{ "markdown": "# Syntavra 0.0.1 — Pre-Release Token & Context Optimization Skill\n\nSyntavra is a local-first **Agent Skill and runtime middleware** designed to reduce token and context overhead in existing AI coding agents. It does not replace Codex, Claude Code, Cursor, OpenCode or another agent/model.\n\nSyntavra targets five cost surfaces while preserving exact recovery:\n\n| Surface | What Syntavra does |\n|---|---|\n| Repository context | Retrieves exact definitions, impact paths, tests and verifiers before broad file reads |\n| Tool output | Stores full stdout/stderr as exact evidence and returns bounded task-relevant views |\n| MCP schemas | Exposes a small profile and deterministically compiles verbose discovery schemas |\n| Session memory | Keeps active context bounded while exact external history remains searchable and recoverable |\n| Measurement | Separates provider-observed usage from locally tokenized or estimated attribution |\n\n> The only active product identity is **0.0.1 / pre-release**. The owner explicitly authorizes any version or release-channel change. External superiority, live certification, long-context quality, adoption and production maturity remain evidence-gated.\n\n## Install\n\nThe registry package is prepared but not published yet:\n\n```bash\nnpx @syntavra/install\n```\n\nUntil registry publication, use the repository installer:\n\n```bash\nnpx github:Naveax/Syntavra\n```\n\nThe installer prefers checksum-verified portable binaries, falls back to Python 3.11+, configures detected hosts transactionally and runs a final health check.\n\n## Engine selection\n\nSyntavra is migrating to one dual-engine bundle containing the Python runtime, native Rust runtime and a native selector. Engine preference is explicit and persisted without hidden fallback:\n\n```bash\nsyntavra engine status\nsyntavra engine use python --scope project\nsyntavra engine use rust --scope project\nsyntavra engine use auto --scope project\n\nsyntavra --engine python version\nsyntavra --engine rust version\n```\n\nSelection precedence is command override, `SYNTAVRA_ENGINE`, project configuration, user configuration and then the explicit `auto` policy. Selecting `python` never invokes Rust. Selecting `rust` never invokes Python; a command without an independent native Rust handler fails closed.\n\nThe canonical installed Python surface currently contains 245 executable public command paths. Full dual-engine parity inventory is tracked against those 245 paths in `contracts/engine/dual-engine-public-surface-v2.json`; bounded R23–R37 contract parity is not treated as a complete native product rewrite. The canonical route-coverage claim is:\n\n```text\nFULL_DUAL_ENGINE_PARITY_PROVEN\n```\n\nThis route-level parity claim does not reactivate Rust feature/parity development and does not grant production-promotion credit. Rust development/resume authority and the 174/245 production-promotion boundary are governed separately.\n\n## Product surface\n\n```bash\nsyntavra setup # plan installation\nsyntavra setup --apply # install and configure detected hosts\nsyntavra setup --repair --apply\nsyntavra status # health, profile, evidence and observed savings\nsyntavra status --savings\nsyntavra status --doctor\nsyntavra run manifest\nsyntavra prove plan\n```\n\nNormal daily work still happens in the existing coding agent. Syntavra is intended to activate through its skill, MCP integration and host hooks rather than requiring every command to be prefixed manually.\n\nFor Codex, the managed MCP entry starts through the same Syntavra launcher using the internal `codex-mcp-bridge` route. User/global installs begin repository-unbound and require `syntavra.project.bind` before repository or process tools can run; trusted project-scope installs may auto-bind only to their exact project.