ultimate-pi 0.1.7 → 0.2.2

package/vault/wiki/decisions/adr-027.md

@@ -1,94 +0,0 @@
----
-type: decision
-title: "ADR-027: pi-vcc Overrides Built-in Auto-Compaction"
-status: accepted
-priority: 1
-date: "2026-05-07"
-tags: [adr, compaction, pi-vcc, vcc, context]
-sources:
-  - "[[pi-vcc-github-repo]]"
-  - "[[vcc-conversation-compaction-for-pi]]"
-  - "https://github.com/sting8k/pi-vcc/issues/3"
-related:
-  - "[[adr-012]]"
-  - "[[adr-015]]"
-created: 2026-05-07
-updated: 2026-05-07
----
-
-# ADR-027: pi-vcc Overrides Built-in Auto-Compaction
-
-## Context
-
-Pi's default compaction uses LLM-generated summaries. This is:
-- **Non-deterministic** — can hallucinate or lose context
-- **Expensive** — burns tokens on every compaction
-- **Slow** — waits for LLM call (seconds vs milliseconds)
-- **Destructive** — pre-compaction history is permanently lost
-
-[pi-vcc](https://github.com/sting8k/pi-vcc) is an algorithmic compaction extension inspired by lllyasviel's VCC. It provides:
-- **Deterministic** — same input = same output, no LLM
-- **Zero cost** — no API calls, algorithmic extraction
-- **Fast** — 30-470ms per compaction
-- **99% token reduction** on long sessions
-- **Lossless recall** — `vcc_recall` reads raw session JSONL
-
-### Problem: Model Stops After Auto-Compaction
-
-GitHub issue [sting8k/pi-vcc#3](https://github.com/sting8k/pi-vcc/issues/3): when `overrideDefaultCompaction: true`, the AI model stops generating after auto-compaction. The maintainer confirmed it is "by design for now."
-
-**Root cause**: pi-vcc's `buildOwnCut` keeps the last user message and everything after it as the "kept tail." When that tail contains a completed turn (user message + assistant's finished response), the LLM sees a complete conversation after session reload and has no signal to continue working.
-
-## Decision
-
-### 1. Project-level pi-vcc activation (not global)
-
-pi-vcc is registered as a **project-level package** in `.pi/settings.json`, not in `~/.pi/agent/settings.json`. The config lives at `.pi/pi-vcc-config.json`, referenced via `PI_VCC_CONFIG_PATH` in `.env`.
-
-This keeps the VCC override scoped to this project only. Other projects use pi's default LLM compaction.
-
-### 2. Enable `overrideDefaultCompaction: true`
-
-All compaction paths (auto-threshold, `/compact`, `/pi-vcc`) route through pi-vcc's algorithmic compactor.
-
-### 3. Fix "model stops" with continuation directive
-
-For auto-compactions (not manual `/pi-vcc`), append `[Continue]` section to the summary:
-
-```
-[Continue]
-Continue working on the current task. The user's request is not yet complete. Use vcc_recall if you need context from earlier in the session.
-```
-
-This is patched in `node_modules/@sting8k/pi-vcc/src/hooks/before-compact.ts`, applied only when `isPiVcc === false` (auto-compact or `/compact`).
-
-## Alternatives Considered
-
-| Option | Pros | Cons |
-|--------|------|------|
-| **Keep default LLM compaction** | No risk of model stopping | Expensive, non-deterministic, slow |
-| **`overrideDefaultCompaction: false` (manual only)** | Safer, model never stops unexpectedly | User must run `/pi-vcc` manually; auto-compaction still burns LLM tokens |
-| **Fork pi-vcc repo** | Clean patch management | Overhead of maintaining fork, slow to upstream updates |
-| **Patch node_modules directly** (chosen) | Zero overhead, immediate fix | Patch lost on `npm update`; requires re-apply |
-| **Send synthetic "continue" message post-compaction** | Clean separation of concerns | Requires new `session_compact` handler; more invasive |
-| **Append continuation to summary** (chosen) | Simplest fix; minimal code change | Slightly longer summary text |
-
-## Consequences
-
-### Positive
-- **99% context reduction** on auto-compaction (vs ~70-80% with LLM)
-- **Zero compaction cost** — no LLM calls, saves tokens
-- **30-470ms latency** vs 2-10s LLM wait
-- **Deterministic output** — same session = same summary
-- **Lossless recall** via `vcc_recall` — pre-compaction history searchable
-- **Model continues working** after auto-compaction via continuation directive
-
-### Negative
-- Patch in node_modules is fragile — `npm update` removes it
-- Continuation directive may prompt unnecessary continuation if task IS actually complete (minor risk)
-- VCC config is project-scoped only — other projects don't benefit
-
-### Mitigations
-- Document the patch location so it can be re-applied after npm updates
-- Consider upstreaming the fix to pi-vcc repo
-- Continuation directive only fires on auto-compaction, not `/pi-vcc`

