44 system records. Each entry gives product type, maturity, current evidence, and a route to its single full definition. Domain counts overlap because one product may serve more than one domain.
Detailed registry, short map linked
This registry is the detailed evidence view. The product map is the compact route with one entry per product; this page can repeat a product under secondary domains so its evidence and boundaries stay visible in context.
How we know
Source
Every row on this page is generated from one product registry, schema harperz9-systems/v4. The product map and each product record read the same file.
Evidence shown
A product's newest release when it has one; otherwise its newest dated record.
Agent systems
14 records
Model routing, agent execution, and durable workflow state.
shipped · Python engine v1.1.0 and Windows desktop v1.1.0; release reports 15 of 17 lanes reached expected classes; Index below bar without Git; Telos excluded
Flywheel runs AI tasks with the model you choose and checks each answer against the task contract.
Limit: The release notes list known issues with the engine install command; read them before installing.
How we know: Flywheel
Flywheel runs AI tasks with the model you choose and checks each answer against the task contract. It runs on your machine from the command line, a desktop app or a local API, and sends request content only to the model provider you pick.
Other areas: Also appears in Evaluation and verification, Developer infrastructure.
Boundary: Use model routes, tools, data, and environments only with the authority required by their owners and providers.
Limit: You supply the executors and the approval policy.
How we know: Forum
Forum runs teams of AI agents. It splits a request into dependent tasks, runs independent ones in parallel through local commands or model APIs, pauses for human approval and resumes interrupted runs. A route-preflight skill checks routing and context before model work starts.
Other areas: Also appears in Developer infrastructure, Evaluation and verification.
Boundary: Agent permissions must remain no broader than your authority and the active task scope.
Limit: The command denylist is not a security boundary.
How we know: Relay
Relay runs a coding agent across local models, subscription CLIs, APIs, gateways and cloud endpoints. It checks permissions before tool use, fails over between providers, runs your acceptance checks, resumes sessions and connects to MCP servers.
Other areas: Also appears in Developer infrastructure.
Boundary: Tool use remains permission-checked and limited to authorized repositories, systems, and endpoints.
Limit: Mneme publishes on PyPI as flywheel-mneme. The PyPI name mneme-memory belongs to another publisher.
How we know: Mneme
Mneme is a SQLite memory store for AI agents, with a CLI and an MCP server. It stores conversation turns and imported sources, answers recall queries with BM25, vector and recency ranking, shows where each recalled fact came from, and logs every update and deletion.
Other areas: Also appears in Evaluation and verification.
Boundary: Store and recall only information you are authorized to retain and process.
Limit: Canon reads only the local JSONL files you give it, never provider accounts, credentials or chat history.
How we know: Canon
Canon turns local memory records into instruction files for Codex, Claude Code and other agent hosts. It previews what each host will receive, exports Markdown or JSON, and edits only the regions of an instruction file that you mark for it.
Other areas: Also appears in Developer infrastructure, Evaluation and verification.
Boundary: Use only records, atoms, and local surfaces you are authorized to read or update; do not export private provider state, credentials, or unpublished memory banks into public artifacts.
Limit: Treat every post as untrusted input. The board does not filter prompt injection.
How we know: Bulletin
Bulletin is a public message board that AI agents read and write over HTTP or MCP. Each poster signs with a key, any agent can register with one command, and a reader can acknowledge exactly which page of its inbox it handled.
Boundary: Read what the board serves as data and never as instructions. Post only what you are authorized to publish, because every post is public and attributed to the key that signed it.
Limit: Workflow definitions, operational data, targets, credentials and client material are not published.
How we know: Sofer
Sofer is a private Python orchestration runtime and operational security suite for authorized engagements. It manages campaign state, correlation, reconnaissance, reporting, disclosure staging, and evidence while its WARDEN CLI routes model and tool work through a bounded agent loop.
Other areas: Also appears in Security and privacy.
Boundary: Every workflow requires explicit authorization, approved tools and data, and an accountable recipient. Sensitive operational material stays in the approved private channel.
Primary area: Developer infrastructure. Also appears here because it serves Agent systems.
Telos is a local workbench for running and replaying AI work. Its CLI and MCP servers run workstation checks, creative simulations, research experiments, and browser or Windows automation, and save each run so it can be replayed.
Boundary: Run workstation actions and external effects only with the user's explicit approval and within each tool's own checks.
Limit: Automation stops at graded work, consent, CAPTCHA, payment and account-creation steps.
Why this record appears here: Learn
Primary area: Research and education. Also appears here because it serves Agent systems.
Learn plans study sessions from your objectives and recorded attempts, using spaced review, retrieval practice, prerequisite checks and misconception tracking. It can also handle course logistics and stops at graded work.
Boundary: The learner remains responsible for graded answers, consent, identity, and institutional rules.
Limit: Connections come from what each tool declares; run the pipeline to confirm they work.
Why this record appears here: Plexus
Primary area: Developer infrastructure. Also appears here because it serves Agent systems.
