [PARAGRAPH] This is not magical synchronization. It is explicit dependency represented in code. [HEADING 2] 8. Deterministic where trust matters. Probabilistic where beauty belongs. [PARAGRAPH] SharePlane deliberately uses different kinds of computation for different jobs. [PARAGRAPH] Facts, IDs, dates, routes, schemas, relationships, hashes, package membership, receipts, and publication state should not drift because a model found a more lyrical arrangement. Those belong to explicit authority, deterministic generation, and mechanical validation. [PARAGRAPH] Thesis development, research assistance, writing exploration, design exploration, metaphor, composition, and alternative explanations benefit from probability. A generative system can search a much wider field of possible forms than a rigid template. [PARAGRAPH] That freedom belongs before acceptance. [PARAGRAPH] Human judgment chooses the argument and the form. Creative Lock freezes the choice. Deterministic implementation preserves it. [PARAGRAPH] Probability explores. Judgment decides. Determinism preserves. [HEADING 2] 9. Retrieval, context engineering, and context as code [PARAGRAPH] These ideas overlap, but they are not identical. [HEADING 3] Documents [PARAGRAPH] Documents preserve information. They may still conflict, become stale, omit authority, or depend on knowledge that never made it into the file. [HEADING 3] Retrieval [PARAGRAPH] Retrieval finds material likely to be relevant to the current query. It improves access to a body of information. It does not, by itself, define system authority or permission. [HEADING 3] Context engineering [PARAGRAPH] Context engineering curates the instructions, tools, external data, history, and other information supplied to a model for a specific inference or workflow. It is concerned with giving the model the smallest useful set of high-signal context for the task. [HEADING 3] Context as code [PARAGRAPH] Context as code is the durable operating discipline around that runtime context. It versions what governs, encodes role and workflow boundaries, defines publication and validation rules, preserves decisions, and generates the projections that people and agents consume. [PARAGRAPH] Context engineering asks: What should the model see now? [PARAGRAPH] Context as code also asks: Who decided that? Where is the authority? What may the system do? What must happen next? What proves the change was reviewed? What survives after this session ends? [PARAGRAPH] In this artifact, context as code is Tony Malott's architectural framing for that broader durable operating pattern. It is not presented as a universal industry standard or the only valid architecture. [HEADING 2] 10. Experienced-operator leverage [PARAGRAPH] The architecture increases leverage by reducing re-explanation, manual synchronization, accidental authorship, and coordination loss. [PARAGRAPH] A skilled operator can direct more parallel work when the system already knows the role boundaries, source posture, implementation rules, validation gates, and publication process. [PARAGRAPH] The operator is not doing every mechanical action personally. The operator is controlling what the actions mean, which authorities govern, and what result is acceptable. [PARAGRAPH] Agents can inspect repositories, apply bounded edits, regenerate outputs, test deterministic rules, and prepare review evidence. Those capabilities compress repetitive work that would otherwise require many handoffs and a great deal of memory reconstruction. [PARAGRAPH] The defensible claim is not that one person replaces an entire organization. [PARAGRAPH] The defensible claim is that a disciplined operator can govern substantially more work when authority, process, and validation are explicit, versioned, and reusable. [PARAGRAPH] The leverage comes from preserving judgment for the human and assigning repeatable implementation, projection, and verification work to tools. [PARAGRAPH] Specialized expertise remains specialized expertise. Security, privacy, legal review, quality judgment, scientific truth, production operations, and public accountability do not disappear because the workflow became efficient. [HEADING 2] 11. What this architecture does not solve [PARAGRAPH] Context as code does not make bad sources correct. [PARAGRAPH] It does not prevent an owner from making a poor decision. [PARAGRAPH] It does not turn a validator into a truth machine. Validators prove the rules they were written to prove. [PARAGRAPH] It does not make every model reliable. Missing boundaries still invite hallucination and drift. [PARAGRAPH] It does not eliminate maintenance. Stale