OpenProse Standard Library

Reusable programs, patterns, and roles for evaluation, operations, and memory management in OpenProse.

Usage

use "std/evals/inspector"
use "std/ops/profiler"
use "std/memory/project-memory"
use "std/composites/worker-critic"
use "std/controls/pipeline"
use "std/roles/critic"

Install with prose install. See deps.md for details.

Programs

evals/

Programs for evaluating and improving prose runs.

Program	Purpose
inspector	Post-run analysis — runtime fidelity and task effectiveness
calibrator	Validates light evals against deep evals for reliability
vm-improver	Analyzes inspections, proposes VM improvements
program-improver	Analyzes inspections, proposes program improvements

ops/

Programs for profiling and debugging.

Program	Purpose
profiler	Cost, token usage, and time profiling for completed runs
cost-analyzer	Token usage and cost pattern analysis across runs
error-forensics	Root cause analysis for failed runs
lint	Validate structure, schema, shapes, and contract consistency for a program and its service tree
preflight	Check that all runtime dependencies are satisfied before executing a program
status	Summarize recent runs from .prose/runs/
wire	Run Forme wiring to produce an execution manifest

delivery/

Services for delivering program outputs to humans and external systems.

Service	Purpose
human-gate	Present output for human review, block until approved or rejected
slack-notifier	Format and deliver content to Slack via webhook or API
email-notifier	Format and deliver content via email
dashboard-builder	Render an HTML dashboard from a template and structured data

composites/

Named multi-agent topology patterns for RLM programs.

Pattern	Purpose
observer-actor-arbiter	Three-role OAA composite: observer tracks state, actor proposes actions, arbiter decides
ensemble-synthesizer	N-agent ensemble that independently solves then synthesizes into a consensus answer
worker-critic	Two-agent loop: worker produces, critic evaluates, repeat until accepted
ratchet	Iterative improvement loop with regression prevention
witness	Meta-observer composite for recording and reflecting on agent behavior
dialectic	Two-agent debate structure producing a synthesized conclusion
proposer-adversary	Adversarial proposal testing: proposer generates, adversary stress-tests
assumption-miner	Surfaces implicit assumptions in a solution for explicit evaluation
blind-review	Independent evaluation without exposure to other agents' assessments
contrastive-probe	Generates contrasting solutions to expose decision boundaries
drift-detector	Monitors agent behavior over iterations to detect drift from objectives
stochastic-probe	Introduces controlled randomness to test solution robustness
synchronization-probe	Tests coordination quality between collaborating agents

controls/

Flow control patterns for structuring RLM program execution.

Pattern	Purpose
pipeline	Sequential stage pipeline: output of each stage feeds the next
map-reduce	Parallel fan-out across inputs followed by aggregation
gate	Conditional branching: proceed only if a quality or validity threshold is met
progressive-refinement	Iterative solution improvement with expanding scope each round
retry-with-learning	Retry loop that accumulates failure context to guide subsequent attempts

roles/

Single-agent role guides for use in RLM programs and composites.

Role	Purpose
classifier	Classifies inputs into a fixed set of categories; returns structured label
critic	Evaluates a proposed solution against criteria; returns scored feedback
extractor	Extracts structured data from unstructured text; returns JSON
summarizer	Condenses long content into a concise representation
verifier	Checks a solution for correctness against known constraints or test cases

memory/

Programs for persistent agent memory management.

Program	Purpose
user-memory	Cross-project memory (user-scoped)
project-memory	Project-scoped memory