Install
openclaw skills install rollbarMonitor and manage Rollbar error tracking. List recent items, get item details, resolve/mute issues, track deployments, and manage project access tokens via the Rollbar API.
openclaw skills install rollbarMonitor and manage Rollbar errors directly from OpenClaw.
The script resolves your Rollbar token from the first matching source:
| Priority | Method | How |
|---|---|---|
| 1 | $PWD/secrets/rollbar | One-line file in your agent workspace containing the token |
| 2 | .rollbar-mcp.json | MCP-compatible JSON config (cwd, ~/.rollbar-mcp.json, or $ROLLBAR_CONFIG_FILE) |
| 3 | $PWD/.env | File in your agent workspace with ROLLBAR_ACCESS_TOKEN=your-token |
| 4 | ROLLBAR_ACCESS_TOKEN | Environment variable (injected or shell) |
Create .rollbar-mcp.json in your agent workspace:
// Single project
{ "token": "tok_abc123" }
// Multiple projects
{
"projects": [
{ "name": "linkz-api", "token": "tok_abc123" },
{ "name": "linkz-dashboard", "token": "tok_xyz789" },
{ "name": "linkz-php", "token": "tok_def456" }
]
}
Use --project-name <name> to select a specific project's token when running commands.
Tip: Use
project-token-create --saveto have the agent auto-create and persist tokens to.rollbar-mcp.json.
# workspace-myagent/.env
ROLLBAR_ACCESS_TOKEN=your-token-here
read scope for monitoring; add write scope to resolve/mute. Best for single-project use.projects, project-tokens, and project-token-create. Use --project-id to target specific projects.All commands use the helper script rollbar.sh in this skill directory.
./skills/rollbar/rollbar.sh projects
./skills/rollbar/rollbar.sh items [--project-id <id>] [--project-name <name>] [--status active|resolved|muted] [--level critical|error|warning|info] [--limit 20]
./skills/rollbar/rollbar.sh item <item_id>
./skills/rollbar/rollbar.sh occurrences <item_id> [--limit 5]
./skills/rollbar/rollbar.sh resolve <item_id> --yes
# or preview without making changes:
./skills/rollbar/rollbar.sh resolve <item_id> --dry-run
./skills/rollbar/rollbar.sh mute <item_id> --yes
./skills/rollbar/rollbar.sh activate <item_id> --yes
Safety interlock:
resolve,mute,activate,project-token-create, andproject-token-updaterequire--yesto execute or--dry-runto preview. Omitting both prints a warning and exits with an error — preventing silent state changes from mistaken or manipulated invocations.
./skills/rollbar/rollbar.sh deploys [--limit 10]
./skills/rollbar/rollbar.sh project
./skills/rollbar/rollbar.sh top [--limit 10] [--hours 24]
These commands require an account-level token with write scope.
./skills/rollbar/rollbar.sh project-tokens --project-id <id>
Token values are truncated in output (first 8 chars + ...) for safety.
./skills/rollbar/rollbar.sh project-token-create \
--project-id <id> \
--name "openclaw-agent" \
--scopes read,write \
[--project-name <name>] \
[--save]
--scopes — comma-separated, e.g. read or read,write (default: read,write)--save — automatically saves the new token to .rollbar-mcp.json using --project-name (or --name) as the key--project-name — name used as the key in .rollbar-mcp.json when --save is setBootstrap workflow — let the agent provision its own project tokens:
# Set account token first
export ROLLBAR_ACCESS_TOKEN=your-account-token
# Create tokens for each project and save to MCP config
./rollbar.sh project-token-create --project-id 378962 --name "linkz-api-agent" --scopes read,write --project-name linkz-api --save --yes
./rollbar.sh project-token-create --project-id 462118 --name "linkz-dashboard-agent" --scopes read,write --project-name linkz-dashboard --save --yes
./rollbar.sh project-token-create --project-id 755542 --name "linkz-php-agent" --scopes read,write --project-name linkz-php --save --yes
# Now use per-project tokens directly from config
./rollbar.sh items --project-name linkz-api
./skills/rollbar/rollbar.sh project-token-update <token_id> \
--project-id <id> \
--token-status enabled|disabled
To get automatic alerts for new critical/error items, set up a cron job in OpenClaw:
"Check Rollbar for new critical or error-level items in the last hour. If any new items appeared, summarize them and alert me."
Recommended schedule: every 30–60 minutes during work hours.
top command sorts active items by occurrence count — useful for daily triage.project-tokens output are truncated for safety; full values are only returned on creation.