Browse
Installable packages, the resource types built into kdeps, and the versioned assets kdeps runs on, in one catalog. The colored badge on each entry says which kind it is.
254 entries
kdeps event: efficiency-failures
kdeps event: efficiency-file
kdeps event: efficiency-reads
kdeps event: efficiency-stops
kdeps event: efficiency-tighten
kdeps event: efficiency-verbose
kdeps event: efficiency-web
kdeps theme: emacs
kdeps kdeps-component: email
Search for documents semantically similar to a query in the local embedding database. Returns ranked results with similarity scores. Requires: query (natural language search query), collection (name of the document collection). Optional: limit (max results, default 5).
Convert text into vector embeddings using an embedding model. Returns the embedding vectors. Use for indexing documents into the local embedding database or computing semantic similarity. Requires: texts (list of strings to embed). Optional: model, backend.
kdeps provider: ernie
kdeps harness: errors
Search the web using Exa (formerly Metaphor) neural search. Finds highly relevant URLs and content using AI-powered link prediction. Best for research, finding authoritative sources, and content discovery. Requires EXA_API_KEY.
kdeps defaults: executors
kdeps command: exit
kdeps command: explain
End a session-integrity handshake challenge as failed once its round budget is spent without the model producing the expected grounding call.
Escalate a stalled goal-directed task (reanchor, narrow tools, force close, then fail forward) and eventually close it so the loop can never get trapped on one task.
kdeps kdeps-component: file
kdeps event: file-call-budget
kdeps event: file-read-limit
kdeps command: fold
kdeps event: fold
kdeps event: force-answer-digest
End the current tool-calling loop and force a final text answer from whatever was gathered, instead of letting the model keep spinning.
kdeps command: frugal
Minimizes token usage: compacts and folds history sooner, injects far less memory context, truncates tool output harder, and caps tool-call budgets tighter. Best for cost-sensitive sessions or a small-context- window local model.
kdeps model: gguf
Alias for llama-server — GGUF weights via llama.cpp OpenAI-compatible /v1