Transcript Lake

Documentation / CLI

transcript-lake goal label

Infer a goal from a session’s first masked user prompt and persist it as a model label.

Invocation#

sh
transcript-lake [--data-dir <path>] goal label <session-id> [--runtime <r>] [--json]

Purpose#

Infer a goal from a session’s first masked user prompt and persist it as a model label.

Inputs and options#

  • Required <session-id> identifies an existing session with a non-empty masked user event.
  • Optional --runtime <r> disambiguates a repeated session ID; --json emits the result object.
  • Uses the same qualified local model and runtime requirements as goal title.

Output and state effect#

Reads only the first masked user prompt. When inference returns a goal, appends a goal label with source model:jeden-goal-qwen3-4b-2512d7a4; no label is written for <goal/>. JSON reports session, runtime, goal, applied state, source, and SHA-256.

Exact refusals#

  • Wrong positional count: usage: transcript-lake goal label <session-id> [--runtime <r>] [--json].
  • No eligible prompt: session <session-id> has no masked user prompt in this Lake.
  • Ambiguous identity: session <session-id> exists in multiple runtimes; pass --runtime.
  • Model/runtime validation refusals are the same exact local failures documented for goal title.

All long flags are exact and single-use. A value flag never consumes another --flag. Global --data-dir may appear anywhere in the invocation, must appear once, and selects an absolute root.

Back to the complete CLI command tree.