package/vault/wiki/decisions/colocate-wiki.md

@@ -1,34 +0,0 @@
----
-type: decision
-status: active
-priority: 2
-date: "2026-04-28"
-owner: "aryaniyaps"
-due_date: ""
-context: "Should the architecture wiki be co-located in the same repository as the application source code?"
-tags: [decision, architecture, documentation]
-created: "2026-04-28"
-updated: "2026-04-28"
-title: "Co-locating Wiki with Codebase"
----
-# Co-locating Wiki with Codebase
-
-## Context
-Deciding whether to store the codebase map and architecture wiki inside the source repository or in a separate documentation vault.
-
-## Options Considered
-1. **Co-location (Monorepo approach)**: Wiki lives in `wiki/` and `CLAUDE.md` in the root.
-2. **Separate Vault**: Codebase is purely code; wiki is an isolated Obsidian vault elsewhere.
-
-## Decision
-**Option 1: Co-location.**
-
-## Rationale
-- **Proximity**: Developers can update code and architecture documentation in the same commit, preventing the wiki from becoming stale (the "doc rot" problem).
-- **Tooling Integration**: Agent skills (`wiki`, `wiki-ingest`) can seamlessly traverse the codebase and cross-reference real files because they share the same filesystem context.
-- **Git as Single Source of Truth**: Branching, PR reviews, and versioning apply to code and its architectural reasoning simultaneously.
-
-## Consequences
-- The repository clone size slightly increases (markdown is negligible, but attachments/images could add up).
-- Non-code contributors (if any) need git access to update documentation.
-- We must enforce policies (e.g., `.gitignore` rules or pre-commit hooks) so that generated wiki artifacts don't interfere with the build process.

package/vault/wiki/entities/Anders Hejlsberg.md

@@ -1,29 +0,0 @@
----
-type: entity
-entity_type: person
-title: "Anders Hejlsberg"
-role: "Creator of Turbo Pascal, Delphi, C#, TypeScript"
-active: true
-tags: [person, language-design, typescript, csharp, microsoft]
-related:
-  - "[[hejlsberg-7-learnings]]"
-  - "[[fast-feedback-loops]]"
-  - "[[behavioral-compatibility-over-purity]]"
----
-
-# Anders Hejlsberg
-
-Creator of Turbo Pascal (Borland, 1983), Delphi (1995), lead architect of C# (Microsoft, 2000), and designer of TypeScript (2012). Self-taught compiler writer. Still writes code daily — "my calling is writing code."
-
-## Engineering Practices
-
-- **Fast feedback as first principle**: Turbo Pascal's impact came from shortening edit-compile-run to instants. TypeScript's value equally from tooling responsiveness as from the type system.
-- **Stay hands-on**: Discovered hash tables from a book while writing Turbo Pascal 2.0 — doubled compiler speed. Wrote the TypeScript type checker core himself. "I treated work as the place where I wouldn't work and then I go home and work."
-- **Compromise over purity**: TypeScript extended JavaScript instead of replacing it. C# merged VB6 ease with C++ power. "Languages do not succeed because they are perfectly designed. They succeed because they accommodate the way teams actually work."
-- **Behavioral compatibility**: When porting TypeScript compiler to Go, the goal was semantic fidelity — "the new compiler needed to behave exactly like the old one, including quirks and edge cases."
-- **Open development**: 2014 move to GitHub was TypeScript's turning point. "Our entire workflow is in the open... no secrets on this project."
-- **On AI coding**: Grounding/constraint matters more than generation. Type systems and refactoring tools become essential guardrails. "The most valuable tools in an AI-assisted workflow aren't the ones that generate the most code, but the ones that constrain it correctly."
-
-## Relevance to AI Coding Harness
-
-Hejlsberg's career is the strongest bridge between legendary programming and AI-assisted development. His 7 learnings (2026 GitHub blog) directly address AI coding workflow: fast feedback matters most, behavioral compatibility over purity, visibility builds trust, and deterministic constraints (type checkers, linters) as guardrails around AI output. The idea that "grounding matters more than generation" is a direct harness design principle.