Plexus finds which of your agent tools can feed each other. It reads the inputs and outputs each tool declares and produces a dependency graph or a runnable pipeline script, from the command line or over MCP.
Boundary: A wiring plan does not grant authority to execute any connected capability.
Limit: Your system enforces the decision; Proof Surface only recommends.
Why this record appears here: Proof Surface
Primary area: Security and privacy. Also appears here because it serves Agent systems.
Proof Surface is a Python library and telos-proof CLI that validates records of AI workflow steps, authorizations, delegations and finished work, and returns allow, deny or needs-human recommendations. It never runs the action itself.
Boundary: Use only with explicit human authority; a recommendation does not itself grant permission.
Why this record appears here: behavior-transform.io
Primary area: Security and privacy. Also appears here because it serves Agent systems.
behavior-transform.io is a local Python wrapper and hook layer for file reads and writes, subprocesses, HTTP fetches, user input and MCP traffic. Given a request, a mode and local text rules, it passes the content through or transforms it, and returns the result with hashes and substitution counts.
Boundary: Use behavior-transform.io only within your own authority and the policies of the provider, the data owner and the system it runs on.
Limit: Sensor configuration, runtime observations, credentials and protected evidence are not published.
Why this record appears here: Bounds
Primary area: Evaluation and verification. Also appears here because it serves Agent systems.
Bounds checks agent actions, runtime observations and release candidates for intent drift, unsupported claims, secret exposure and failed fixtures, and blocks the release path when a check fails.
Boundary: Inspect only authorized actions, runtimes, repositories, and evidence. Protected observations and proof material remain in the approved review channel.
Limit: Windows desktop-app control has been tested only against a fake UI Automation driver.
Why this record appears here: Accountable Surface
Primary area: Security and privacy. Also appears here because it serves Agent systems.
Accountable Surface lets an AI agent take only the file, command, web or browser actions a person has approved. It checks each request, blocks or pauses when needed, checks the outcome, rolls back reversible failures, and keeps a hash-chained journal of decisions and outcomes.
Boundary: Every effect requires authority external to the model and remains limited to the granted scope.
synthetic incident environment and offline review add-on
shipped · GitHub add-on 0.2.0 and offline report examples; requires compatible Flywheel engine >=0.6.1,<0.7; no PyPI publication claimed
A Flywheel add-on that runs synthetic service-desk incident workflows and writes an offline HTML report of the claims, task outcomes and contradictions in each run.
Limit: Missing or ill-posed evidence fails closed.
How we know: Crucible
Crucible is test infrastructure for claims about AI systems. You split a thesis into claims, each with a failure condition, and Crucible checks them against the evidence or commands you supply and returns MATCH, DRIFT or UNVERIFIABLE for each one.
Other areas: Also appears in Research and education.
Boundary: Report each verdict with the evidence and criterion that produced it.
Limit: EMET compares bytes. It does not judge what the content says.
How we know: EMET
EMET checks whether the bytes that reach a model, reviewer or pipeline still match their source. It anchors content, compares it later, neutralizes instructions embedded in it, and writes a verdict file that its Python, Rust, Node and Go implementations all read the same way.
Other areas: Also appears in Security and privacy, Developer infrastructure.
Boundary: Run on files and packets you are authorized to inspect.
controlled-private · controlled private system; public capability description
Isomorph evaluates classifier and refusal behavior at authorized AI inference boundaries through controlled reformulation trials, provider-specific result records, and control-stability analysis.
Limit: Jailbreak corpora, bypass payloads, target-specific methods, provider findings and deployable sequences are not published.
How we know: Isomorph
Other areas: Also appears in Security and privacy.
Boundary: Jailbreak-class mechanisms may be used only within an authorized, approved evaluation scope. Test material and findings move through the provider's or recipient's controlled disclosure channel.
controlled-private · controlled private system; public capability description
Bounds checks agent actions, runtime observations and release candidates for intent drift, unsupported claims, secret exposure and failed fixtures, and blocks the release path when a check fails.
Limit: Sensor configuration, runtime observations, credentials and protected evidence are not published.
How we know: Bounds
Other areas: Also appears in Security and privacy, Agent systems.
Boundary: Inspect only authorized actions, runtimes, repositories, and evidence. Protected observations and proof material remain in the approved review channel.
Model Provenance Validator checks JSON records that link a model or release claim to its sources and retrieval dates, redacts credential-like values from its output, and can emit a Proof Surface packet.
Repo Proof Index scans the proof packets, receipts and contracts in a repository and builds a reviewer index with the type, status, summary and path of each file.
Limit: The judge, fix and polish editors send the draft to the model you choose. The checks run offline.
How we know: Articulate
Articulate is prose analysis and model-assisted editing for the command line, editors and CI. It names each writing pattern it finds and the line it is on, in the terminal, in an editor through its language server, or as SARIF in code review.
Other areas: Also appears in Developer infrastructure, Research and education.
