Discover the Best AI Tools

Pi Jev Helm is a public-preview Pi extension that classifies new work with Jev and temporarily selects a configured model. Automatic routing sends the current message to a paid provider; failures retain the baseline model, Windows is unverified, and proposed safety/verifier features are not implemented.

OMP TypeSafe adds advisory or adversarial Jev review, typed questions and optional context pruning to the omp agent. Review notes do not block tools, while separate configured gates can intervene. It requires credentials and sends selected transcript and diff content to TypeSafe.

Jev Flash Router exposes an MCP decision tool for binary, categorical and rubric judgments over supplied context through OpenRouter. It requires an OpenRouter key with Jev access; no generated output text does not mean that input inference is free or that judgments are certain.

Ask Jev provides explicit advisory audits and binary questions over recorded local Codex conversations through Windows PowerShell hooks. Requested audits send selected transcript content to paid TypeSafe inference; evidence is bounded and probabilities are not guarantees or automatic fixes.

Limpet is a Python stop hook for Claude Code and Codex that checks the stopping response against user-written rules with Jev. It needs provider credentials and can push back once per stop chain; judgments can miss problems and do not verify unseen execution results.

TypeSafe Code Guard is a Python MCP server that uses Jev to rank candidate files and code blocks and flag potential issues in diffs. It requires a TypeSafe key and sends selected code context to the service; rankings and flags are advisory, not proof of correctness.

Stepwarden is an experimental Claude Code hook that compares pending tool calls with task intent using Jev. It starts in audit-only mode; enforcement must be enabled explicitly, and early-access hooks and uncalibrated thresholds are not deterministic security controls.

Typeful Triage is a collaborative dashboard that syncs public GitHub issues and pull requests, uses Jev to suggest classifications and next actions, and retains human corrections. It requires its database stack and TypeSafe credentials; queue actions do not write back to GitHub.

FTP Deploy MCP lets agents browse, transfer and deploy files on FTP, FTPS or SFTP servers with dry runs and configured local-root limits. Source installation and server credentials are required; the npm package is not yet published, and secure transport settings need configuration.

Text File Read and Refactor MCP searches, reads and edits text by lines, slices or boundary patterns while preserving encoding and BOM. It is a local Python stdio server; the public repository receives selected development snapshots.

Skillful routes coding prompts to installed skills, MCP servers or commands using a local shortlist and TypeSafe Jev, injecting a small suggestion set through agent hooks. Prompts are uploaded by default. Retrieval targets and task-outcome benefits remain unproven; failures fall back to a reminder without blocking the agent.

jev-scout is a Rust CLI and MCP server that retrieves GitHub and crates.io candidates and uses TypeSafe Jev to rank their relevance and maintenance signals. It requires a TypeSafe key and network access; registry metadata and model scores do not guarantee installability, licensing suitability or absence of errors.

jev-git is a Rust Git extension that scores staged or piped diffs for secrets and harmful payloads through TypeSafe Jev. It needs an API key, sends diff content externally and truncates large inputs. Its pre-push wiring and chained-hook exit handling need review; it is not a comprehensive security gate.

jear is an early Rust client that routes requests toward NEAR AI Cloud models or agent plans using budget, quality and sensitivity policies with optional Jev judgments. Its default demo is offline; live inference needs provider keys and Jev endpoint configuration. The multi-tenant proxy is a design, not an implemented service.

decision-first combines an agent skill with Python runners for testing typed Jev decisions, comparing labels, and documenting reusable experiments. It requires Python and a TypeSafe key for live tests, sends supplied state to the API, and relies on human evaluation before adopting an integration.

typesafe-ai-firewall is a runnable shadow-evaluation harness for AI tool-call risk decisions, with Jev hazard scores, deterministic policy and offline log replay. It requires a TypeSafe key for live scoring and does not enforce production tool execution. Synthetic test success does not establish real-world protection; calibration and latency targets remain unmet.

Ailerix is a model-routing prototype with an OpenAI-shaped completion API: Jev classifies task families and code selects from a cost/quality catalog. Development uses a fixed benchmark snapshot and local or mocked responses until service keys are configured. It is not evidence of a fully operational hosted routing service.

jev-skill-gate scores installed Claude Code skills against project or prompt context and writes visibility overrides to reduce the skill manifest. It supports remote Jev scoring and local fallback; thresholds may hide relevant skills, and published savings or small evaluation results are not general guarantees.

This Pi extension uses TypeSafe Jev to prune old tool calls and results without rewriting ordinary conversation or the session file. It offers background scoring and conservative preservation of side-effecting calls. Context is sent to an external API; errors retain unmodified messages, and pruning can lose useful evidence or increase cache costs.

MCP File Tools lets AI clients read, search and edit legacy-encoded text while preserving original encoding, BOM and line endings. Access is limited to configured directories. Encoding detection is heuristic and exposes confidence; users should check ambiguous files before rewriting them.

hledit-mcp exposes hash-anchored file reads and edits to MCP coding clients, rejecting edits when referenced lines have changed. It requires the separate hledit CLI and Node.js; a stale result requires rereading the file, and workspace scope must be configured.

Agent Coherence coordinates shared state and files across AI agents using versioned reads and conditional writes, with framework adapters and MCP access. Cache coherence alone does not prevent stale writes: agents must use the guarded write interfaces. External side effects remain outside this protocol.

Typesafe Migration Guard sends SQL to Jev and returns a CI-facing safety verdict, with 403 for DANGER when the supplied environment is production. It needs explicit pipeline integration and trusts the caller’s environment label. The model judgment is probabilistic, not database enforcement; a SAFE response does not prove a migration harmless.

Jev Workbench defines, tests and publishes versioned Jev judgment functions through a local UI, HTTP and MCP with scoped client tokens. Production requires provider credentials and never substitutes simulated answers on failure. Demo answers are simulated; runtime integrations and business accuracy need separate validation.