package/vault/wiki/entities/Anthropic.md

@@ -1,17 +0,0 @@
----
-type: entity
-status: stub
-created: 2026-05-02
-updated: 2026-05-02
-tags: [entity, ai-lab]
----
-
-# Anthropic
-
-AI research company. Creator of Claude (Opus, Sonnet, Haiku models) and the Claude Code agentic coding tool.
-
-## Relevance to Harness
-
-- [[anthropic-prompt-best-practices]] — official prompting guidance
-- [[anthropic2026-harness-design]] — GAN-inspired generator-evaluator architecture
-- [[claude-code-architecture-vila-lab-2026]] — Claude Code architecture analysis

package/vault/wiki/entities/Augment Code.md

@@ -1,49 +0,0 @@
----
-type: entity
-entity_type: organization
-title: "Augment Code"
-created: 2026-04-30
-status: developing
-tags:
-  - ai-coding
-  - startup
-  - context-engine
-  - swe-bench
-website: https://www.augmentcode.com
-related:
-  - "[[Context Engine (AI Coding)]]"
-  - "[[Augment SWE-bench Agent GitHub]]"
-  - "[[Augment SWE-bench Pro Blog]]"
-updated: 2026-05-02
-
----# Augment Code
-
-Augment Code (also Augment Computing Inc.) builds AI coding assistants specifically designed for large, complex codebases. Their core differentiator is the Context Engine — a semantic search engine for code that understands relationships across hundreds of thousands of files.
-
-## Products
-- **Auggie Agent**: AI coding agent for VS Code, JetBrains, and CLI.
-- **Context Engine**: Semantic codebase indexing and retrieval (core technology).
-- **Context Engine MCP**: MCP server making the Context Engine available to any AI agent.
-- **Intent**: AI-powered code review.
-- **Slack integration**: Agent accessible via Slack.
-
-## Key Metrics
-- #1 on SWE-bench Pro (51.80%, Feb 2026).
-- #1 open-source SWE-bench Verified agent (65.4%).
-- 60-80% code review acceptance rate.
-- 1M+ files indexed.
-- 870 GitHub stars on open-source agent.
-
-## Key People
-- **Chris Kelly**: Presented at ERC 2025.
-- **Vinay Perneti**: VP of Engineering.
-- **Arash (AJ) Joobandi**: DevRel.
-
-## Philosophy
-- "Context is the new compiler."
-- "Contractor vs Employee" model: context makes the difference, not model intelligence.
-- Only 3 model choices — they curate models for users.
-- Planning in quarters, not years.
-
-## Relevance to Our Harness
-Augment Code is the closest analog to what we're building with the agentic coding harness. Their Context Engine's capabilities — semantic indexing, prompt enhancement, multi-source context — are directly implementable in our system. Their open-source SWE-bench agent provides a reference architecture for dual-model agent design.

