Menu
Use when ingesting, cognifying, or querying a codebase with Cognee CLI from Codex.
Installer avec Codex ou Claude Copiez ce prompt, collez-le dans Codex, Claude ou un autre assistant, puis laissez-le vérifier la page du skill et l'installer pour vous.
Sync session cache entries into the permanent Cognee knowledge graph. Run this to make session memory searchable, or it runs automatically at session end.
Store data permanently in the Cognee knowledge graph. Accepts a data category (user, project, or agent) to tag the data with the correct node_set for filtered retrieval.
Search Cognee memory. Session memory is automatically searched on every prompt via hooks. Use this skill explicitly for permanent knowledge graph search, filtered category search, or when you need more results than the automatic lookup provides.
Sync session cache entries into the permanent Cognee knowledge graph. Run this to make session memory searchable, or it runs automatically at session end.
Store data permanently in the Cognee knowledge graph. Accepts a data category (user, project, or agent) to tag the data with the correct node_set for filtered retrieval.
Use when Codex should remember, recall, search, improve, or forget information using Cognee.
Basé sur la classification professionnelle SOC
| name | codebase |
| description | Use when ingesting, cognifying, or querying a codebase with Cognee CLI from Codex. |
Use this skill when the user asks Codex to build a Cognee memory of a repository, index code, query implementation details, or create a code-aware knowledge graph.
uv run cognee-cli ...; do not use MCP..env, private keys, credentials, local database files, virtualenvs, dependency caches, or generated build output.rg --files first to inspect candidate paths.Create a dataset name that is stable and readable, such as:
codebase-cognee
codebase-frontend
codebase-api
Inspect candidate files:
rg --files
Common exclusions include:
.git/
.venv/
node_modules/
dist/
build/
.next/
coverage/
.env
*.sqlite
*.db
*.key
*.pem
For a focused set of source paths:
uv run cognee-cli add <path-1> <path-2> <path-3> -d <dataset-name>
uv run cognee-cli cognify -d <dataset-name> --background
For small docs or architectural notes:
uv run cognee-cli remember <path-or-note> -d <dataset-name>
If command length becomes unwieldy, ingest in batches by directory or feature
area, then run cognify once for the dataset.
For code-specific questions:
uv run cognee-cli search "<implementation question>" -d <dataset-name> -t CODE -k 10 -f pretty
For architecture and reasoning questions:
uv run cognee-cli recall "<architecture question>" -d <dataset-name> -t GRAPH_COMPLETION -f pretty
For citation-like output:
uv run cognee-cli search "<specific symbol or behavior>" -d <dataset-name> -t CHUNKS -k 10 -f json
Use results as supporting context. Verify important claims against the actual files before editing code.
The server is the source of truth. cognee-cli can print empty stdout even when content exists, so never conclude "not found" from an empty CLI run — confirm against the server directly (authoritative), and omit -d <dataset> to search all datasets:
curl -s -X POST "${COGNEE_BASE_URL:-http://localhost:8011}/api/v1/recall" \
-H "Content-Type: application/json" \
-H "X-Api-Key: ${COGNEE_API_KEY:-}" \
-d '{"query": "<question>", "top_k": 10, "only_context": true, "scope": ["graph"]}'