\n\n## MCP profiles\n\n| Profile | Purpose | Maximum public surface |\n|---|---|---:|\n| `minimal` | Default hot-loop token saver | 10 tools |\n| `balanced` | Repository context, output, memory and provider controls | 36 tools |\n| `audit` | Full inspection and administration | Entire installed catalog |\n\n`tiny`, `optimized` and `full` remain compatibility aliases only. The installed profile, listed tools, callable tools and benchmark profile are derived from the same canonical registry.\n\n## Implemented surfaces and evidence boundary\n\nThe repository contains implementations for command rewriting and output compaction, structural repository indexing, bounded session memory, host/MCP integration, provider-usage receipts, and optional local UI or editor integrations.\n\nImplementation presence is not evidence of real-world savings, integration certification, adoption, reliability, or superiority. Registry counts are treated as internal inventory only and are intentionally omitted from the public overview because they can drift and do not measure task success.\n\n```bash\nsyntavra run rewrite -- git status\nsyntavra run code-intel report\nsyntavra run memory-search \"cache decision\"\nsyntavra prove plan\n```\n\nEvidence rules are defined in `docs/CLAIMS_POLICY.md`; the detailed implementation inventory is in `docs/IMPLEMENTATION_STATUS_001.md`.\n\n## Exact-recovery rule\n\nSyntavra does not treat deletion as compression:\n\n```text\nfull repository/tool/session evidence\n -> content-addressed local artifact\n -> bounded task-relevant view\n -> recovery handle\n -> exact reveal/verification when required\n```\n\nCorrectness, verifier success and security boundaries outrank nominal token reduction.\n\n## Measurement rule\n\nThe primary optimization metric is:\n\n> **provider-observed cost per verified successful task**\n\nRaw byte reduction is reported separately. Source-level attribution identifies schema, repository, tool-output, memory, conversation and output tokens with an explicit confidence level:\n\n- `PROVIDER_OBSERVED`\n- `LOCALLY_TOKENIZED`\n- `ESTIMATED`\n- `UNKNOWN`\n\nSynthetic fixtures and internal component tests cannot open public superiority claims.\n\n## Primary certification targets\n\nThe first live-certification targets are:\n\n1. Codex\n2. Claude Code\n3. Cursor\n\nOther adapters remain contract-tested or declared bridges until external execution receipts prove live behavior.\n\n## Benchmarking\n\nSignalBench compares independent external arms under the same frozen repository, prompt, model, reasoning mode, context window, verifier, permissions, timeout, cache policy and hardware class. Templates include plain host, Caveman, RTK, Token Savior, repository-context tools, a combined competitor pack and Syntavra minimal/balanced arms.\n\nMissing competitors, provider usage or verifier output fail closed; they are never replaced with synthetic results.\n\n## Canonical documentation\n\n- `docs/001_PRE_RELEASE.md`\n- `docs/ARCHITECTURE.md`\n- `docs/TOKEN_SAVER_PLAN_001.md`\n- `docs/COMPETITIVE_GAP_CLOSURE_001.md`\n- `docs/SECURITY_MODEL.md`\n- `docs/ADAPTER_PLATFORM.md`\n- `docs/SIGNALBENCH.md`\n- `docs/OPERATIONS.md`\n- `docs/CLAIMS_POLICY.md`\n- `docs/IMPLEMENTATION_STATUS_001.md`\n\n## Current claim boundary\n\n```text\nFULL_DUAL_ENGINE_PARITY_PROVEN\nEXTERNAL_SUPERIORITY_NOT_PROVEN\nLONG_CONTEXT_QUALITY_NOT_PROVEN\nMEASURED_AGENT_BENCHMARK_NOT_PROVEN\nLIVE_INTEGRATION_CERTIFICATION_NOT_PROVEN\nDAILY_CODING_AGENT_READINESS_NOT_PROVEN\nPUBLIC_PRODUCT_MATURITY_NOT_PROVEN\n```\n\n\n", "bytes": 7668, "sha": "84bf438e2e416766d5df9e3f4b02ac8fc4622cae1c8bdd927f22d39a47505d7a", "repo_slug": "naveax/syntavra", "fonte": "repo", "truncated": false, "api": "https://agentalog.com/api/listings/plg_naveax_syntavra_6ff49fbe/readme" }