Use Z1P in every coding session: one global MCP entry for Codex, Claude and Kimi, with host-verified packets for direct Ollama workers. Every request explicitly selects a checkout; source stays local to the tools.
Local code graphs and bounded evidence for coding agents.
Local graph viewer: export a private offline source graph, follow static relationships and inspect exact evidence. Rust/Kotlin syntax spans are available in exploration and packet planning. Installed usage accounting reconciles selected host receipts with an explicit task inventory.
Shared workspace: encrypted reusable indexes, resumable batch warm-up, and advisory agent activity and addressed notes across worktrees.
Local ticket dashboard: view agent activity, overlapping paths, context health and checkpoint summaries, with explicit Girnel card links. Read-only, loopback-only; no ticket writes or automatic acceptance.
Local task console: explicitly enable task entry, a managed Codex, Claude or Kimi host session, approvals/questions, result review and saved conversation continuation. An optional bounded Ollama worker lane uses an operator-configured helper profile, reviewed source dispatch/replacements and configured checks. Live worker availability requires an eligible model on that endpoint.
Bounded host workflows: optional local task contracts, dependency/review checks, deterministic result reduction and resumable planning. The coding host retains execution, permissions and model selection. The local dashboard's task plans connect dependencies, assigned hosts, check evidence and owner review to explicit managed runs. Versioned requirement changes preserve earlier work, flag old results and notify registered workers.
Use Context on your own machine and repositories: follow the Codex and Claude Code setup guide, including current source-build availability, worktrees and first-request verification.
Z1P Core is the free, open-source foundation of Z1P: signed, encrypted project evidence, bounded retrieval and authorised relationship graphs. It runs locally, requires no account and does not send source code or records to a hosted service.
The product goal is an independently built, MIT-licensed Graphify alternative for developers on their own projects, with less repeated discovery and smaller, sufficient model context. Feature parity and whole-task savings remain unproven. See the product direction for priorities and current gaps. ForgeSworn projects are early users; Oathrun is an optional consumer, not a required runtime.
The Node tools include deterministic, bounded TypeScript and JavaScript source analysis that emits reviewable file, declaration, import and call evidence into the same signed graph format. It runs locally without executing a repository or contacting a model.
The bounded source tools cover TypeScript/JavaScript through the compiler API, plus conservative lexical navigation for Python, Rust, Go, Java, Kotlin, Swift, C/C++, C#, Ruby and PHP. Every derived record can carry signed method, derivation and integer confidence metadata.
An explicit ecosystem manifest can also join package dependencies and Markdown design rationale across multiple repositories. The resulting append-safe graph uses the existing bounded query and path traversal; extraction remains local and deterministic, so routine scans consume no model tokens.
packages/context: browser-safe core, published as@forgesworn/context.packages/context-tools: Node persistence, CLI and MCP, published as@forgesworn/context-tools.
Both are MIT licensed. Provider routing, worker execution and coding-agent UI belong to consumers. Local graph exploration belongs in the MIT developer tool; the separate multi-repository ecosystem viewer remains a prototype. No model provider is required by core.
Local stdio MCP servers already provide repository navigation and separate signed context/graph tools. Claude Code and Codex are first-class target clients; actual Codex use and a first Claude read-only task are recorded; full Claude lifecycle and routine coding acceptance remain open. See the client and language plan for current capabilities, deeper Rust/TS/Kotlin work and the provider-authentication boundary.
This repository will remain the complete MIT local developer tool: formats, cryptography, extraction, verification, indexing and refresh, bounded graph operations, local exploration, CLI and MCP tools. Some of that scope remains planned. The commercial offering can provide managed shared graphs, repository connections, company administration, deployment and support around that engine.
The free core is intended to be useful on its own, not a time-limited trial. See OPEN_CORE.md for the durable product boundary and compatibility policy.
See GOALS.md for the core-first release gates, required evidence and whole-task inference-cost evaluation. See the execution plan for ordered work, owners, locations and model/effort assignments. Enterprise development comes after a working open-source release with demonstrated benefits. The current candidate results and blockers are recorded in RELEASE_EVIDENCE.md.
To try a disposable local scan → signed cache → MCP retrieval workflow from this checkout, see the dogfood walkthrough. It reports scan omissions and checks restart persistence; it does not measure inference savings.
For the remaining work to use this across ForgeSworn, see the dogfooding goals and model assignments. The immediate path uses the local MCP bridge while measuring complete tasks. For coding handoffs from this checkout, the worker packet helper assembles bounded source excerpts and rejects stale packets before reuse.
For repository navigation beyond the signed collection's 128-record limit, see local repository navigation: a separate unsigned, in-memory MCP index with explicit refresh, larger response budgets and pagination.
The formats and APIs are project-agnostic. A collection can describe one repository or an explicitly assembled ecosystem; graph operations never make another collection visible or turn an extracted relationship into authority.
Use Node 24 LTS. The repository pins the currently validated release in .nvmrc:
npm ci --ignore-scripts
npm run build
npm test
npm run test:packages
npm run benchmark:tokens:check
npm run benchmark:tokens:parityThe measured token-reduction benchmarks keep complete raw-evidence retrieval separate from compact source navigation. Both count exact bounded payloads and require all predeclared sources. The navigation gate tracks Graphify's published 71.5x figure without treating different corpora or undisclosed tokenisation methodology as directly comparable or claiming that a source pointer contains enough code to answer.
The smoke test builds real tarballs and installs them outside the workspace to check exports, browser isolation, CLI/MCP behaviour and licence notices. See extraction evidence for provenance and publication status.
The optional local task console supports Codex, Claude, Kimi and bounded Ollama workers, with isolated task worktrees, concurrent runs, a shared addressed inbox and a review queue. Native worker notes use existing repository tools; Ollama receives queued notes on explicit continuation. Host permissions and model accounts remain operator-owned.