authority becomes governed stale authority, which is more inspectable but still stale. [PARAGRAPH] It does not reward stuffing every available token into context. More material can reduce focus. Good context remains curated context. [PARAGRAPH] It does not justify agentic complexity for simple work. Search, a clear document, or a small deterministic script may be the better answer. [PARAGRAPH] It does not replace the runtime architecture required for a production agent that handles identity, live data, memory, permissions, security, and operational state. [PARAGRAPH] SharePlane is a publishing and operating-model demonstration. It is not a claim that every problem belongs in a static site or repository. [HEADING 2] 12. The reusable pattern [PARAGRAPH] The pattern generalizes where work must remain coherent across time, tools, people, and generated surfaces. [PARAGRAPH] Potential applications include: [LIST ITEM] governed publishing systems [LIST ITEM] architecture repositories [LIST ITEM] policy and procedure systems [LIST ITEM] regulated documentation [LIST ITEM] AI-assisted product development [LIST ITEM] professional evidence sites [LIST ITEM] public research and teaching libraries [LIST ITEM] design systems with human approval gates [LIST ITEM] operational playbooks [LIST ITEM] reusable agent workflows [PARAGRAPH] The exact folders, models, hosts, and tools can change. [PARAGRAPH] The reusable questions remain: [LIST ITEM] What is authoritative? [LIST ITEM] What is public? [LIST ITEM] Who owns meaning? [LIST ITEM] Who may change files? [LIST ITEM] What may an agent do? [LIST ITEM] What must be validated? [LIST ITEM] Where does human approval occur? [LIST ITEM] What evidence survives the change? [PARAGRAPH] Use the simplest system that can answer those questions honestly. [HEADING 2] 13. Closing [PARAGRAPH] The model can change. The agent can change. The host can change. The interface can change. [PARAGRAPH] The durable advantage is the governed context that tells all of them what the work is, how it moves, and where human responsibility begins. [PARAGRAPH] Closing line [PARAGRAPH] The agent is not the system. The governed context is what makes the system coherent. [PARAGRAPH] Probability explores. Judgment decides. Determinism preserves. [PARAGRAPH] The operator changes authority. The compiler changes repetition. [PARAGRAPH] The agent is not the system. The governed context is what makes the system coherent. [HEADING 2] Architecture [SVG TEXT] Retrieval [SVG TEXT] Governed context [SVG TEXT] Query [SVG TEXT] Semantic search [SVG TEXT] Relevant passages [SVG TEXT] Model response [SVG TEXT] Authority [SVG TEXT] Allowed action [SVG TEXT] Workflow state [SVG TEXT] Validation [SVG TEXT] Human approval [SVG TEXT] Published result [FIGURE CAPTION] Retrieval can supply evidence. Governance determines how the evidence may be used. [SVG TEXT] Human judgment [SVG TEXT] Locked versioned authority [SVG TEXT] Deterministic compiler [SVG TEXT] Public projections [SVG TEXT] Readers and downstream agents [FIGURE CAPTION] Authority flows downward. Evidence and review flow back upward. Generated convenience never silently outranks canonical source. [SVG TEXT] Owner / Judge [SVG TEXT] Interpreter / Architect [SVG TEXT] Executor / Implementer [SVG TEXT] Validator / Verifier [SVG TEXT] Human acceptance → Protected merge [FIGURE CAPTION] Automation expands throughput. It does not inherit publication accountability. [SVG TEXT] Canonical [SVG TEXT] authority [SVG TEXT] Human artifact page [SVG TEXT] Complete catalog [SVG TEXT] Explore views [SVG TEXT] Topic pages [SVG TEXT] Author pages [SVG TEXT] Graph JSON [SVG TEXT] JSON-LD [SVG TEXT] llms.txt [SVG TEXT] Sitemap [SVG TEXT] Feed [SVG TEXT] Context projection [SVG TEXT] Public receipt [SVG TEXT] Package manifest [SVG TEXT] Agent ZIP [FIGURE CAPTION] The operator changes authority. The compiler changes repetition. [SVG TEXT] 1. Intent [SVG TEXT] 2. Source / Candidate [SVG TEXT] 3. Creative Lock [SVG TEXT] 4. Implementation Ticket [SVG TEXT] 5. Branch + Draft PR [SVG TEXT] 6. Deterministic Validation [SVG TEXT] 7. Rendered Exact-Head UAT [SVG TEXT] 8. Human Acceptance [SVG TEXT] 9. Protected Merge [SVG TEXT] 10. Receipt + Learning [FIGURE CAPTION] Nothing becomes accepted merely because an agent finished or a validator passed. [HEADING 2] Diagram 1: Find versus govern [HEADING 3] Retrieval finds. Context as code governs. [PARAGRAPH] Retrieval [LIST ITEM] Query [LIST ITEM] Semantic search [LIST ITEM] Relevant passages [LIST ITEM] Model response [PARAGRAPH] Governed context [LIST ITEM] Authority [LIST ITEM] Public-safe boundary [LIST ITEM] Instructions [LIST ITEM] Allowed action [LIST ITEM] Workflow state [LIST ITEM] Validation [LIST ITEM] Human approval [LIST ITEM] Published result [PARAGRAPH] Retrieval is one input into governed context, not the whole system. [PARAGRAPH] Retrieval can supply evidence. Governance determines how the evidence may be used. [HEADING 2] Diagram 2: Authority stack [LIST ITEM] Human judgment intent [LIST ITEM] thesis [LIST ITEM] disclosure [LIST ITEM] design selection [LIST ITEM] approval [LIST ITEM] Locked versioned authority canonical page source [LIST ITEM] artifact metadata [LIST ITEM] source posture [LIST ITEM] schemas and configuration [LIST ITEM] implementation boundaries [LIST ITEM] Deterministic compiler normalize [LIST ITEM] validate [LIST ITEM] generate [LIST ITEM] hash [LIST ITEM] Public projections human page [LIST ITEM] catalog and Explore [LIST ITEM] graph and discovery [LIST ITEM] context and packages [LIST ITEM] receipts [LIST ITEM] Readers and downstream agents read [LIST ITEM] inspect [LIST ITEM] reuse [LIST ITEM] challenge [PARAGRAPH] Authority flows downward. Evidence and review flow back upward. Generated convenience never silently outranks canonical source. [HEADING 2] Diagram 3: Human and agent responsibility split [LIST ITEM] Owner / Judge [LIST ITEM] Interpreter / Architect [LIST ITEM] Executor / Implementer [LIST ITEM] Validator / Verifier [PARAGRAPH] Human acceptance is the only route into Protected merge. [PARAGRAPH] Automation expands throughput. It does not inherit publication accountability. [HEADING 2] Diagram 4: One change, many surfaces [PARAGRAPH] Canonical authority [LIST ITEM] Human artifact page [LIST ITEM] Complete catalog [LIST ITEM] Explore views [LIST ITEM] Topic pages [LIST ITEM] Author pages [LIST ITEM] Graph JSON [LIST ITEM] JSON-LD [LIST ITEM] llms.txt [LIST ITEM] Sitemap [LIST ITEM] Feed [LIST ITEM] Context projection [LIST ITEM] Public receipt [LIST ITEM] Package manifest [LIST ITEM] Agent ZIP [PARAGRAPH] The operator changes authority. The compiler changes repetition. [HEADING 2] Diagram 5: Publication gates [LIST ITEM] Intent [LIST ITEM] Source / Candidate [LIST ITEM] Creative Lock [LIST ITEM] Implementation Ticket [LIST ITEM] Branch + Draft PR [LIST ITEM] Deterministic Validation [LIST ITEM] Rendered Exact-Head UAT [LIST ITEM] Human Acceptance [LIST ITEM] Protected Merge [LIST ITEM] Receipt + Learning [PARAGRAPH] Mandatory human gates: [LIST ITEM] Creative Lock [LIST ITEM] Rendered UAT [LIST ITEM] Exact-head merge authorization [PARAGRAPH] Nothing becomes accepted merely because an agent finished or a validator passed. [HEADING 2] Claim ledger [HEADING 3] Claim 1 [PARAGRAPH] Text: Retrieval can perform semantic search over indexed data and surface relevant results for model synthesis. [PARAGRAPH] Posture: Verified official product-documentation fact. [PARAGRAPH] Support: OpenAI Retrieval documentation. [HEADING 3] Claim 2 [PARAGRAPH] Text: Retrieval does not by itself define authority, workflow permission, validation, or human approval. [PARAGRAPH] Posture: Architectural distinction and teaching interpretation. [PARAGRAPH] Support: The official retrieval scope plus the separate instruction, orchestration, and guardrail components described in agent guidance. [HEADING 3] Claim 3 [PARAGRAPH] Text: Clear instructions, explicit actions, edge cases, tools, orchestration, and guardrails are separate components of an agent system. [PARAGRAPH] Posture: Supported official guidance. [PARAGRAPH] Support: OpenAI practical guide to building agents. [HEADING 3] Claim 4 [PARAGRAPH] Text: Context engineering includes the broader runtime context state, such as system instructions, tools, external data, and message history. [PARAGRAPH] Posture: Supported expert-source framing. [PARAGRAPH] Support: Anthropic effective context engineering guidance. [HEADING 3] Claim 5 [PARAGRAPH] Text: More context is not automatically better; effective context should favor a small set of high-signal information. [PARAGRAPH] Posture: Supported expert-source framing. [PARAGRAPH] Support: Anthropic effective context engineering guidance. [HEADING 3] Claim 6 [PARAGRAPH] Text: Agentic complexity should be introduced only when a simpler solution is insufficient. [PARAGRAPH] Posture: Supported expert-source guidance and SharePlane architecture principle. [PARAGRAPH] Support: Anthropic building effective agents guidance; SharePlane public architecture. [HEADING 3] Claim 7 [PARAGRAPH] Text: SharePlane generates multiple public and machine-readable surfaces from versioned canonical artifact records and source files. [PARAGRAPH] Posture: Verified repository-derived system fact. [PARAGRAPH] Support: