Back to Agno

Environments

cookbook/environments/README.md

2.8.06.6 KB
Original Source

Environments

Verification and dataset generation for agents. 28 progressive folders contain 79 single-file runnable examples: run an agent K times against difficult tasks, score every attempt, inspect the pass-rate grid, and export passing text trajectories as a supervised fine-tuning dataset.

Each subfolder covers one theme. Its basic.py is the smallest complete example; variants add one task-meaningful option at a time.

The central signal is the learning zone: tasks with 0 < pass_rate < 1. Tasks that always pass are already saturated, while tasks that always fail provide no successful trajectory to export. The useful middle band shows where the policy is capable but inconsistent. The examples use tasks calibrated against gpt-5.5; an all-full grid is a prompt to make the task harder, not a successful demonstration.

This release performs independent rollouts and scores them after completion. It does not run a live RL reward loop, and exporting JSONL does not train a model. A live turn-by-turn environment is a later release.

Start with _01_first_environment/basic.py. Every other cookbook mirrors its structure and builds on the vocabulary introduced there.

Layout

cookbook/environments/
├── README.md
├── <theme>/
│   ├── README.md
│   ├── basic.py            # smallest readable example
│   ├── <variant>.py        # one file per task-meaningful option
│   ├── schemas.py          # shared Pydantic types, if any
│   ├── data/               # checked-in tasks; generated/ is ignored
│   └── TEST_LOG.md         # observed live pass rates for every file
└── ...

Cookbooks

Verification basics

  • _01_first_environment/: create an Environment, run K isolated attempts, and read the grid and summary().
  • _02_task_sets/: declare tasks inline, load strict JSONL, and select metadata-defined slices without changing environment identity.
  • _03_code_scorer/: verify typed outputs with Boolean, graded, and explicit Score results.
  • _04_judge_scorer/: grade criteria that code cannot express with binary and numeric rubrics.
  • _05_tool_call_scorer/: require clean tool executions, exact arguments, and no unexpected tools.
  • _06_learning_zone/: surface the partial pass-rate band and separate it from saturated and failed tasks.
  • _07_difficulty_calibration/: grow task difficulty until a strong model stops producing a wall of full bars.
  • _08_async_rollouts/: use arun_rollouts and the async SFT exporter inside an existing event loop.
  • _09_task_selection/: run a proven subset and rerun only tasks that need more evidence.

Dataset export

  • _10_export_sft/: select learnable tasks, keep passing attempts, and write portable conversational JSONL.
  • _11_export_provenance/: inspect the score and fingerprint sidecar that keeps training rows auditable.
  • _12_trainer_loader/: validate and stream exported messages through a small trainer-facing loader without pretending training occurred.

Comparing runs

  • _13_saved_baselines/: save, reload, and protect plaintext rollout evidence for later comparison.
  • _14_environment_diff/: diff identical environments under different gpt-5.5 policy settings and handle fingerprint mismatches.
  • _15_prompt_comparison/: compare before/after prompt summaries when the environment fingerprint changes by design.
  • _16_policy_settings/: compare low and high reasoning effort while keeping the model family fixed.

Reliability and evidence

  • _17_tool_reliability/: measure tool grounding over a distribution and compare repeated ReliabilityEval verdicts with the scorer.
  • _18_execution_matching/: distinguish clean executions from requested, failed, or wrong-argument calls.
  • _19_error_analysis/: inspect unscored attempts, scorer errors, and public StopReason values without folding them into failures.
  • _20_report_drilldown/: move from the grid to failed-only reports and a single attempt's full transcript.

Task domains

  • _21_math/: exact arithmetic ladders whose difficulty grows past single-operation saturation.
  • _22_sql_generation/: execute generated SQL against in-memory fixtures, including joins and window functions.
  • _23_code_fixes/: verify constrained bug fixes against explicit regression cases.
  • _24_structured_extraction/: score typed extraction when dates, fields, and nested records conflict.
  • _25_support_triage/: apply precedence rules to genuinely multi-intent support tickets.
  • _26_multi_step_tools/: verify required tool chains, arguments, and execution order.

From evidence to a gate

  • _27_verified_dataset/: run, curate the middle band, export passing text attempts, and inspect the resulting manifest end to end.
  • _28_ci_gating/: turn summary() and per-task floors into a process exit decision suitable for CI.

Running a cookbook

From the Agno repository root, create the demo environment if needed:

bash
./scripts/demo_setup.sh

Load the repository environment and run the first file:

bash
direnv exec . .venvs/demo/bin/python cookbook/environments/_01_first_environment/basic.py

Every runnable file uses OpenAIResponses with gpt-5.5. Folder READMEs list all commands and call out any local fixture they use.

VariableUsed by
OPENAI_API_KEYEvery environment cookbook

Reading “learning zone” precisely

For Boolean scores, results.learning_zone() and 0 < pass_rate < 1 select the same tasks. Numeric scorers can vary in score while every attempt remains on the same side of the pass threshold; those examples call that score variation, not a partial pass-rate learning zone. SFT examples use Boolean verdicts before exporting.

Tool-using rollouts can be verified but are not exportable with the current text-only SFT format. The exporter skips them rather than dropping the tool evidence and teaching the model to answer without its tools.