kairos-bundle
kairos-bundle
Export all KAIROS protocols to a directory of .md files, or import all .md files from a directory into KAIROS (personal space).
Prerequisites
- Python 3 — Interpreter to run the script.
- KAIROS CLI — For obtaining a token (
kairos token); must be onPATH. - KAIROS MCP — Server running and reachable (e.g.
KAIROS_API_URLor defaulthttp://localhost:3000). - KAIROS_TOKEN — Bearer token (e.g.
KAIROS_TOKEN=$(kairos token)). - KAIROS_API_URL — Optional; default
http://localhost:3000.
Run the script
From the skill directory (the directory that contains this SKILL.md — e.g. skills/kairos-bundle in the repo or ~/.agents/skills/kairos-bundle when installed):
# Export: save all protocols from spaces into --dir
KAIROS_TOKEN=$(kairos token) python3 scripts/kairos-bundle.py export [--dir DIR]
# Import: mint all .md files from --dir into KAIROS (personal space)
KAIROS_TOKEN=$(kairos token) python3 scripts/kairos-bundle.py import [--dir DIR]
So the script path is always scripts/kairos-bundle.py relative to the skill directory; the agent should run these commands with the current working directory set to the skill directory.
Do not confuse: the skill directory (cwd when running) is where the skill and script live. The --dir option is the bundle directory — where .md files are written (export) or read from (import); it is independent of the skill directory.
Critical — always use an absolute path for --dir when the user means a project path. The script is run with cwd = skill directory (so scripts/kairos-bundle.py is found). Any relative path passed to --dir is then resolved relative to that cwd (e.g. ~/.agents/skills/kairos-bundle when the skill is installed there). So if the user says e.g. "export to kairos/bundles/trunk", they mean relative to the workspace/project root, not the skill directory. The agent must resolve that path against the workspace root and pass the absolute path to --dir (e.g. --dir /path/to/project/kairos/bundles/trunk or --dir "$(pwd)/kairos/bundles/trunk" when the shell is already in the project root). Otherwise files will be written under the skill directory or ~/.agents, which is wrong.
Options
| Action | Option | Description |
|---|---|---|
| export | --dir |
Target directory to write .md files (default: .local/cache/dump) |
| export | --space |
Only export chains from this space (e.g. Personal) |
| export | --dry-run |
List chains only, do not save |
| import | --dir |
Target directory to read .md files from (default: .local/cache/dump) |
| import | --force |
Overwrite existing chain with same title |
| import | --dry-run |
List .md files only, do not mint |
Examples
When the bundle directory is inside the user's project, use an absolute path (resolve from workspace root):
# From project root; BUNDLE_DIR is absolute so it doesn't depend on skill cwd
cd /path/to/project
KAIROS_TOKEN=$(kairos token) python3 /path/to/skill/scripts/kairos-bundle.py export --dir "$(pwd)/kairos/bundles/trunk" --space "Personal"
KAIROS_TOKEN=$(kairos token) python3 /path/to/skill/scripts/kairos-bundle.py import --dir "$(pwd)/kairos/bundles/trunk" --force
Or call from the skill directory and pass an absolute bundle path:
cd ~/.agents/skills/kairos-bundle # or skills/kairos-bundle in repo
KAIROS_TOKEN=$(kairos token) python3 scripts/kairos-bundle.py export --dir /path/to/project/kairos/bundles/trunk
More from debian777/kairos-mcp
kairos
>-
49kairos-install
>-
18kairos-bug-report
>-
13kairos-ai-docs
Find and run the zero-drift minimal template for writing AI instructions via KAIROS. Use when the user invokes /ai-docs or asks to write, generate, or update AI instructions using the zero-drift template.
7kairos-code
Run the KAIROS ELITE AI CODING STANDARDS protocol. Use when the user invokes /code, asks for AI coding rules, or wants code changes to follow the full protocol (feature branch, baseline tests, full suite, deploy to dev, validate, then promote).
7kairos-create-skill
Run the KAIROS protocol for creating a KAIROS skill (Agent Skill with optional references/KAIROS.md or multiple references/KAIROS-{alias}.md protocols). Use when the user wants to create a KAIROS skill, create a skill with a KAIROS protocol, write SKILL.md, or asks about skill structure and Agent Skills format (agentskills.io).
7