Switch this session to another Cognee dataset, or back to the previous one, so later saves and recalls use it. Takes a dataset name (created if new) or lists writable datasets to pick from; syncs this session first.
The English summary is being prepared. Showing the original description.
See how to installReview the instructions the agent will receive before you install.
Move this Claude Code session to a different Cognee dataset. A Cognee session never spans two datasets, so the switch retires the current session (after syncing it into the graph) and starts a new one on the chosen dataset.
Where Claude Code loads the plugin's hooks module (2.1.287+), the user can switch
without a model turn: /cognee-dataset <name> from the prompt (/cognee-dataset
alone lists the datasets), or the dataset name in the status row above the prompt,
which is a picker (in a terminal, focus it with ctrl+x tab). If the user asks you
to switch, run this skill as below; if they ask
how to switch themselves, point them there.
When $ARGUMENTS contains a dataset name, skip the picker:
python3 "${CLAUDE_PLUGIN_ROOT}/scripts/switch-dataset.py" "$ARGUMENTS" --json
python3 "${CLAUDE_PLUGIN_ROOT}/scripts/switch-dataset.py" --list --json
The JSON has current, datasets ([{name, id, writable, current}]),
hidden_readonly (datasets you can read but not write — never offered) and
filtered (true when write access was verified for every row).
Present the choice with AskUserQuestion (single select), listing the
dataset names from datasets with the current one marked "(current)". The
tool allows four options per question, so if there are more than four, put
the first three plus a "More…" option and page on the next question. The
user can always type a name via the built-in "Other" — a name that is not
listed is created for them on switch. If filtered is false, say so in the
question text ("write access could not be verified for these").
Then run the switch with the chosen name:
python3 "${CLAUDE_PLUGIN_ROOT}/scripts/switch-dataset.py" "<chosen name>" --json
On success the JSON is
{"switched": true, "dataset", "session_id", "previous": {"dataset", "session_id", "synced": true|false}}.
Tell the user in one line which dataset and Cognee session are now active and
that the previous session was synced. The status line updates within a couple
of seconds. If "switched": false with "reason": "already_active", say the
session is already on that dataset.
On failure the JSON is {"error", "code"}:
| code | meaning | what to do |
|---|---|---|
| 2 | launch record not found | the plugin did not initialise this session; run "${CLAUDE_PLUGIN_ROOT}/scripts/cognee-doctor.sh". If several Claude Code sessions share this directory, rerun with --session-key <host session id> |
| 3 | syncing the current session failed | nothing was changed; show the error. Re-run with --force only if the user explicitly accepts that unsynced entries of the current session are retried at session end instead |
| 4 | registering the new session failed | nothing was changed; the server rejected the registration — check the connection |
| 5 | dataset not writable | the name is readable but owned by another principal; pick from --list |
sync-session-to-graph.py --strict) — aborts if this fails.touched list; the session-end sync covers them again as a safety net.A switch is not needed just to look in another dataset: a one-off graph search
leaves the session where it is — "${CLAUDE_PLUGIN_ROOT}/scripts/cognee-search.sh" "<query>" 10 --graph --dataset-id <id> (see the cognee-search skill).
COGNEE_PLUGIN_DATASET only seeds the dataset at launch; a switch overrides it for the rest of the session and survives --resume.
Recall is scoped to the active dataset, so after a switch earlier context from the previous dataset is no longer injected — switch back to see it again.
No reviews yet. Tell others what you thought.
No login needed. Share how it worked for you, ask a question, or note where it did not run.
No comments yet. Be the first to write one.
Use when indexing a code repository into Cognee's code graph or querying it (callers, impact analysis, paths, endpoints) from Codex.
The English summary is being prepared. Showing the original description.
Use when indexing a code repository into Cognee's code graph or querying it (callers, impact analysis, paths, endpoints) from Antigravity.
The English summary is being prepared. Showing the original description.
Explain a repository's architecture (modules, routes, storage, services), find risky hotspots, draw a Mermaid diagram of the code graph, or review how far a change reaches, from Cognee's code graph.
The English summary is being prepared. Showing the original description.
Index a repository (local path or git URL) into Cognee's code graph and answer exact structural queries with no LLM calls: callers, dependencies, impact of changing a symbol, call paths, endpoints, re-index changes.
The English summary is being prepared. Showing the original description.
A multi-day digest of the user's own coding-agent sessions — sessions per day and project, most-edited files, and every lesson Cognee distilled into the knowledge graph in that window. Use only when the user explicitly asks for a weekly digest, a retro, a recap or report of their recent sessions, or asks for the `cognee-digest` skill by name. Not for questions about code or the current task.
The English summary is being prepared. Showing the original description.
A digest of a week or more of the user's own coding-agent sessions (default 7 days): sessions per day and project, most-edited files and the lessons Cognee learned, written up as shipped, decided and still open.
The English summary is being prepared. Showing the original description.