The OpenAI-compatible runner
A second agent runner drives any OpenAI-compatible chat endpoint (POST <base>/chat/completions)
in a tool-use loop over the same six tools Claude Code has: Bash, Read, Edit, Write, Glob, Grep.
The default is unchanged: without the switch below every run uses Claude Code.
Switch it on, in the forge's environment:
FORGE_AGENT_RUNNER=openaichooses the runner (any other value keeps Claude).OPENAI_BASE_URLis the endpoint's base,http://orhttps://, for examplehttps://api.openai.com/v1.OPENAI_MODELnames the model, unless the run names one.OPENAI_API_KEYis sent as a bearer token; leave it unset for an endpoint that needs none.
Without a base URL and a model a run ends as an error saying so. Read it as the run's result.
What it does: the file tools resolve every path inside the run's clone, following symlinks, and refuse
anything outside. Bash runs under bubblewrap with the run's sandbox policy and needs an enforced
sandbox: where the host cannot enforce one, Bash answers with a refusal and the model uses the file
tools. The shell sees only PATH, HOME, TMPDIR and the run's env less anything named like a secret
(OPENAI_*, FORGE_*, KEY, TOKEN, SECRET, PASSWORD). Bounds: an endpoint answer over 4 MB
ends the run, a turn runs at most 16 tool calls, Bash is killed at 120 s and its output cut at 30 000
bytes, Grep skips files over 1 MB, reads 1 000 characters of a line and refuses a pattern with a repeated
group that can backtrack without bound. A run ends at 60 turns.
Not carried over from the Claude runner: resuming a session (each run starts fresh), steering a run while it works, images, and cost: an endpoint reports tokens, not dollars, so cost is 0 and the turn limit holds the budget.
Pinned by test/openai-runner.test.ts; the code is src/crew/openai_runner.ts.