package/vault/wiki/entities/Bjarne Stroustrup.md

@@ -1,26 +0,0 @@
----
-type: entity
-entity_type: person
-title: "Bjarne Stroustrup"
-role: "Creator of C++"
-active: true
-tags: [person, c++, language-design, columbia]
-related:
-  - "[[pragmatic-language-design]]"
----
-
-# Bjarne Stroustrup
-
-Creator of C++ (1985). Professor at Columbia University. Member of US National Academy of Engineering. Recipient of the 2018 Charles Stark Draper Prize.
-
-## Engineering Practices
-
-- **C compatibility as pragmatic choice**: "Within C++, there is a much smaller and cleaner language struggling to get out... [which] would have been an unimportant cult language." Stroustrup prioritized real-world adoption over theoretical purity.
-- **Evolutionary design**: The Design and Evolution of C++ documents decisions across decades. "Language design issues, fundamental design decisions, and history don't change very often."
-- **Programming is for others**: "Assumes that your aim is to eventually write programs that are good enough for others to use and maintain." Focus on fundamental concepts, not language-technical details.
-- **Static typing as safety and efficiency**: committed to static type checking over dynamic typing for systems programming. Type safety enables both performance and correctness guarantees.
-- **C++ Core Guidelines**: community-driven standardization of best practices, maintained on GitHub.
-
-## Relevance to AI Coding Harness
-
-Stroustrup's evolutionary approach maps to incremental harness improvement rather than rewrites. C++ compatibility with C demonstrates that working within existing ecosystem constraints produces wider adoption than clean-slate designs. The focus on programs "good enough for others to use and maintain" maps directly to harness quality gates: AI output must pass the same maintainability standards as human code. Static typing as guardrail echoes Hejlsberg's view that types constrain AI output correctly.

package/vault/wiki/entities/Bolt.new (StackBlitz).md

@@ -1,39 +0,0 @@
----
-type: entity
-title: "Bolt.new (StackBlitz)"
-created: 2026-05-03
-updated: 2026-05-03
-tags:
-  - company
-  - ai-coding
-  - platform
-  - open-source
-related:
-  - "[[Source: Bolt.new Architecture & Case Study]]"
-  - "[[Lovable (company)]]"
-  - "[[Claude Code]]"
-sources:
-  - "[[Source: Bolt.new Architecture & Case Study]]"
-
----# Bolt.new (StackBlitz)
-
-AI-powered full-stack web development platform by StackBlitz. Users prompt in natural language, AI builds, code executes instantly in browser-based WebContainers.
-
-## Key Numbers
-- Launched 2024. 0→$4M ARR in 4 weeks
-- 3M+ monthly developers (StackBlitz platform)
-- 99% reduction in development costs for users
-- Open source: [github.com/stackblitz/bolt.new](https://github.com/stackblitz/bolt.new)
-
-## Technology Stack
-- **Frontend**: Remix + React + Radix UI + Framer Motion + UnoCSS
-- **Core**: WebContainers (WebAssembly browser-based OS)
-- **AI**: Anthropic Claude (originally Claude 3.5 Sonnet)
-- **Backend**: Ruby on Rails (users, permissions, billing)
-- **Deployment**: Cloudflare Pages, Netlify
-
-## Key Innovation
-**WebContainers** give AI complete control over the development environment: filesystem, Node.js runtime, package manager, terminal, browser console. This is what enables the AI to not just generate code but **run it, verify it, and fix issues before the user sees them.**
-
-## Founding
-Founded 2018 by Eric Simons and Albert Pai. Seed: $7.9M (Greylock, Google Ventures, Tom Preston-Werner). Evil Martians as key technical partner.

package/vault/wiki/entities/Boris Cherny.md

@@ -1,11 +0,0 @@
----
-type: entity
-status: stub
-created: 2026-05-02
-updated: 2026-05-02
-tags: [entity, person]
----
-
-# Boris Cherny
-
-Author of "TypeScript Cookbook" and engineering blog posts on TypeScript architecture. Referenced in Claude Code architecture analysis for TypeScript design insights.

