8c2970e70e
Review found five real defects in the configuration-source work, all confirmed against the code rather than argued: 1. The note claimed --config outranks settings.yaml. It does not: the settings seam registers a plugin's cordis entry config as the `base` layer and the user section layers over it, and the seam cannot tell a shipped value from a --config one. The note now states shipped reality and names --config-replace as the lever for a deployment that must win. Separately, a literal `apiKey` in settings outranked both the environment and .credentials.yaml — the field is removed, so configuration carries a reference and nothing else. 2. DEEPSEEK_SEARCH_BASE_URL was functionally deleted: the shipped inline went away without the provider learning to read it. It now resolves from the environment snapshot, as the README always claimed. 3. The bootstrap deny list missed the interpreter start-up hooks. BASH_ENV is the sharpest: `bash -c` sources it on every bash tool call, so a project .env could run a file of its choosing before every command. The list now covers BASH_ENV and its per-language siblings, the Git hook commands, and the remaining preload and CA variables, organised by what a variable does rather than which runtime owns it. 4. YAML parse errors quoted the offending source line — which in a credentials document is the secret — into boot stderr and the watcher's logger. Only the error code and position are reported now, in credentials-local and settings-local alike, pinned by a test that asserts the secret is absent. 5. 0600 governed only files the harness wrote. A hand-created 0644 document was read normally. POSIX now checks the mode before reading contents, at boot and on every reload; Windows has no mode to inspect and is skipped rather than faked. The project a session is launched in is trusted by default, with no prompt and no stored trust record: it may supply its own endpoint, ordinary variables, and a key ranked below the managed store. Trust stops at the harness itself — a discovered file still cannot set DSH_PERMISSION_MODE, PATH, BASH_ENV, or the rest, because those take effect with no user action, before any turn, outside the permission policy and the sandbox.
headless-agent
English | 中文
Headless one-shot agent wiring: DeepSeek V4 + local bash and filesystem tools + subagent delegation + workflows and fresh-agent Ralph iteration + todo_write + JSONL persistence, with @deepseek-ai/dsh-cli-demo as the app 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 demo:headless "fix the failing test in this workspace"
pnpm run demo:headless --output-format json -- "summarize the implementation"
pnpm run demo:headless --output-format stream-json -- "run the focused tests"
Exactly one nonblank positional task is required; quote tasks containing spaces. There is no -p flag. text prints the last text-bearing assistant message, json prints one DSH-native result record, and stream-json emits the top-level session's canonical task-turn events before that record. Child sessions surface only through parent tool events and results.
Each invocation creates and persists a fresh session, runs all model and tool steps in one turn, flushes, disposes, and exits. This is non-interactive automation: there is no prompt, approval, resume, second turn, or stdin context. The configured tools can mutate the launch workspace, run commands, spawn child agents, and consume provider tokens.
Advanced and snapshot wiring
advanced.cordis.yml adds Code Mode and the Cordis tools to the shipped leaf. advanced.cordis.snapshot.yml replaces only the live LLM with replay. The tests under tests/ own the keyless real-Loader smoke, key-gated world-verified smoke, and the stream-json replay snapshot with its parent and child session fixtures.
The package-level CLI contract documents output records, exit status, cancellation, persistence, and model/token effects.