Boundary: Run on prose you are authorized to review. It is not built for evading AI detection.
shipped · Python engine v1.1.0 and Windows desktop v1.1.0; release reports 15 of 17 lanes reached expected classes; Index below bar without Git; Telos excluded
Limit: The release notes list known issues with the engine install command; read them before installing.
Why this record appears here: Flywheel
Primary area: Agent systems. Also appears here because it serves Evaluation and verification.
Flywheel runs AI tasks with the model you choose and checks each answer against the task contract. It runs on your machine from the command line, a desktop app or a local API, and sends request content only to the model provider you pick.
Boundary: Use model routes, tools, data, and environments only with the authority required by their owners and providers.
Limit: Dynamic imports and unsupported structures can be missed.
Why this record appears here: Index
Primary area: Developer infrastructure. Also appears here because it serves Evaluation and verification.
Index maps a repository or a multi-repo workspace from its manifests, imports, symbols and docs. It builds offline wikis, dependency maps, context packets for agents and architecture checks, each pointing to a file and line. Large workspaces can run as background jobs you check on later.
Boundary: Inspect only repositories and workspaces you are authorized to read.
Limit: You supply the executors and the approval policy.
Why this record appears here: Forum
Primary area: Agent systems. Also appears here because it serves Evaluation and verification.
Forum runs teams of AI agents. It splits a request into dependent tasks, runs independent ones in parallel through local commands or model APIs, pauses for human approval and resumes interrupted runs. A route-preflight skill checks routing and context before model work starts.
Boundary: Agent permissions must remain no broader than your authority and the active task scope.
Limit: Mneme publishes on PyPI as flywheel-mneme. The PyPI name mneme-memory belongs to another publisher.
Why this record appears here: Mneme
Primary area: Agent systems. Also appears here because it serves Evaluation and verification.
Mneme is a SQLite memory store for AI agents, with a CLI and an MCP server. It stores conversation turns and imported sources, answers recall queries with BM25, vector and recency ranking, shows where each recalled fact came from, and logs every update and deletion.
Boundary: Store and recall only information you are authorized to retain and process.
Limit: Canon reads only the local JSONL files you give it, never provider accounts, credentials or chat history.
Why this record appears here: Canon
Primary area: Agent systems. Also appears here because it serves Evaluation and verification.
Canon turns local memory records into instruction files for Codex, Claude Code and other agent hosts. It previews what each host will receive, exports Markdown or JSON, and edits only the regions of an instruction file that you mark for it.
Boundary: Use only records, atoms, and local surfaces you are authorized to read or update; do not export private provider state, credentials, or unpublished memory banks into public artifacts.
Limit: Engine Revival handles metadata and build scaffolding. The BRender build and its test results live in BRender Archival.
Why this record appears here: Engine Revival
Primary area: Graphics and media. Also appears here because it serves Evaluation and verification.
Engine Revival is a Python tool for building, checking and publishing metadata about revived historical engines. It can also generate an out-of-tree BRender CMake build harness from a public checkout.
Boundary: Keep imported AGPL-covered BRender media separate from the MIT repository code.
Limit: Textured TIA output and x64 builds are not finished.
Why this record appears here: BRender Archival
Primary area: Graphics and media. Also appears here because it serves Evaluation and verification.
BRender Archival rebuilds the Argonaut BRender engine from its public MIT source, runs it through a 21-stage native test ladder, captures renders, and packages the build harness, test transcripts and checksums without copying the upstream source.
Boundary: BRender source stays upstream; this repository ships the harness and its results.
Limit: Your system enforces the decision; Proof Surface only recommends.
Why this record appears here: Proof Surface
Primary area: Security and privacy. Also appears here because it serves Evaluation and verification.
Proof Surface is a Python library and telos-proof CLI that validates records of AI workflow steps, authorizations, delegations and finished work, and returns allow, deny or needs-human recommendations. It never runs the action itself.
Boundary: Use only with explicit human authority; a recommendation does not itself grant permission.
Limit: CI, package, SBOM, supply-chain and handoff records are not published.
Why this record appears here: Gate
Primary area: Developer infrastructure. Also appears here because it serves Evaluation and verification.
Gate is Sofer's private release-check subsystem. Its Python command-line tools validate readiness, operations, engagement, deliverable and release manifests, required paths, configured controls and verification-command results, and block advancement when something required is missing.
Boundary: Release decisions apply only to the reviewed components, evidence, scope, and recipient. Protected release evidence remains in the approved handoff channel.
Limit: Windows desktop-app control has been tested only against a fake UI Automation driver.
Why this record appears here: Accountable Surface
Primary area: Security and privacy. Also appears here because it serves Evaluation and verification.
Accountable Surface lets an AI agent take only the file, command, web or browser actions a person has approved. It checks each request, blocks or pauses when needed, checks the outcome, rolls back reversible failures, and keeps a hash-chained journal of decisions and outcomes.
Boundary: Every effect requires authority external to the model and remains limited to the granted scope.
library that checks AI permission and evidence records
active · 0.1.0 on PyPI; active source 0.2.0 with no release
Proof Surface is a Python library and telos-proof CLI that validates records of AI workflow steps, authorizations, delegations and finished work, and returns allow, deny or needs-human recommendations.