package/vault/wiki/entities/Claude Code.md

@@ -1,19 +0,0 @@
----
-type: entity
-status: stub
-created: 2026-05-02
-updated: 2026-05-02
-tags: [entity, tool, coding-agent]
----
-
-# Claude Code
-
-Anthropic's agentic coding tool (82K+ GitHub stars). TypeScript codebase with sophisticated harness architecture.
-
-## Relevance to Harness
-
-- [[claude-code-architecture-vila-lab-2026]] — comprehensive architecture analysis
-- [[claude-code-architecture-qubytes-2026]] — five-layer architecture breakdown
-- [[claude-code-architecture-karaxai-2026]] — full stack walkthrough
-- [[claude-code-security-architecture-penligent-2026]] — security architecture
-- [[Research: Claude Code State-of-the-Art Harness Improvements]]

package/vault/wiki/entities/Dennis Ritchie.md

@@ -1,26 +0,0 @@
----
-type: entity
-entity_type: person
-title: "Dennis Ritchie"
-role: "Co-creator of Unix and C programming language"
-active: false
-tags: [person, unix, c-language, bell-labs]
-related:
-  - "[[Ken Thompson]]"
-  - "[[unix-philosophy]]"
-  - "[[birth-of-unix-kernighan-interview]]"
----
-
-# Dennis Ritchie
-
-Co-creator of Unix and creator of the C programming language at Bell Labs. Turing Award recipient (1983) alongside Ken Thompson. C became the foundation language for systems programming, directly influencing C++, Java, C#, Go, Rust, and virtually every major systems language. K&R C (Kernighan & Ritchie, 1978) defined programming style for generations.
-
-## Engineering Practices
-
-- **1974 Unix paper design considerations** (with Thompson): Make it easy to write, test, and run programs. Interactive use over batch processing. Economy and elegance of design due to size constraints ("salvation through suffering"). Self-supporting system: all Unix software maintained under Unix.
-- **Community via shared code**: Brian Kernighan recalled Ritchie would bring copies of Private Eye (British satire magazine) to the Unix Room — culture of shared space and shared interests fostering collaboration.
-- **K&R brace style**: still the dominant C brace style in systems programming, adopted by the Linux kernel.
-
-## Relevance to AI Coding Harness
-
-Ritchie's C language philosophy — give programmers power without unnecessary abstraction — maps to harness design that exposes clear, simple interfaces rather than opaque abstraction layers. "Salvation through suffering" (economy from constraint) maps to token budgeting as a design constraint that produces better output. The K&R style influence shows that coding conventions, once established and enforced, create ecosystem-wide readability benefits.

package/vault/wiki/entities/Emergent Labs.md

@@ -1,32 +0,0 @@
----
-type: entity
-title: "Emergent Labs"
-created: 2026-05-03
-updated: 2026-05-03
-tags:
-  - company
-  - ai-coding
-  - platform
-  - yc
-related:
-  - "[[Lovable (company)]]"
-  - "[[Bolt.new (StackBlitz)]]"
-  - "[[Rocket.new]]"
-
----# Emergent Labs
-
-YC S24. Builds autonomous coding agents that "replace traditional software development by generating, testing, and deploying production applications directly from plain-language intent."
-
-## Product
-- Full-stack web and mobile apps from conversation
-- AI agents that "design, code, and deploy your application from start to finish"
-- Production-ready, running at global scale, used to build "millions of real applications"
-
-## Key Claim
-"Software is being rewritten. Emergent builds autonomous coding agents that replace traditional software development."
-
-## Hiring Focus
-"AI Agent Reliability Engineer" — signals they're building infrastructure for agent reliability at scale. Systems run in production globally.
-
-## Note
-Limited public technical information available. Website (emergent.sh) is JS-heavy with minimal scraped content. Primary signals: YC backing, production scale claims, agent reliability engineering focus.

