razroo

@agent-pattern-labs/actionproof

Action-bound, signed proof challenges for verifying fresh autonomous agent work.

@agent-pattern-labs/agent-harness-fixtures

Reusable fixture projects and eval suites for ISO-based agent harness tooling.

@agent-pattern-labs/agentmd

Agent prompts as structured markdown, with a linter for structure and a harness that scores per-rule adherence — so iteration produces numbers, not vibes.

@agent-pattern-labs/agent-workbench

Local workbench for discovering and inspecting ISO-based agent harness projects.

@agent-pattern-labs/create-agent-harness

Scaffold portable ISO-based agent harness projects.

@agent-pattern-labs/iso

One command that chains agentmd → isolint → iso-harness: author one prompt, ship it to every coding agent.

@agent-pattern-labs/iso-cache

Deterministic content-addressed artifact cache for AI-agent workflows: key, store, retrieve, verify, and prune local artifacts without model calls.

@agent-pattern-labs/iso-canon

Deterministic canonicalization and entity matching for AI-agent workflows: normalize URLs, companies, roles, and stable dedupe keys without model calls.

@agent-pattern-labs/iso-capabilities

Deterministic role capability policies for AI-agent workflows: resolve, check, and render tool/MCP/filesystem permissions without model calls.

@agent-pattern-labs/iso-context

Deterministic context bundle planning for AI-agent workflows: select files, estimate tokens, check budgets, and render prompt context without model calls.

@agent-pattern-labs/iso-contract

Deterministic artifact contracts for AI-agent workflows: validate, parse, and render structured records without model calls.

@agent-pattern-labs/iso-eval

Behavioral eval runner for AI coding agents — snapshot a workspace, hand it to a runner with a task prompt, score the resulting filesystem/git state.

@agent-pattern-labs/iso-facts

Deterministic fact materialization for AI-agent workflows: extract, query, verify, and check provenance-rich facts from local artifacts without model calls.

@agent-pattern-labs/iso-guard

Deterministic policy checks for AI-agent workflows: enforce orchestration invariants against traces and event logs without adding prompt tokens.

@agent-pattern-labs/iso-harness

One config for every coding agent — Cursor, Claude Code, Codex, OpenCode, Pi.

@agent-pattern-labs/iso-index

Deterministic local artifact index for AI-agent workflows: build, query, and verify where authoritative facts live without model calls.

@agent-pattern-labs/iso-ledger

Deterministic append-only event ledger for AI-agent workflows: idempotent writes, local queries, verification, and materialized state views.

@agent-pattern-labs/iso-lineage

Deterministic artifact lineage tracking for AI-agent workflows: record dependencies, verify graph hashes, and detect stale outputs without model calls.

@agent-pattern-labs/isolint

Lint and rewrite AI harness markdown so weak small models (Minimax 2.5, Nemotron 3, Mistral 7B) can execute it reliably.

@agent-pattern-labs/iso-migrate

Deterministic project migrations for AI-agent workflow packages: plan, apply, check, and explain JSON/text file upgrades without model calls.

@agent-pattern-labs/iso-orchestrator

Durable workflow primitives for AI-agent harnesses: resumable steps, keyed mutexes, and bounded parallel fan-out with file-backed state.

@agent-pattern-labs/iso-postflight

Deterministic postflight settlement for AI-agent workflows: reconcile dispatch plans, observed outcomes, artifacts, and post-steps without model calls.

@agent-pattern-labs/iso-preflight

Deterministic preflight planning for AI-agent workflows: validate file-backed facts, apply gates, and produce bounded dispatch rounds without model calls.

@agent-pattern-labs/iso-prioritize

Deterministic policy-based queue prioritization for AI-agent workflows: rank, gate, quota, and select local work items without model calls.

@agent-pattern-labs/iso-receipts

Portable signed-work receipt artifacts for AI-agent workflows: capture, pack, verify, inspect, and redact local evidence bundles.