hister

hister [globaloptions] command [args_]

hister is a private search engine for pages you visit and files you keep. One binary contains the HTTP server, a command-line client, and a terminal UI. It indexes full text so you can search from the web interface at http://127.0.0.1:4433, from the terminal, or through an MCP endpoint for AI assistants.A typical local setup is hister listen, then the Firefox or Chrome extension to capture newly visited pages. The server must stay running for indexing and search. No configuration is required for a single-user bind on localhost.Queries support field filters (`title:`, `text:`, `url:`, `domain:`, `label:`, `language:`, `type:`, `visits:`, `added:`, `updated:`, `user_id:`), quoted phrases, wildcards, negation, `sort:`, and `(a|b)` alternation. hister index can fetch individual URLs or run persistent, resumable crawls whose queues live in the database; hister crawl only inspects those jobs.Local commands such as crawl list and most user-admin operations do not contact the HTTP server. Remote commands such as search, export, and cleanup talk only to the configured server. Hybrid commands such as persistent index can use both.

--config PATH

Config file. Search order without this flag: `./config.yml`, `~/.histerrc`, `~/.config/hister/config.yml`. Also settable via HISTER_CONFIG.

Hister server URL (overrides server.base_url).

Access token (overrides app.access_token).

error, warning, info (default), debug, or trace.

Default external search engine URL containing a `{query}` placeholder.

HTTP timeout in seconds for server communication (0 = none; default 10).

Start the search server. Default bind is 127.0.0.1:4433. Keep the process running while clients index or search.

With no arguments, open the TUI. With a query, print matching documents. --format / -f is text, json, jsonl, or csv. --fields selects columns; --limit caps the result count.

Fetch pages and add them to the running server. Direct URLs are fetched once. --recursive / -r starts a persistent crawl; --input FILE queues one URL per line (- reads stdin); --job-id NAME creates or resumes a named job.

Inspect persistent crawl jobs without fetching. Subcommands include list, show, urls, queue, errors, and delete operations.

Import JSON exports, 7z archives of those exports, saved HTML, and local file snapshots (PDF, DOCX, Markdown, Org, UTF-8 text). Talks to a running server.

Write indexed documents as JSON. Use - for stdout. Optional query and --start-date / --end-date (`YYYY-MM-DD`) limit the dump.

Change user_id, label, title, or language on matching documents. Prompts unless --yes or --dry.

create [file] writes defaults (refuses to overwrite). path prints the selected file. show prints effective YAML with credentials redacted. validate checks keys and types.

Check local config, extractor executables, and server connectivity. --format selects text, json, jsonl, or csv. Exit 1 if any check fails.

Compare the installed version with the latest release. Does not download an update.

List commands. Root help labels each operation as local, remote, or hybrid.

Without --config or HISTER_CONFIG, Hister looks for `./config.yml`, then `~/.histerrc`, then `~/.config/hister/config.yml`. TUI settings live in a sibling tui.yaml (created on first hister search). After editing tui.yaml, restart the TUI.`hister config create ~/.config/hister/config.yml` writes a starter file. `hister config show` prints the effective YAML. Server listen address, database path, authentication, directory watchers, crawler backends, and optional embeddings are documented in the configuration reference.

Indexing and most search operations require a running hister listen (or equivalent service). Closing that terminal stops the server. Browser capture needs the official extension; the CLI can still index URLs and import files without it.index exit status is 0 on success (skipped URLs allowed), 1 on setup/execution failure, and 2 when item errors occurred. Search JSON/JSONL can be partial if a later page fails; check the exit status. Optional semantic search sends document text to whatever embeddings endpoint you configure.

Hister is developed by Adam Tauber (asciimoo), author of Searx. It is written in Go and released under AGPLv3. Source is published on GitHub and mirrored on Codeberg.

curl(1), wget(1), lynx(1), rg(1), grep(1)

Source code · Homepage · Documentation