package/vault/wiki/entities/Google Cloud.md

@@ -1,16 +0,0 @@
----
-type: entity
-status: stub
-created: 2026-05-02
-updated: 2026-05-02
-tags: [entity, cloud]
----
-
-# Google Cloud
-
-Google's cloud computing platform. Hosts Gemini models and Vertex AI.
-
-## Relevance to Harness
-
-- [[gemini-3-prompting-guide]] — official Gemini prompting on Vertex AI
-- [[Source: Google Blog - Gemini CLI Announcement]] — Gemini CLI launch

package/vault/wiki/entities/Guido van Rossum.md

@@ -1,28 +0,0 @@
----
-type: entity
-entity_type: person
-title: "Guido van Rossum"
-role: "Creator of Python"
-active: true
-tags: [person, python, language-design, microsoft]
-related:
-  - "[[guido-python-design-philosophy]]"
-  - "[[pragmatic-language-design]]"
----
-
-# Guido van Rossum
-
-Creator of Python (1991). Python's BDFL (Benevolent Dictator For Life) until 2018. Distinguished Engineer at Microsoft since 2020.
-
-## Engineering Practices
-
-- **"Good enough" over perfection**: Python was a skunkworks project. "Don't try for perfection because good enough is often just that. It's okay to cut corners sometimes, especially if you can do it right later."
-- **Simplicity as core value**: "Python fits in your brain" (Bruce Eckel). The Zen of Python: "Simple is better than complex. Readability counts."
-- **Library integration as success factor**: Python's success came from being easy to understand yet powerful, and from great third-party library integration (NumPy developed independently).
-- **Pragmatic evolution**: type hints above 10K lines, not for beginners. Deliberate departure from ABC's perfectionism — Python was open, evolutionary, and community-driven from the start.
-- **Cautionary on AI**: "I am definitely not looking forward to an AI-driven future. I'm not worried about AI wanting to kill us all, but I see too many people without ethics or morals getting enabled to do much more damage to society with less effort."
-- **On AI coding**: Doesn't use "vibe coding" — "we stay in control where it comes to architecture and API design."
-
-## Relevance to AI Coding Harness
-
-Van Rossum's philosophy directly challenges AI hype: maintain architectural control, write code humans can read, don't sacrifice simplicity for automation. The type-hint threshold (10K lines) suggests tiered harness behavior: lightweight dynamic checking for small modules, strict typing for larger codebases. Python's extensibility model (multiple levels of extensibility) maps to harness skill systems: progressively loaded capabilities rather than monolithic context.

package/vault/wiki/entities/Ken Thompson.md

@@ -1,28 +0,0 @@
----
-type: entity
-entity_type: person
-title: "Ken Thompson"
-role: "Co-creator of Unix, inventor of B language, UTF-8, Go"
-active: true
-tags: [person, unix, bell-labs, go]
-related:
-  - "[[Dennis Ritchie]]"
-  - "[[unix-philosophy]]"
-  - "[[birth-of-unix-kernighan-interview]]"
-  - "[[subtractive-design]]"
----
-
-# Ken Thompson
-
-Co-creator of Unix at Bell Labs (1969), inventor of the B programming language (precursor to C), co-inventor of UTF-8 encoding, and co-designer of the Go programming language at Google. Turing Award recipient (1983).
-
-## Engineering Practices
-
-- **Extreme productivity from deep understanding**: built first Unix in 3 weeks. Reverse-engineered a typesetter in hours — wrote disassembler, assembler, and B interpreter for an unfamiliar CPU in a single day.
-- **Subtractive design**: McIlroy recalled the Unix Room culture: "We used to sit around saying, 'What can we throw out? Why is there this option?'"
-- **Make it easy to write, test, and run programs**: core design principle from 1974 Ritchie-Thompson Unix paper.
-- **Pipes**: implemented the breakthrough composability mechanism that enabled the Unix tool ecosystem.
-
-## Relevance to AI Coding Harness
-
-Thompson's approach demonstrates that deep system understanding enables extreme leverage. For the harness: deep codebase understanding (semantic indexing, call graphs) must precede code generation. The "what can we throw out" ethic maps to a harness that proactively suggests deletions and simplifications, not just additions. Pipes map to composable agent architectures where specialized sub-tools chain together.