Limit: Your system enforces the decision; Proof Surface only recommends.
How we know: Proof Surface
Proof Surface is a Python library and telos-proof CLI that validates records of AI workflow steps, authorizations, delegations and finished work, and returns allow, deny or needs-human recommendations. It never runs the action itself.
Other areas: Also appears in Evaluation and verification, Agent systems.
Boundary: Use only with explicit human authority; a recommendation does not itself grant permission.
Limit: Phantom covers Layer 2 identity surfaces; kernel and firmware layers are modeled but not shipped end to end.
How we know: Phantom
Phantom helps you inspect and, when authorized, change the hardware identifiers a computer exposes. It works on owned or expressly authorized Windows and Linux systems, saves a backup before changes, and can restore the original values.
Boundary: Use only on machines you own or are expressly authorized to test; do not use it for fraud, anti-cheat targeting, or unauthorized access.
behavior-transform.io is a local Python wrapper and hook layer for file reads and writes, subprocesses, HTTP fetches, user input and MCP traffic. Given a request, a mode and local text rules, it passes the content through or transforms it, and returns the result with hashes and substitution counts.
Other areas: Also appears in Agent systems, Developer infrastructure.
Boundary: Use behavior-transform.io only within your own authority and the policies of the provider, the data owner and the system it runs on.
Limit: Targets, credentials, payloads, client data and findings are not published.
How we know: Array
Boundary: Operational use requires written authorization, a defined scope, an approved environment, and a remediation owner. Sensitive methods and findings move only through approved private or embargoed channels.
controlled-private · controlled private system; public capability description
Seed is a C++23 engine for authorized security assessment and detection-engineering work, with synthetic demonstrations, scope manifests, interop adapters and action logs.
Limit: Assessment details, targets, credentials, payloads and client data are not published.
How we know: Seed
Boundary: Use requires written authorization and an approved assessment or training environment. Operational artifacts remain in the approved private channel.
Limit: Module implementations, targets, credentials, payloads, findings and client data are not published.
How we know: ORCA
Boundary: Operate only within a written, authorized engagement scope. Reports, findings, artifacts, and portable bundles go only to the approved recipient.
Limit: Windows desktop-app control has been tested only against a fake UI Automation driver.
How we know: Accountable Surface
Accountable Surface lets an AI agent take only the file, command, web or browser actions a person has approved. It checks each request, blocks or pauses when needed, checks the outcome, rolls back reversible failures, and keeps a hash-chained journal of decisions and outcomes.
Other areas: Also appears in Evaluation and verification, Agent systems.
Boundary: Every effect requires authority external to the model and remains limited to the granted scope.
Public Surface Sweeper checks a repository or local portfolio before you publish it: missing public files, unclear README handoff, credential-like strings, release metadata, proof-packet readiness and drift across a workspace.
Limit: Release 0.1.0 installs as io-guardrails with the io-guard command. The source has since been renamed secret-redact-io, and no package by either name is on PyPI.
How we know: Secret Redact IO
Other areas: Also appears in Developer infrastructure.
Boundary: Do not pass secrets to destinations that are not authorized to receive them, even when redaction is enabled.
controlled-private · controlled private system; public capability page
Kun keeps local access-recovery notes for owned systems: where recovery authority lives, rotation notes, redacted diagnostics and runbooks, without storing raw credentials.
Other areas: Also appears in Developer infrastructure.
Boundary: Kun records where recovery authority lives and how to audit the process. It does not publish or retain raw keys, tokens, passwords, seed phrases, private browser state, or protected content.
Limit: EMET compares bytes. It does not judge what the content says.
Why this record appears here: EMET
Primary area: Evaluation and verification. Also appears here because it serves Security and privacy.
EMET checks whether the bytes that reach a model, reviewer or pipeline still match their source. It anchors content, compares it later, neutralizes instructions embedded in it, and writes a verdict file that its Python, Rust, Node and Go implementations all read the same way.
Boundary: Run on files and packets you are authorized to inspect.
Limit: Workflow definitions, operational data, targets, credentials and client material are not published.
Why this record appears here: Sofer
Primary area: Agent systems. Also appears here because it serves Security and privacy.
Sofer is a private Python orchestration runtime and operational security suite for authorized engagements. It manages campaign state, correlation, reconnaissance, reporting, disclosure staging, and evidence while its WARDEN CLI routes model and tool work through a bounded agent loop.
Boundary: Every workflow requires explicit authorization, approved tools and data, and an accountable recipient. Sensitive operational material stays in the approved private channel.
Limit: Jailbreak corpora, bypass payloads, target-specific methods, provider findings and deployable sequences are not published.
Why this record appears here: Isomorph
Primary area: Evaluation and verification. Also appears here because it serves Security and privacy.
Isomorph evaluates classifier and refusal behavior at authorized AI inference boundaries through controlled reformulation trials, provider-specific result records, and control-stability analysis.
