Packages

codex_sdk

0.4.1
0.17.0 0.16.1 0.16.0 0.14.0 0.13.0 0.12.0 0.11.0 0.10.1 0.10.0 0.9.0 0.8.0 0.7.2 0.7.1 0.7.0 0.6.0 0.5.0 0.4.5 0.4.4 0.4.3 0.4.2 0.4.1 0.4.0 0.3.0 0.2.5 0.2.4 0.2.3 0.2.2 0.2.1 0.2.0 0.1.0

Idiomatic Elixir SDK for OpenAI's Codex agent. Provides a complete, production-ready interface with streaming support, comprehensive event handling, and robust testing utilities.

HexDocs

Current section

Files

Jump to
Readme
30 Versions
6 Dependencies
Dependants
Files
Activity
codex_sdk docs 20251202 python-parity 002-config-and-flags.md
Raw

docs/20251202/python-parity/002-config-and-flags.md 

Python

- `src/agents/run.py:179-267` `RunConfig` controls model override, provider, model_settings overlay, handoff history nesting/mapping, guardrails, tracing toggles (`tracing_disabled`, `trace_include_sensitive_data`), workflow/group metadata, custom trace ids, session handling callbacks, and `call_model_input_filter` to mutate instructions/input before model invocation.

- `src/agents/run.py:106-110` defaults `trace_include_sensitive_data` from `OPENAI_AGENTS_TRACE_INCLUDE_SENSITIVE_DATA` env (true unless explicitly falsey).

- `src/agents/_config.py:8-26` and `src/agents/__init__.py:172-200` expose setters for default OpenAI key/client and default API flavor (chat_completions vs responses), optionally reused for tracing exports.

- `src/agents/models/default_models.py:9-58` picks default model from `OPENAI_DEFAULT_MODEL` (fallback `gpt-4.1`), and adapts default `ModelSettings` for GPT-5 reasoning requirements.

- `src/agents/model_settings.py:69-190` provides per-call knobs: temperature, top_p, penalties, tool_choice/parallel_tool_calls, truncation, max_tokens, reasoning effort, verbosity, metadata, store flag, prompt cache retention, include_usage, response_include fields, top_logprobs, and extra headers/body/query/args merging via `resolve`.

- `src/agents/models/openai_provider.py:28-97` allows provider-level api_key/base_url/organization/project overrides or custom AsyncOpenAI client; switches between Responses and Chat Completions based on `use_responses` default or override.



Elixir status

- `lib/codex/options.ex:11-179` offers `api_key`, `base_url`, `codex_path_override`, telemetry prefix, model, and reasoning_effort defaults; derives key from env or codex CLI credentials. No per-call model settings beyond model/reasoning_effort.

- No equivalent env-driven tracing toggles; telemetry uses `Codex.Telemetry` with prefix configuration only.

- No provider abstraction; always shells out to codex binary using configured base_url/model.

- No hook to filter or mutate model payloads before execution.



Gaps/deltas

- Missing fine-grained model settings (temperature/top_p/penalties/tool_choice/parallel calls/truncation/prompt cache/response_include).

- No configuration for trace metadata, workflow/group IDs, sensitive-data inclusion, or disabling tracing per run.

- Lacks run-level mutation hooks (`call_model_input_filter`) and session input callbacks.

- Provider selection (chat_completions vs responses) and default model env parity absent.



Porting steps + test plan

- Extend `Codex.Options`/`Codex.Thread.Options` with a struct mirroring `ModelSettings` fields, mapping to codex CLI flags or request body; add validation similar to `model_settings.py` and tests akin to `tests/test_model_payload_iterators.py`.

- Add run-level config struct with trace metadata fields and a hook to pre-process payloads (Elixir function accepting instructions/input), validating behavior with parity cases from `tests/test_call_model_input_filter.py`.

- Introduce env toggles for tracing sensitive data and default model selection (`OPENAI_AGENTS_TRACE_INCLUDE_SENSITIVE_DATA`, `OPENAI_DEFAULT_MODEL`) with unit coverage.

- Add provider-style overrides for base URL/api key per run; ensure compatibility with codex exec args and cover via option tests similar to `tests/test_config.py` and `tests/test_run_config.py`.***