Files
deepseek-harness/examples/headless-agent
Yichen Jiang d5af34b022 fix(headless): align the one-shot runner with the direct rosterless profile
The shipped headless profile is base + headless (the direct-core front
door), and its installation-owned migration rewrites the old three-bundle
tuple down to that template, so a one-shot run composes no preset roster.
The runner's optional preset composition was compensating in the wrong
direction: it re-added roster reads the shipped profile never satisfies,
and its golden pinned a header field only a roster-bearing deployment
produces. The runner, the headless bundle manifest, and the dsh-run golden
return to the direct shape; skills in a direct run flow through the base
host rows' global layer, which is what the repository-plugin e2e proves.
2026-08-09 23:12:29 +08:00
..
2026-08-08 15:08:07 +08:00

headless-agent

English | 中文

This directory owns the replay and real-model test composition for a headless coding agent: DeepSeek V4 + local bash and filesystem tools + subagent delegation + workflows and fresh-agent Ralph iteration + todo_write + JSONL persistence. It explicitly mounts the shared agent spine, one root agent, persistence, and checkpoint policy; it is not a second product front door.

Run it

# repo root .env (gitignored) or exported env:
#   DEEPSEEK_API_KEY=sk-…
#   DEEPSEEK_BASE_URL=https://…   # optional; defaults to the public API
pnpm run dsh run "fix the failing test in this workspace"

The product command is dsh run: it accepts one nonblank task, creates and persists a fresh session, prints the final assistant text, and exits. The root demo:headless script is only an alias of that command.

Snapshot suites run this directory's configuration through tests/fixtures/headless-driver.ts, an unexported test-only process that emits canonical session events as JSONL before its result record. That stream is test infrastructure, not a supported CLI output format. Child sessions surface only through parent tool events and results.

E2B POC overlay

e2b.cordis.yml replaces the local filesystem and subprocess providers with one shared E2B sandbox while retaining dsh-bash-local and the same model-facing tools. Put E2B_API_KEY beside DEEPSEEK_API_KEY in the gitignored root .env, then run the credential-gated live composition, which drives FS, Bash, PTY, and LSP in one sandbox and proves final deletion:

pnpm exec vitest run --config vitest.e2e.config.ts packages/e2b/e2b/tests/composition.e2e.ts

The overlay creates the same absolute cwd inside the sandbox, but it does not upload or mount the host workspace. File and Bash mutations exist only in E2B; Cordis, model calls, agent/session state, session logs, skills, and SDK buffers remain on the host. The composition kills its sandbox on timeout and disposal. It is a provider-composition POC, not a whole-harness migration or a workspace-sync feature.

Advanced configuration

advanced.cordis.yml adds Code Mode and the Cordis tools to the test composition.