Automation entrypoint
Run Kinic Wiki from the CLI
kinic-vfs-cli is the command-line interface for Kinic Wiki databases. It lets automation search, read, edit, connect databases, and record Skill Registry evidence without opening the browser.
Install
Install the published npm package. It downloads the matching kinic-vfs-cli release binary and verifies its SHA-256 checksum.
npm install -g kinic-vfs-cliCLI workflow
- 1InstallAdd the CLI package once per machine.
- 2ConnectLink a database or pass a database id per command.
- 3Use JSONRequest structured output for agents and scripts.
- 4Guard writesRead etags before editing existing nodes.
First Check
Confirm the installed binary, then use --help to inspect the command index before connecting automation to a database.
kinic-vfs-cli --version
kinic-vfs-cli --helpConnect Database
Pass --database-id for one command, run database link once for a workspace, or set VFS_DATABASE_ID in scripted environments.
kinic-vfs-cli --database-id <database-id> status --json
kinic-vfs-cli database link <database-id>
VFS_DATABASE_ID=<database-id> kinic-vfs-cli status --jsonRead Workflow
Search first, read exact paths next, then request context when link relationships matter. Use --json so agents can parse results safely.
kinic-vfs-cli search-remote "query text" --prefix /Knowledge --top-k 10 --json
kinic-vfs-cli read-node --path /Knowledge/page.md --json
kinic-vfs-cli read-node-context --path /Knowledge/page.md --jsonStore API Reads
Use Store API commands for task-scoped recall, source evidence references, full-scope snapshots, and trusted snapshot deltas.
kinic-vfs-cli memory-manifest --json
kinic-vfs-cli query-context --task "answer auth question" --namespace /Knowledge --entity auth --budget-tokens 8000 --depth 1 --json
kinic-vfs-cli source-evidence --node-path /Knowledge/page.md --json
kinic-vfs-cli export-snapshot --prefix /Knowledge --limit 100 --json
kinic-vfs-cli fetch-updates --known-snapshot-revision <revision> --prefix /Knowledge --limit 100 --jsonSafe Write Workflow
Read the node first, keep its etag, edit with --expected-etag, then read again to verify the stored content.
kinic-vfs-cli read-node --path /Knowledge/page.md --json
kinic-vfs-cli edit-node --path /Knowledge/page.md --old-text before --new-text after --expected-etag <etag> --json
kinic-vfs-cli read-node --path /Knowledge/page.md --jsonSkill Registry
Find a skill, inspect the package before use, then record run evidence after the agent completes the task.
kinic-vfs-cli skill find "contract review" --json
kinic-vfs-cli skill inspect legal-review --json
kinic-vfs-cli skill record-run legal-review --task "review contract" --outcome success --notes-file ./notes.md --jsonRaw canister calls
Use raw calls for endpoint debugging and database SQL without installing kinic-vfs-cli. Use kinic-vfs-cli for scripted reads and safe writes.
Safety Notes
- Public reads can run with --identity-mode anonymous when the database grants anonymous reader access.
- Writes, database grants, archive operations, and private Skill Registry writes require identity mode.
- Non-Internet Identity credentials require the explicit --allow-non-ii-identity opt-in.
- Agents should request JSON output and use etag guards before mutating existing nodes.
- The CLI intentionally omits delete_database, canister_health, wiki_metrics, and wiki_metrics_series.