intertwine

Discipline for developing Hive itself — PR sizing, review handling, merge discipline, release workflow, delegation, and cleanup hygiene. Use this skill when working on Hive repo changes that span multiple PRs or review cycles.

Coordinate work across multiple agents and projects in Hive. Covers task claims, blockers, handoffs, campaigns, portfolio management, briefs, and shared memory.

Hive mental model and orientation. Read this first before using any other Hive skill. Covers the entity hierarchy, observe-and-steer pattern, drivers, sandboxes, console vs CLI, and workspace conventions.

Use the Hive MCP server for the thin v2 search and execute tool surface. Use this skill when configuring MCP access to Hive or when an agent needs workspace search and bounded local execution.

Set up and configure Hive workspaces, projects, tasks, and evaluator policy. Use this skill when bootstrapping, creating projects, managing tasks, or configuring PROGRAM.md.

The core agent work cycle in Hive — from finding a task through claiming, launching a run, handling approvals, finishing, and promoting. Use this skill for task-first project work, governed runs, and clean handoff.

Run and analyze Security Verifiers evaluations. Use when asked to evaluate models on E1 (network-logs) or E2 (config-verification), generate metrics reports, compare model performance, or analyze eval results.

Generate SV-Bench metrics reports (summary.json + report.md) for E1/E2 runs, validate metrics contracts, and produce comparison-friendly artifacts from outputs/evals/.

Deploy Security Verifiers environments and packages. Use when asked to deploy to Prime Intellect Environments Hub, publish to PyPI, bump versions, build wheels, or manage releases.

Build and manage Security Verifiers datasets. Use when asked to build E1 or E2 datasets, create test fixtures, validate data, or manage dataset files for network-logs or config-verification environments.

Development workflow for Security Verifiers. Use when asked to run tests, lint code, format files, set up the development environment, or perform CI checks on the codebase.

Manage HuggingFace datasets for Security Verifiers. Use when asked to push datasets to HuggingFace, manage metadata, configure gated access, or set up user HF repositories for E1/E2 datasets.

Drive a complete DSPy 3.2.x project end-to-end — spec → program → metric → baseline → GEPA optimize → export → deploy. Orchestrates the other four DSPy skills (dspy-fundamentals, dspy-evaluation-harness, dspy-gepa-optimizer, dspy-rlm-module) in the correct order. Use this for any non-trivial DSPy build from scratch.

Build DSPy evaluation harnesses with rich-feedback metrics that are essential for GEPA optimization. Use when writing a metric function, calling dspy.Evaluate, splitting dev/val sets, debugging "why is my optimizer not improving?", or designing CI-ready DSPy eval suites.

Write idiomatic DSPy 3.2.x programs — typed Signatures, dspy.Module subclasses, Predict/ChainOfThought/ReAct/ProgramOfThought, and save/load. Use this when starting any new DSPy project or when fixing non-idiomatic DSPy code (hard-coded prompts, ad-hoc string templates, untyped outputs, non-serializable classes).

Optimize DSPy programs with dspy.GEPA — the reflective/evolutionary optimizer that is the 2026 gold standard for DSPy (beats MIPROv2 on complex tasks with far fewer rollouts when the metric returns rich feedback). Use when the user says optimize, compile, GEPA, reflective optimization, or "make this program better" and a DSPy program + metric + trainset exist.

Use dspy.RLM (Recursive Language Model) for reasoning over contexts too large to fit in an LLM's working window — entire codebases, long logs, massive documents, or multi-step data exploration that needs a sandboxed Python REPL. Use when the input is >100k tokens, needs recursive chunking, or benefits from the LLM writing and running code to probe data.

Cross-session observational memory system. Use when the user asks to "recall something from a past session", "search memory", "what do you know about me", "remember when we discussed", or needs context from previous conversations. Also triggers when the user mentions "observational memory", "OM", or asks about cross-session context.

Convert software repositories from plaintext dotenv files to SOPS + age encrypted dotenv files. Use when asked to replace .env files with encrypted secrets, migrate a repo to sops/age, update Node/Python/Ruby/Go/Docker/Make dev and test commands to decrypt at runtime, add agent-facing docs such as AGENTS.md or CLAUDE.md, or verify that a repo no longer depends on long-lived plaintext .env files.

Two-tier compressed memory system using Observer and Reflector background agents. Replaces raw conversation history with dense, prioritized observations and long-term reflections.

Create, manage, and run scheduled Claude Code tasks. Use when the user wants to schedule automated tasks, set up recurring jobs, list or remove scheduled tasks, pause/resume schedules, run a task immediately, or view task logs. Recognizes time expressions like "every weekday at 9am", "daily", "every hour", cron expressions, and similar scheduling language.

Cross-session observational memory for Grok Build. Use when the user asks to "recall something from a past session", "search memory", "what do you know about me", "remember when we discussed", or needs context from previous conversations. Also triggers when the user mentions "observational memory", "OM", or asks about cross-session context.