package/vault/wiki/entities/Lee et al.md

@@ -1,16 +0,0 @@
----
-type: entity
-status: stub
-created: 2026-05-02
-updated: 2026-05-02
-tags: [entity, academic]
----
-
-# Lee et al.
-
-Authors from Stanford / Together AI. Published Meta-Harness paper: outer-loop harness optimization framework.
-
-## Relevance to Harness
-
-- [[lee2026-meta-harness]] — Meta-Harness paper
-- [[self-evolving-harness]] — applied concept

package/vault/wiki/entities/Linus Torvalds.md

@@ -1,28 +0,0 @@
----
-type: entity
-entity_type: person
-title: "Linus Torvalds"
-role: "Creator of Linux kernel and Git"
-active: true
-tags: [person, linux, kernel, open-source]
-related:
-  - "[[linux-kernel-coding-workflow]]"
-  - "[[chain-of-trust-software]]"
-  - "[[fast-feedback-loops]]"
----
-
-# Linus Torvalds
-
-Creator of the Linux kernel (1991) and Git version control system (2005). Maintains the kernel through a hierarchical chain-of-trust model: subsystem maintainers review and merge patches, then Torvalds pulls from them during 2-week merge windows.
-
-## Key Engineering Practices
-
-- **Opinionated coding standards**: 8-char tabs, short functions, K&R braces. Enforced by style guide, not tooling alone.
-- **Don't break userspace**: backward compatibility is sacred. Regressions are the primary metric for release readiness.
-- **"Good taste" in code**: eliminating edge cases at the design level rather than handling them.
-- **Anti-hype on AI**: vibe coding is "horrible for production" but "fairly positive" for learning (2026). AI-generated code must carry human accountability.
-- **"Code is cheap. Show me the talk."** — values demonstrated understanding over volume of output.
-
-## Relevance to AI Coding Harness
-
-Torvalds' kernel workflow maps to harness quality gates: every AI-generated change must be reviewable, diffable, and attributable to a human gatekeeper. The chain-of-trust model suggests a tiered verification pipeline where simpler checks run before human review. The "don't break userspace" principle maps to behavioral regression testing as a mandatory harness layer.

package/vault/wiki/entities/Lovable (company).md

@@ -1,40 +0,0 @@
----
-type: entity
-title: "Lovable (company)"
-created: 2026-05-03
-updated: 2026-05-03
-tags:
-  - company
-  - ai-coding
-  - platform
-related:
-  - "[[Source: Lovable Architecture & Clone Analysis]]"
-  - "[[Bolt.new (StackBlitz)]]"
-  - "[[Rocket.new]]"
-  - "[[Emergent Labs]]"
-sources:
-  - "[[Source: Lovable Architecture & Clone Analysis]]"
-
----# Lovable (company)
-
-Formerly GPT Engineer. A full-stack AI development platform that transforms natural language into production-ready web applications.
-
-## Core Product
-- Full-stack web apps from natural language prompts
-- Frontend, backend, database, authentication, integrations
-- Code ownership via GitHub sync — integrates into existing engineering workflows
-- Browser-based with enterprise compliance (SOC 2 Type II, ISO 27001, GDPR)
-- SSO/SCIM, workspace roles, built-in security checks
-
-## Key Differentiator
-Lovable's breakthrough is the **orchestration layer on top of models**, not the models themselves. The system architecture bridges the "intent-to-execution chasm" that raw AI code generators fail at.
-
-## Architecture Pattern
-Multi-agent pipeline: Planner → Architect → Coder with Pydantic-typed structured handoffs between agents.
-
-## Competitors
-- Bolt.new (StackBlitz)
-- Rocket.new
-- Emergent Labs
-- Replit
-- Cursor