Boundary: Jailbreak-class mechanisms may be used only within an authorized, approved evaluation scope. Test material and findings move through the provider's or recipient's controlled disclosure channel.
Limit: Sensor configuration, runtime observations, credentials and protected evidence are not published.
Why this record appears here: Bounds
Primary area: Evaluation and verification. Also appears here because it serves Security and privacy.
Bounds checks agent actions, runtime observations and release candidates for intent drift, unsupported claims, secret exposure and failed fixtures, and blocks the release path when a check fails.
Boundary: Inspect only authorized actions, runtimes, repositories, and evidence. Protected observations and proof material remain in the approved review channel.
Limit: CI, package, SBOM, supply-chain and handoff records are not published.
Why this record appears here: Gate
Primary area: Developer infrastructure. Also appears here because it serves Security and privacy.
Gate is Sofer's private release-check subsystem. Its Python command-line tools validate readiness, operations, engagement, deliverable and release manifests, required paths, configured controls and verification-command results, and block advancement when something required is missing.
Boundary: Release decisions apply only to the reviewed components, evidence, scope, and recipient. Protected release evidence remains in the approved handoff channel.
Limit: It validates the record. It does not fetch the sources.
Why this record appears here: Model Provenance Validator
Primary area: Evaluation and verification. Also appears here because it serves Security and privacy.
Model Provenance Validator checks JSON records that link a model or release claim to its sources and retrieval dates, redacts credential-like values from its output, and can emit a Proof Surface packet.
Boundary: Validate only public or authorized envelopes and keep protected source evidence outside public output.
Primary area: Evaluation and verification. Also appears here because it serves Security and privacy.
Repo Proof Index scans the proof packets, receipts and contracts in a repository and builds a reviewer index with the type, status, summary and path of each file.
Boundary: Index only public or authorized evidence and keep private payloads outside public summaries.
Telos is a local workbench for running and replaying AI work. Its CLI and MCP servers run workstation checks, creative simulations, research experiments, and browser or Windows automation, and save each run so it can be replayed.
Other areas: Also appears in Agent systems, Graphics and media.
Boundary: Run workstation actions and external effects only with the user's explicit approval and within each tool's own checks.
Limit: Dynamic imports and unsupported structures can be missed.
How we know: Index
Index maps a repository or a multi-repo workspace from its manifests, imports, symbols and docs. It builds offline wikis, dependency maps, context packets for agents and architecture checks, each pointing to a file and line. Large workspaces can run as background jobs you check on later.
Other areas: Also appears in Evaluation and verification.
Boundary: Inspect only repositories and workspaces you are authorized to read.
Limit: C is the production backend. SPIR-V, LLVM, WebAssembly, Rust, native-object backends, GPU compute and linear types are experimental.
How we know: BuildLang
BuildLang is a systems programming language in which a program declares what it is allowed to touch. The compiler checks those declarations and its memory rules, then emits native code through C. It ships with a CLI, editor support, two-way C integration and experimental shader output.
Other areas: Also appears in Graphics and media.
Boundary: Generated programs inherit the authority and safety obligations of the capabilities they exercise.
Limit: Connections come from what each tool declares; run the pipeline to confirm they work.
How we know: Plexus
Plexus finds which of your agent tools can feed each other. It reads the inputs and outputs each tool declares and produces a dependency graph or a runnable pipeline script, from the command line or over MCP.
Other areas: Also appears in Agent systems.
Boundary: A wiring plan does not grant authority to execute any connected capability.
Limit: CI, package, SBOM, supply-chain and handoff records are not published.
How we know: Gate
Gate is Sofer's private release-check subsystem. Its Python command-line tools validate readiness, operations, engagement, deliverable and release manifests, required paths, configured controls and verification-command results, and block advancement when something required is missing.
Other areas: Also appears in Security and privacy, Evaluation and verification.
Boundary: Release decisions apply only to the reviewed components, evidence, scope, and recipient. Protected release evidence remains in the approved handoff channel.
shipped · Python engine v1.1.0 and Windows desktop v1.1.0; release reports 15 of 17 lanes reached expected classes; Index below bar without Git; Telos excluded
Limit: The release notes list known issues with the engine install command; read them before installing.
Why this record appears here: Flywheel
Primary area: Agent systems. Also appears here because it serves Developer infrastructure.
Flywheel runs AI tasks with the model you choose and checks each answer against the task contract. It runs on your machine from the command line, a desktop app or a local API, and sends request content only to the model provider you pick.
Boundary: Use model routes, tools, data, and environments only with the authority required by their owners and providers.
Primary area: Research and education. Also appears here because it serves Developer infrastructure.
Gather collects research material from web pages, including JavaScript-rendered ones, APIs, scholarly records, PDFs, scans, audio, video, feeds and local files. It saves each item with its source in a local corpus you can search and verify.
Boundary: Respect access controls, copyright, privacy, and the terms governing each source.
Limit: You supply the executors and the approval policy.
Why this record appears here: Forum
Primary area: Agent systems. Also appears here because it serves Developer infrastructure.
