Back to corpus
architecturetechnical paper candidatescore 36

KARL Architecture

``` Recording -> Scoring -> Analysis -> Training -> Improved Routing ^ | | | +------------- Better trajectories <-------------+ ```

Full HTML reader

Read the full artifact

Open in new tab

Extracted abstract or opening context

KARL is a trajectory-based learning system for AI coding agents. It implements a closed-loop pipeline: - **Tap A** (`init_session_buffer`): Opens a JSON buffer when a session starts or a skill is activated. Captures the prompt, working directory, and skill name. - **Tap B** (`append_tool_event`): After every tool call (Read, Edit, Write, Bash, Grep, Glob, Task), appends the tool name, key parameters, success/failure, and timing to the buffer. - **Tap C** (`flush_session`): When the agent finishes responding, flushes the buffer to `trajectories.jsonl`. Runs the reward engine to compute a score before writing. - **Tap D** (`annotate_previous`): On the next prompt, examines the text for correction signals ("no, I meant...", "try again"). Retroactively annotates the previous trajectory with a failure signal.

Promotion decision

What has to happen next

Promote into a technical note or architecture paper with implementation anchors.

Why this is not always a full paper yet

Corpus pages are public-safe readers for discovered workspace artifacts. They are not automatically final papers. A corpus item becomes a polished paper only after the editable source, evidence checkpoints, references, figures, render path, and release status are attached through the paper schema.