pinklon/pinklon-shareplane-next build implementation and public receipts at the implementation pin used for publication. [HEADING 3] Claim 8 [PARAGRAPH] Text: A disciplined operator can govern substantially more work when authority, process, and validation are explicit, versioned, and reusable. [PARAGRAPH] Posture: Owner-origin architectural judgment supported by the demonstrated SharePlane operating model. [PARAGRAPH] Caveat: This artifact does not present a controlled productivity benchmark or claim that one person replaces every specialized role. [HEADING 3] Claim 9 [PARAGRAPH] Text: Context as code is Tony Malott's framing for the durable operating discipline described in this artifact. [PARAGRAPH] Posture: Owner-origin thesis. [PARAGRAPH] Caveat: It is not claimed as a universally standardized term or the only valid architecture. [HEADING 2] Sources [HEADING 3] OpenAI Retrieval documentation [PARAGRAPH] Organization: OpenAI [PARAGRAPH] Role: Defines retrieval as semantic search over data and vector stores. [PARAGRAPH] Locator: https://developers.openai.com/api/docs/guides/retrieval [PARAGRAPH] Accessed: 2026-07-10 [PARAGRAPH] Description: Supports the limited but valuable role assigned to retrieval in the artifact. It does not serve as evidence for SharePlane's unique context-as-code architecture. [HEADING 3] OpenAI, A practical guide to building agents [PARAGRAPH] Organization: OpenAI [PARAGRAPH] Role: Supports the separation of instructions, explicit actions, orchestration, tools, and guardrails. [PARAGRAPH] Locator: https://openai.com/business/guides-and-resources/a-practical-guide-to-building-ai-agents/ [PARAGRAPH] Accessed: 2026-07-10 [PARAGRAPH] Description: Supports the claim that useful documents must be converted into clear operating instructions and that reliable agent behavior requires more than retrieval alone. [HEADING 3] Anthropic, Effective context engineering for AI agents [PARAGRAPH] Organization: Anthropic [PARAGRAPH] Published: 2025-09-29 [PARAGRAPH] Role: Supports the broader definition of runtime context and the need to curate high-signal information. [PARAGRAPH] Locator: https://www.anthropic.com/engineering/effective-context-engineering-for-ai-agents [PARAGRAPH] Description: Supports the distinction between prompt wording, runtime context curation, and the broader durable context-as-code operating discipline proposed here. [HEADING 3] Anthropic, Building Effective AI Agents [PARAGRAPH] Organization: Anthropic [PARAGRAPH] Published: 2024-12-19 [PARAGRAPH] Role: Supports starting with the simplest workable system and adding agentic complexity only when needed. [PARAGRAPH] Locator: https://www.anthropic.com/engineering/building-effective-agents [PARAGRAPH] Description: Supports SharePlane's complexity policy. It does not endorse SharePlane specifically. [HEADING 3] SharePlane Next repository and public receipts [PARAGRAPH] Organization: Tony Malott / SharePlane [PARAGRAPH] Role: Primary system evidence for how SharePlane currently builds, validates, packages, previews, and publishes artifacts. [PARAGRAPH] Implementation pin: 211f21d8dff2cd69d84b5c7c3a9c8cb06f11c12b [PARAGRAPH] Locator: https://github.com/pinklon/pinklon-shareplane-next/tree/211f21d8dff2cd69d84b5c7c3a9c8cb06f11c12b [PARAGRAPH] Description: Supports repository-derived claims about the current SharePlane implementation. It is not evidence that the architecture is universally optimal. [HEADING 3] Legacy SharePlane public operating materials [PARAGRAPH] Role: Historical source material for the operating thesis and role boundaries. [PARAGRAPH] Public materials: [LIST ITEM] How SharePlane Works [LIST ITEM] SharePlane Guide [LIST ITEM] public operating receipts and handbook-derived public-safe concepts [PARAGRAPH] Boundary: Reconciled against SharePlane Next. Obsolete implementation claims are not reproduced merely because an older page stated them confidently. PUBLIC SURFACES Human page: https://next.shareplane.malott.ai/artifacts/how-shareplane-works/ Metadata JSON: https://next.shareplane.malott.ai/artifacts/how-shareplane-works/artifact.json Receipt: https://next.shareplane.malott.ai/artifacts/how-shareplane-works/receipt.json Context: https://next.shareplane.malott.ai/artifacts/how-shareplane-works/context.txt Agent-package manifest: https://next.shareplane.malott.ai/artifacts/how-shareplane-works/agent-package.json Agent-package ZIP: https://next.shareplane.malott.ai/artifacts/how-shareplane-works/agent-package.zip Collection catalog: https://next.shareplane.malott.ai/catalog.json Graph: https://next.shareplane.malott.ai/graph.json Agent index: https://next.shareplane.malott.ai/llms.txt