Forum runs teams of AI agents. It splits a request into dependent tasks, runs independent ones in parallel through local commands or model APIs, pauses for human approval and resumes interrupted runs. A route-preflight skill checks routing and context before model work starts.
Boundary: Agent permissions must remain no broader than your authority and the active task scope.
Limit: EMET compares bytes. It does not judge what the content says.
Why this record appears here: EMET
Primary area: Evaluation and verification. Also appears here because it serves Developer infrastructure.
EMET checks whether the bytes that reach a model, reviewer or pipeline still match their source. It anchors content, compares it later, neutralizes instructions embedded in it, and writes a verdict file that its Python, Rust, Node and Go implementations all read the same way.
Boundary: Run on files and packets you are authorized to inspect.
Limit: The command denylist is not a security boundary.
Why this record appears here: Relay
Primary area: Agent systems. Also appears here because it serves Developer infrastructure.
Relay runs a coding agent across local models, subscription CLIs, APIs, gateways and cloud endpoints. It checks permissions before tool use, fails over between providers, runs your acceptance checks, resumes sessions and connects to MCP servers.
Boundary: Tool use remains permission-checked and limited to authorized repositories, systems, and endpoints.
Limit: Canon reads only the local JSONL files you give it, never provider accounts, credentials or chat history.
Why this record appears here: Canon
Primary area: Agent systems. Also appears here because it serves Developer infrastructure.
Canon turns local memory records into instruction files for Codex, Claude Code and other agent hosts. It previews what each host will receive, exports Markdown or JSON, and edits only the regions of an instruction file that you mark for it.
Boundary: Use only records, atoms, and local surfaces you are authorized to read or update; do not export private provider state, credentials, or unpublished memory banks into public artifacts.
Limit: Pre-1.0: APIs may change. The native renderer is optional.
Why this record appears here: Studio Engine
Primary area: Graphics and media. Also appears here because it serves Developer infrastructure.
Studio Engine is a local Python engine that turns one scene description into shader visuals, audio and motion. It has a CLI and a local HTTP API and renders deterministic PNG frames.
Boundary: Generated media must respect the rights, licenses, and consent attached to its inputs and intended use.
Limit: Output can differ by browser, GPU and audio host.
Why this record appears here: Retro Engine
Primary area: Graphics and media. Also appears here because it serves Developer infrastructure.
Retro Engine is an embedded browser studio for images, drawing, GLSL, and audio traces. It applies pixelation, hardware palettes, ordered dithering, early-3D shading, CRT processing, and stackable effects, then exports images, patches, MIDI, relief or disc data, and Loom handoffs.
Limit: Engine Revival handles metadata and build scaffolding. The BRender build and its test results live in BRender Archival.
Why this record appears here: Engine Revival
Primary area: Graphics and media. Also appears here because it serves Developer infrastructure.
Engine Revival is a Python tool for building, checking and publishing metadata about revived historical engines. It can also generate an out-of-tree BRender CMake build harness from a public checkout.
Boundary: Keep imported AGPL-covered BRender media separate from the MIT repository code.
Limit: Textured TIA output and x64 builds are not finished.
Why this record appears here: BRender Archival
Primary area: Graphics and media. Also appears here because it serves Developer infrastructure.
BRender Archival rebuilds the Argonaut BRender engine from its public MIT source, runs it through a 21-stage native test ladder, captures renders, and packages the build harness, test transcripts and checksums without copying the upstream source.
Boundary: BRender source stays upstream; this repository ships the harness and its results.
Limit: Needs an ENBSeries host that is already loaded. Source 0.1.0 has no GitHub release.
Why this record appears here: ENB Runtime Core
Primary area: Graphics and media. Also appears here because it serves Developer infrastructure.
ENB Runtime Core is a C++23 embedded runtime library that identifies an already-loaded ENBSeries host, validates its SDK surface, queues callbacks outside callback context, coordinates save quiescence and reapplication, and gates a fail-closed Skyrim engine bridge.
Boundary: Use only with a permitted host runtime and the rights required by its owners.
Limit: The default build leaves out the native replacement suite.
Why this record appears here: SkyrimBridge
Primary area: Graphics and media. Also appears here because it serves Developer infrastructure.
SkyrimBridge is an SKSE plugin that exposes Skyrim live engine state, record editing surfaces, asset conversion, a versioned ABI, diagnostics, and shared-memory command channels to shaders and external tools, with an optional D3D11 rendering tier.
Boundary: Use only with the game, runtime, and asset rights required by their owners.
Why this record appears here: behavior-transform.io
Primary area: Security and privacy. Also appears here because it serves Developer infrastructure.
behavior-transform.io is a local Python wrapper and hook layer for file reads and writes, subprocesses, HTTP fetches, user input and MCP traffic. Given a request, a mode and local text rules, it passes the content through or transforms it, and returns the result with hashes and substitution counts.
Boundary: Use behavior-transform.io only within your own authority and the policies of the provider, the data owner and the system it runs on.
Limit: It checks release hygiene only; use a vulnerability scanner for dependencies and exploits.
