# Add to your Claude Code skills
git clone https://github.com/sysid/bkmrGuides for using ai agents skills like bkmr.
Last scanned: 5/30/2026
{
"issues": [],
"status": "PASSED",
"scannedAt": "2026-05-30T15:19:56.059Z",
"npmAuditRan": true,
"pipAuditRan": true
}See how bkmr compares with popular alternatives.
bkmr is an open-source ai agents skill for AI coding assistants such as Claude Code, Codex CLI, and ChatGPT, built by sysid. Knowledge Management for Humans and Agents. It has 269 GitHub stars.
Yes. bkmr passed SkillsLLM's automated security scan — a dependency vulnerability audit plus prompt-injection heuristics — with no high-severity issues. You can read the full report in the Security Report section on this page.
Clone the repository with "git clone https://github.com/sysid/bkmr" and add it to your Claude Code skills directory (see the Installation section above).
bkmr is primarily written in Rust. It is open-source under sysid on GitHub, so you can review or fork the full source.
Yes. SkillsLLM lists many other AI Agents skills you can browse and compare side by side. Open the AI Agents category from the badge at the top of this page, or use the Related Skills and comparison links further down to weigh bkmr against similar tools.
No comments yet. Be the first to share your thoughts!
From $49 · 30 days · terms & refunds
⚠️ Third-Party Software Notice
This skill is third-party open-source software developed and hosted independently on GitHub. SkillsLLM lists this project for discovery and does not control or maintain the underlying repository.
Any security checks, ratings, or warnings displayed by SkillsLLM are automated and limited in scope. They do not constitute a security certification or guarantee that the software is safe, error-free, or free from malicious code, vulnerabilities, compromised dependencies, or prompt-injection risks.
Review the source code, permissions, dependencies, and configuration before installing or running any third-party skill. Use is at your own risk. To the maximum extent permitted by applicable law, SkillsLLM is not liable for losses arising from third-party software.
Automated security scan — not a certification.
[![Build Status][build-image]][build-url]
Store anything, find it by meaning, act on it instantly.
bkmr - crate of the week 482 - memories, bookmarks, snippets, text - search it, invoke it!
Organize, find, and apply various content types:
_mem_ system tag for AI agent memory--json, --np, --stdout for pipelines and integrationsPersistent long-term memory for AI agents. The _mem_ system tag and hsearch (hybrid FTS + semantic search) create a complete read/write memory interface:
# Agent stores memory:
bkmr add 'Prod DB is PostgreSQL 15 on port 5433' fact,database \
--title 'Production database config' -t mem --no-web
# Agent queries memories with natural language (hybrid search)
bkmr hsearch "database configuration" -t _mem_ --json --np
# All output is structured JSON — designed for programmatic consumption
Install the bkmr-memory skill into your agent's skills directory. It defines a comprehensive memory protocol with taxonomy, deduplication, and session workflows. The copy embedded in the binary matches its CLI flags:
bkmr skill install ~/.claude/skills # or ~/.agents/skills, or any agent's skills dir
# alternative, tracks the latest release:
gh skill install sysid/bkmr bkmr-memory --agent claude-code --scope user
See Agent Integration.
# Quick fuzzy search with interactive selection
bkmr search --fzf
# Add URL with automatic metadata extraction
bkmr add https://example.com tag1,tag2
# Store code snippet
bkmr add "SELECT * FROM users" sql,_snip_ --title "User Query"
# Shell script with interactive execution
bkmr add "#!/bin/bash\necho 'Hello'" utils,_shell_ --title "Greeting"
# Render markdown in browser with TOC
bkmr add "# Notes\n## Section 1" docs,_md_ --title "Project Notes"
# Import files with frontmatter
bkmr import-files ~/scripts/ --base-path SCRIPTS_HOME
# Local semantic search (no API keys needed)
bkmr hsearch --mode semantic "containerized application security"
General Usage:

Fuzzy Search with FZF:

Agent Memory:

Detailed walkthroughs: Overview | Getting Started | Search and Filter | Edit and Update | Tag Management
# Via cargo
cargo install bkmr
# Via pip/pipx/uv
pip install bkmr
# Via brew
brew install bkmr
export ORT_DYLIB_PATH=/opt/homebrew/lib/libonnxruntime.dylib
See Getting Started for details and Troubleshooting if it fails.
# Generate configuration
bkmr --generate-config > ~/.config/bkmr/config.toml
# Create database
bkmr create-db ~/.config/bkmr/bkmr.db
# Optional: Configure location
export BKMR_DB_URL=~/path/to/db
# Add your first bookmark
bkmr add https://github.com/yourusername/yourrepo github,project
# Search and find
bkmr search github
# Interactive fuzzy search
bkmr search --fzf
5-minute tutorial: Getting Started.
| Command | Description |
|---|---|
search |
Full-text search with tag filtering, FZF, JSON output |
hsearch |
Hybrid search: FTS + semantic with RRF fusion; --mode exact: keywords only; --mode semantic: meaning only, using local embeddings (offline, no API keys). Each hit reports its cosine similarity and whether it was a keyword_match |
add |
Add bookmarks (URLs, snippets, scripts, markdown, env vars) |
open |
Smart action dispatch based on content type |
edit |
Edit bookmarks (smart: opens source file for imports) |
update |
Change tags, title, description, content, opener or embedding, non-interactively |
delete |
Delete bookmarks by ID (--yes: no prompt; --force: also purge history, permanently) |
history |
Show saved old states of bookmarks of every type, deleted ones included; --type: one type only; -p: each change as a unified diff |
restore |
Bring back a saved state from history, also of a deleted bookmark; --from FILE: changed and deleted bookmarks of any type from a search --json backup; --dry-run: show the diff only |
verify |
Record that bookmarks were checked against reality (verified_at) |
show |
Display bookmark details |
import-files |
Import files/directories with frontmatter parsing |
load-json |
Add bookmarks in bulk from a JSON array; skips entries whose URL/content exists, leaves existing bookmarks as they are |
tags |
View tag taxonomy with usage counts |
info |
Show configuration, database path, embedding status |
create-db |
Initialize a bookmark database |
backfill |
Generate missing embeddings; --force: recompute all; --prune: remove vectors of non-embeddable bookmarks; --clear: delete all embeddings and content hashes, without re-embedding |
skill install |
Install the bkmr-memory agent skill into an agent's skills directory |
lsp |
Start LSP server for editor snippet completion |
completion |
Generate shell completions (bash, zsh, fish, nushell) |
surprise |
Open random web links (plain http/https, no custom opener) |
Complete command documentation: See Basic Usage for detailed examples.
A backup is the JSON export of search. restore --from brings back every bookmark that changed or was deleted since, whatever its type, under its own ID:
bkmr search --json > ~/backup/bkmr.json # write a backup
bkmr restore --from ~/backup/bkmr.json # show the plan, then [y]es [n]o [s]elect [d]iff
bkmr restore --from ~/backup/bkmr.json --ids 12,40 # only these bookmarks
bkmr restore --from ~/backup/bkmr.json --dry-run # unified diffs, nothing written
Entries are matched by ID, so a bookmark whose URL/content changed is still recognized. Bookmarks the backup does not contain are never touched. See Restoring from a JSON Backup.
bkmr intelligently handles different content types with appropriate actions:
| Content Type | Default Action | System Tag |
|---|---|---|
| URLs | Open in browser | (none) |
| Snippets | Copy to clipboard | _snip_ |
| Shell Scripts | Interactive edit + execute | _shell_ |
| Markdown | Render in browser with TOC | _md_ |
| Environment Variables | Print for eval/source |
_env_ |
| Text Documents | Copy to clipboard | _imported_ |
| Agent Memory | Display to stdout | _mem_ |
Rule: A bookmark can have at most one system tag. Local files without a system tag open with the default application.
URL/content: every bookmark keeps its payload in one field, named URL for historical reasons. For a web bookmark it is a web address; for every other type it is the content itself: the snippet, script, markdown, env vars or memory text. Help texts and docs therefore say URL/content; in JSON the field is
url.
Learn more: Content Types | Core Concepts
Comprehensive documentation is available in the bkmr Wiki: