Skip to main content

Bog Agents Daemon — the patient watcher. Runs your bog-agents on cron schedules, file-change triggers, webhooks, and git push events; survives reboots, persists state, and reports back. Use it when you want an agent that wakes itself rather than waiting for you. Pass through in harmony.

Project description

Bog Agents Daemon

The patient watcher. Wakes itself. Pass through in harmony.

Quiet ambient runner for bog-agents. Schedules. File watches. Inbound webhooks. Git pushes. Sits in the background, fires the agent when something happens, writes the result wherever you point it.

No terminal needed. No hand-holding. It keeps watch through the night and goes the distance.

PyPI Python License


Why a daemon

The CLI is great when you're at the keyboard. The daemon is what you reach for when you want an agent that watches for you — and reports back when something matters.

  • Five trigger types: cron, interval, file_change, webhook, git_push.
  • Seven output targets: log, stdout, file, slack, webhook, email, github_comment.
  • Auth + integrity: token-authenticated REST API; HMAC-validated inbound webhooks; tokens stored with 0o600 permissions.
  • Durable: os.fsync()-durable job persistence so a hard kill never loses a freshly-created job.
  • Cross-platform: POSIX systemd, Windows Task Scheduler, or just bog-agents-daemon run in a shell. Same config either way.

If the CLI passes through in harmony, the daemon is what keeps watch through the night.


Install

pip install bog-agents-daemon

Pulls in bog-agents automatically. Add provider extras you need:

pip install "bog-agents-daemon[anthropic]"
pip install "bog-agents-daemon[openai]"
pip install "bog-agents-daemon[bedrock]"

30-second tour

Start the daemon (foreground for a quick test):

bog-agents-daemon run --port 7878

Add a job that runs every weekday morning:

bog-agents-daemon job add \
  --name morning-brief \
  --cron "0 9 * * 1-5" \
  --prompt "Summarize what changed in this repo since yesterday." \
  --output slack:#engineering

The job persists to ~/.bog-agents/daemon/jobs/. The scheduler picks it up on the next tick and fires it on the configured cadence.


Triggers

# A cron job
triggers:
  - type: cron
    cron: "0 9 * * 1-5"      # 9am Mon–Fri

# An interval
triggers:
  - type: interval
    seconds: 1800            # every 30 min

# A file watcher
triggers:
  - type: file_change
    patterns: ["src/**/*.py"]
    debounce_seconds: 5

# An inbound webhook (HMAC-validated)
triggers:
  - type: webhook
    path: /hooks/incident
    secret_env: WEBHOOK_SECRET

# A git push (post-receive on a remote)
triggers:
  - type: git_push
    repo: /srv/git/main.git
    branch: main

A job can have multiple triggers. The agent fires on any of them.


Outputs

Where the result of an agent run goes. Configure one or many:

outputs:
  - type: log              # systemd journal / stderr
  - type: file
    path: ~/.bog-agents/runs/morning-brief.md
  - type: slack
    channel: "#engineering"
    token_env: SLACK_BOT_TOKEN
  - type: webhook
    url: https://hooks.example.com/agent-output
    hmac_secret_env: OUTBOUND_WEBHOOK_SECRET
  - type: email
    to: oncall@example.com
    from: bog-agents@example.com
    smtp_env_prefix: SMTP_     # SMTP_HOST, SMTP_USER, SMTP_PASSWORD
  - type: github_comment
    repo: example/api
    issue: 1234
    token_env: GITHUB_TOKEN

REST API

Once the daemon is running, you've got a small authenticated REST API for managing jobs:

Endpoint Method What
/jobs GET List all jobs
/jobs POST Create a job
/jobs/{id} GET Job detail
/jobs/{id} DELETE Delete a job
/jobs/{id}/runs GET Run history
/jobs/{id}/run POST Fire the job manually
/health GET Liveness probe
/metrics GET Counter snapshot

Every endpoint requires Authorization: Bearer <daemon_token>. The token is generated on first start, stored at ~/.bog-agents/daemon/.token with 0o600 permissions, and printed once to the foreground log so you can copy it.


Running as a service

systemd (Linux)

bog-agents-daemon install-systemd --user
systemctl --user enable --now bog-agents-daemon
systemctl --user status bog-agents-daemon

Windows Task Scheduler

bog-agents-daemon install-windows-task

Creates a bog-agents-daemon task that starts at logon and restarts on failure. Manage it with schtasks or the Task Scheduler GUI.

macOS launchd

bog-agents-daemon install-launchd --user
launchctl load ~/Library/LaunchAgents/com.bogware.bog-agents-daemon.plist

Security model

  • Token-authenticated API. Every request needs Authorization: Bearer. Tokens generated with secrets.token_urlsafe, compared with hmac.compare_digest, stored at 0o600.
  • HMAC-validated inbound webhooks. Each webhook trigger declares an env-var holding its shared secret; the daemon verifies the HMAC-SHA256 signature on every request.
  • Outbound webhook signing. Outputs that POST to a webhook can sign the body with HMAC-SHA256 so the receiver can verify it came from this daemon.
  • No secrets on disk. Provider keys, Slack tokens, and webhook secrets are all read from env vars — the daemon never persists them.
  • Resource limits. Configurable per-job CPU / memory / wall-clock caps via the same FeatureConfig shape as the SDK.

What's new in 0.9.x

The daemon rides the synchronized monorepo version, so it inherits every SDK hardening as it lands.

  • 0.9.4 — deepagents parity and provider resilience flow up from the SDK: Anthropic, AWS Bedrock, and OpenAI are live-tested, so scheduled jobs survive a provider's bad night.
  • 0.9.1Bedrock, seamless. Automatic inference-profile resolution and auto SSO-credential refresh mean a long-lived daemon keeps firing Bedrock jobs without a stale-credential outage.
  • 0.9.0 — repo-wide security sweep; compliance auditing groundwork.
  • Carried forwardJobRun.dispatch_errors per-target capture (capped at 20 entries with an (overflow) summary) so a wide-fanout outage shows up in the runs table without blowing up the record JSON; patient-by-default execution (provider retries, stdin=/dev/null for interactive commands, virtual_mode=True filesystem confinement, structured event logs ready for log shippers).

When to use this vs. /peat in the CLI

Daemon /peat
Survives reboot ✗ (only while CLI open)
Fires while you're asleep
Webhook / git-push triggers
Slack / email / GitHub-comment outputs
Reuses your interactive agent
Zero ops (no service install)

/peat is the right tool when you're at the keyboard and want a personal assistant for the duration of the session. The daemon is the right tool when you want an agent that wakes itself.


Documentation


License

MIT. See LICENSE.

Pass through in harmony.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

bog_agents_daemon-0.9.5.tar.gz (354.4 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

bog_agents_daemon-0.9.5-py3-none-any.whl (40.9 kB view details)

Uploaded Python 3

File details

Details for the file bog_agents_daemon-0.9.5.tar.gz.

File metadata

  • Download URL: bog_agents_daemon-0.9.5.tar.gz
  • Upload date:
  • Size: 354.4 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: uv/0.11.20 {"installer":{"name":"uv","version":"0.11.20","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"Ubuntu","version":"24.04","id":"noble","libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":true}

File hashes

Hashes for bog_agents_daemon-0.9.5.tar.gz
Algorithm Hash digest
SHA256 f5a25e1b6827fda954ad52cffd099cf238c5f5f1070ba67d2865e6871d25abc5
MD5 bdfe53ce7de63e28045d5cf7414c4011
BLAKE2b-256 8993b4833b2e58285e152a377de056e9ec288375a69aad144ee30c336ffd452b

See more details on using hashes here.

File details

Details for the file bog_agents_daemon-0.9.5-py3-none-any.whl.

File metadata

  • Download URL: bog_agents_daemon-0.9.5-py3-none-any.whl
  • Upload date:
  • Size: 40.9 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: uv/0.11.20 {"installer":{"name":"uv","version":"0.11.20","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"Ubuntu","version":"24.04","id":"noble","libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":true}

File hashes

Hashes for bog_agents_daemon-0.9.5-py3-none-any.whl
Algorithm Hash digest
SHA256 4b5891e305c37a6112c12e022b2cbc74628208d2a700b3aa08e07d7419f5e827
MD5 a62cd1c948ed698480707bcec09a9dfb
BLAKE2b-256 df27c8b9589dc3ecf1130178f5e190ff911d5c47ddf181934f0ab91053739116

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page