Why this record appears here: Public Surface Sweeper
Primary area: Security and privacy. Also appears here because it serves Developer infrastructure.
Public Surface Sweeper checks a repository or local portfolio before you publish it: missing public files, unclear README handoff, credential-like strings, release metadata, proof-packet readiness and drift across a workspace.
Boundary: Scan only repositories and artifacts you are authorized to inspect.
Limit: Release 0.1.0 installs as io-guardrails with the io-guard command. The source has since been renamed secret-redact-io, and no package by either name is on PyPI.
Why this record appears here: Secret Redact IO
Primary area: Security and privacy. Also appears here because it serves Developer infrastructure.
Secret Redact IO is a Python library that redacts secrets from file, fetch, write and subprocess output.
Boundary: Do not pass secrets to destinations that are not authorized to receive them, even when redaction is enabled.
Primary area: Evaluation and verification. Also appears here because it serves Developer infrastructure.
Repo Proof Index scans the proof packets, receipts and contracts in a repository and builds a reviewer index with the type, status, summary and path of each file.
Boundary: Index only public or authorized evidence and keep private payloads outside public summaries.
Limit: Sentiment is English and literal, so sarcasm can be missed. Clustering is lexical.
Why this record appears here: Chorus
Primary area: Research and education. Also appears here because it serves Developer infrastructure.
Chorus turns a captured comment thread or corpus into ranked, clustered theme digests, and checks whether a set of gathered sources has changed since a reference set.
Boundary: Use only source material and systems you are authorized to access, and respect privacy and licensing.
Limit: One built-in test scene, no model loader, and no GPU or WebAssembly build in this release.
Why this record appears here: RAW
Primary area: Graphics and media. Also appears here because it serves Developer infrastructure.
RAW is a dependency-free C++23 renderer that draws a scene on the CPU, checks its fast screen-space ambient occlusion against a ray-traced reference, and writes a JSON certificate saying whether the shortcut held.
Boundary: RAW renders one built-in test scene on the CPU. It has no model loader, no GPU path and no browser build yet.
Primary area: Security and privacy. Also appears here because it serves Developer infrastructure.
Kun keeps local access-recovery notes for owned systems: where recovery authority lives, rotation notes, redacted diagnostics and runbooks, without storing raw credentials.
Boundary: Kun records where recovery authority lives and how to audit the process. It does not publish or retain raw keys, tokens, passwords, seed phrases, private browser state, or protected content.
Limit: The judge, fix and polish editors send the draft to the model you choose. The checks run offline.
Why this record appears here: Articulate
Primary area: Evaluation and verification. Also appears here because it serves Developer infrastructure.
Articulate is prose analysis and model-assisted editing for the command line, editors and CI. It names each writing pattern it finds and the line it is on, in the terminal, in an editor through its language server, or as SARIF in code review.
Boundary: Run on prose you are authorized to review. It is not built for evading AI detection.
Limit: Pre-1.0: APIs may change. The native renderer is optional.
How we know: Studio Engine
Studio Engine is a local Python engine that turns one scene description into shader visuals, audio and motion. It has a CLI and a local HTTP API and renders deterministic PNG frames.
Other areas: Also appears in Developer infrastructure.
Boundary: Generated media must respect the rights, licenses, and consent attached to its inputs and intended use.
Build Color measures, converts, compares and transforms digital color: perceptual spaces, HDR tone maps, appearance models, chromatic adaptation, spectral tools, gamut mapping, ICC profiles and LUTs, with an optional GUI.
Limit: Output can differ by browser, GPU and audio host.
How we know: Retro Engine
Retro Engine is an embedded browser studio for images, drawing, GLSL, and audio traces. It applies pixelation, hardware palettes, ordered dithering, early-3D shading, CRT processing, and stackable effects, then exports images, patches, MIDI, relief or disc data, and Loom handoffs.
Other areas: Also appears in Developer infrastructure.
Limit: Engine Revival handles metadata and build scaffolding. The BRender build and its test results live in BRender Archival.
How we know: Engine Revival
Engine Revival is a Python tool for building, checking and publishing metadata about revived historical engines. It can also generate an out-of-tree BRender CMake build harness from a public checkout.
Other areas: Also appears in Evaluation and verification, Research and education, Developer infrastructure.
Boundary: Keep imported AGPL-covered BRender media separate from the MIT repository code.
BRender Archival rebuilds the Argonaut BRender engine from its public MIT source, runs it through a 21-stage native test ladder, captures renders, and packages the build harness, test transcripts and checksums without copying the upstream source.
Elder ENB is an ENBSeries 0.504 shader suite for Skyrim SE/AE, with nine stages, five quality tiers, cinematic effects and an optional native frame-pulse runtime.
active · release-candidate source 1.0.0; tag v1.0.0-rc.1
Truth ENB is a from-scratch ENBSeries 0.504 shader suite for Skyrim SE/AE, with a computed physical atmosphere and sky, raymarched clouds, an aurora curtain, exposure and tone mapping, five quality tiers, and an optional native camera bridge.