package/vault/wiki/entities/Martin Fowler.md

@@ -1,16 +0,0 @@
----
-type: entity
-status: stub
-created: 2026-05-02
-updated: 2026-05-02
-tags: [entity, person]
----
-
-# Martin Fowler
-
-Software engineer at Thoughtworks. Author of canonical harness engineering patterns (feedforward/feedback controls).
-
-## Relevance to Harness
-
-- [[Source: Martin Fowler - Harness Engineering]] — canonical feedforward/feedback pattern
-- [[feedforward-feedback-harness]] — applied to agent harness

package/vault/wiki/entities/Meng et al.md

@@ -1,16 +0,0 @@
----
-type: entity
-status: stub
-created: 2026-05-02
-updated: 2026-05-02
-tags: [entity, academic]
----
-
-# Meng et al.
-
-Authors of the agent harness survey paper (2026). Defined the formal H=(E,T,C,S,L,V) 6-component model covering 110+ papers and 23 systems.
-
-## Relevance to Harness
-
-- [[meng2026-agent-harness-survey]] — the survey paper
-- [[harness-h-formalism]] — H-formalism concept page

package/vault/wiki/entities/OpenAI.md

@@ -1,16 +0,0 @@
----
-type: entity
-status: stub
-created: 2026-05-02
-updated: 2026-05-02
-tags: [entity, ai-lab]
----
-
-# OpenAI
-
-AI research company. Creator of GPT models and ChatGPT.
-
-## Relevance to Harness
-
-- [[openai-prompt-guidance]] — official prompting guidance
-- [[Source: OpenAI Harness Engineering Five Principles]] — 5 principles from 1M-line agent-built codebase

package/vault/wiki/entities/Rocket.new.md

@@ -1,38 +0,0 @@
----
-type: entity
-title: "Rocket.new"
-created: 2026-05-03
-updated: 2026-05-03
-tags:
-  - company
-  - ai-coding
-  - strategy
-  - competitive-intelligence
-related:
-  - "[[Source: Rocket.new — Vibe Solutioning Platform]]"
-  - "[[Lovable (company)]]"
-  - "[[Bolt.new (StackBlitz)]]"
-sources:
-  - "[[Source: Rocket.new — Vibe Solutioning Platform]]"
-
----# Rocket.new
-
-"World's first Vibe Solutioning platform." Covers the full arc: market research → product strategy → app building → competitive intelligence — in one system with shared context.
-
-## Key Numbers
-- $15M seed (Accel, Salesforce Ventures, Together Fund) — Sept 2025
-- 1.5M+ users across 180 countries
-- 57 employees, HQ in Surat, India
-- Subscriptions: $25-$350/month; ARPU ~$4,000/year
-- Gross margins >50%
-
-## Three Capabilities
-1. **Solve**: Market analysis, what-to-build, GTM strategy, PRDs — consulting-style reports
-2. **Build**: Web apps, mobile apps, landing pages, SaaS, dashboards, Figma import
-3. **Intelligence**: Continuous competitor monitoring, daily briefs, pricing/social/hiring signals
-
-## Key Thesis
-"Code generation has become a commodity. Deciding what to build — and tracking what happens after — is what's missing. Running a business and just building a codebase are two different things."
-
-## Founder
-Vishal Virani, previously co-founder & CEO of DhiWise (which pivoted to Rocket).

package/vault/wiki/entities/VILA-Lab.md

@@ -1,15 +0,0 @@
----
-type: entity
-status: stub
-created: 2026-05-02
-updated: 2026-05-02
-tags: [entity, academic]
----
-
-# VILA-Lab
-
-Academic research lab. Published arxiv 2604.14228: comprehensive Claude Code architecture analysis.
-
-## Relevance to Harness
-
-- [[claude-code-architecture-vila-lab-2026]]