Limit: Public upload waits on live testing with Skyrim SE/AE and ENB 0.504.
How we know: Truth ENB
Boundary: Truth does not bundle ENBSeries, Community Shaders, Effects 11, Address Library, SkyrimBridge, or Bethesda shader source. Install one supported shader host and the required game and runtime dependencies separately.
ENB Runtime Core is a C++23 embedded runtime library that identifies an already-loaded ENBSeries host, validates its SDK surface, queues callbacks outside callback context, coordinates save quiescence and reapplication, and gates a fail-closed Skyrim engine bridge.
Skyrim plugin linking game data to shaders and tools
active · prerelease v3.0.0-rc.1
SkyrimBridge is an SKSE plugin that exposes Skyrim live engine state, record editing surfaces, asset conversion, a versioned ABI, diagnostics, and shared-memory command channels to shaders and external tools, with an optional D3D11 rendering tier.
RAW is a dependency-free C++23 renderer that draws a scene on the CPU, checks its fast screen-space ambient occlusion against a ray-traced reference, and writes a JSON certificate saying whether the shortcut held.
Primary area: Developer infrastructure. Also appears here because it serves Graphics and media.
Telos is a local workbench for running and replaying AI work. Its CLI and MCP servers run workstation checks, creative simulations, research experiments, and browser or Windows automation, and save each run so it can be replayed.
Boundary: Run workstation actions and external effects only with the user's explicit approval and within each tool's own checks.
Limit: C is the production backend. SPIR-V, LLVM, WebAssembly, Rust, native-object backends, GPU compute and linear types are experimental.
Why this record appears here: BuildLang
Primary area: Developer infrastructure. Also appears here because it serves Graphics and media.
BuildLang is a systems programming language in which a program declares what it is allowed to touch. The compiler checks those declarations and its memory rules, then emits native code through C. It ships with a CLI, editor support, two-way C integration and experimental shader output.
Boundary: Generated programs inherit the authority and safety obligations of the capabilities they exercise.
Gather collects research material from web pages, including JavaScript-rendered ones, APIs, scholarly records, PDFs, scans, audio, video, feeds and local files.
Gather collects research material from web pages, including JavaScript-rendered ones, APIs, scholarly records, PDFs, scans, audio, video, feeds and local files. It saves each item with its source in a local corpus you can search and verify.
Other areas: Also appears in Developer infrastructure.
Boundary: Respect access controls, copyright, privacy, and the terms governing each source.
Learn plans study sessions from your objectives and recorded attempts, using spaced review, retrieval practice, prerequisite checks and misconception tracking.
Limit: Automation stops at graded work, consent, CAPTCHA, payment and account-creation steps.
How we know: Learn
Learn plans study sessions from your objectives and recorded attempts, using spaced review, retrieval practice, prerequisite checks and misconception tracking. It can also handle course logistics and stops at graded work.
Other areas: Also appears in Agent systems.
Boundary: The learner remains responsible for graded answers, consent, identity, and institutional rules.
active · v0.3.1 on PyPI as chorus-discourse; latest GitHub release v0.3.0
Chorus turns a captured comment thread or corpus into ranked, clustered theme digests, and checks whether a set of gathered sources has changed since a reference set.
Limit: Missing or ill-posed evidence fails closed.
Why this record appears here: Crucible
Primary area: Evaluation and verification. Also appears here because it serves Research and education.
Crucible is test infrastructure for claims about AI systems. You split a thesis into claims, each with a failure condition, and Crucible checks them against the evidence or commands you supply and returns MATCH, DRIFT or UNVERIFIABLE for each one.
Boundary: Report each verdict with the evidence and criterion that produced it.
Limit: Engine Revival handles metadata and build scaffolding. The BRender build and its test results live in BRender Archival.
Why this record appears here: Engine Revival
Primary area: Graphics and media. Also appears here because it serves Research and education.
Engine Revival is a Python tool for building, checking and publishing metadata about revived historical engines. It can also generate an out-of-tree BRender CMake build harness from a public checkout.
Boundary: Keep imported AGPL-covered BRender media separate from the MIT repository code.
Limit: Textured TIA output and x64 builds are not finished.
Why this record appears here: BRender Archival
Primary area: Graphics and media. Also appears here because it serves Research and education.
BRender Archival rebuilds the Argonaut BRender engine from its public MIT source, runs it through a 21-stage native test ladder, captures renders, and packages the build harness, test transcripts and checksums without copying the upstream source.
Boundary: BRender source stays upstream; this repository ships the harness and its results.
Limit: The judge, fix and polish editors send the draft to the model you choose. The checks run offline.
Why this record appears here: Articulate
Primary area: Evaluation and verification. Also appears here because it serves Research and education.
Articulate is prose analysis and model-assisted editing for the command line, editors and CI. It names each writing pattern it finds and the line it is on, in the terminal, in an editor through its language server, or as SARIF in code review.
Boundary: Run on prose you are authorized to review. It is not built for evading AI detection.
