Add outbound-sales example: batch dial-out bot with text-mode evals

.claude/settings.json

@@ -1,5 +1,17 @@
 {
   "attribution": {
     "commit": ""
+  },
+  "permissions": {
+    "allow": [
+      "mcp__claude_ai_Daily_MCP_Gateway__get_api",
+      "mcp__claude_ai_Daily_MCP_Gateway__execute_code",
+      "Bash(uv run *)",
+      "Bash(rtk ls *)",
+      "Bash(python3 -)",
+      "Bash(rtk git *)",
+      "Bash(rtk read *)",
+      "Bash(rtk wc *)"
+    ]
   }
 }

outbound-sales/.gitignore

@@ -0,0 +1,224 @@
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[codz]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+#   Usually these files are written by a python script from a template
+#   before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py.cover
+.hypothesis/
+.pytest_cache/
+cover/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+db.sqlite3-journal
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+.pybuilder/
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# IPython
+profile_default/
+ipython_config.py
+
+# pyenv
+#   For a library or package, you might want to ignore these files since the code is
+#   intended to run in multiple environments; otherwise, check them in:
+# .python-version
+
+# pipenv
+#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
+#   However, in case of collaboration, if having platform-specific dependencies or dependencies
+#   having no cross-platform support, pipenv may install dependencies that don't work, or not
+#   install all needed dependencies.
+# Pipfile.lock
+
+# UV
+#   Similar to Pipfile.lock, it is generally recommended to include uv.lock in version control.
+#   This is especially recommended for binary packages to ensure reproducibility, and is more
+#   commonly ignored for libraries.
+# uv.lock
+
+# poetry
+#   Similar to Pipfile.lock, it is generally recommended to include poetry.lock in version control.
+#   This is especially recommended for binary packages to ensure reproducibility, and is more
+#   commonly ignored for libraries.
+#   https://python-poetry.org/docs/basic-usage/#commit-your-poetrylock-file-to-version-control
+# poetry.lock
+# poetry.toml
+
+# pdm
+#   Similar to Pipfile.lock, it is generally recommended to include pdm.lock in version control.
+#   pdm recommends including project-wide configuration in pdm.toml, but excluding .pdm-python.
+#   https://pdm-project.org/en/latest/usage/project/#working-with-version-control
+# pdm.lock
+# pdm.toml
+.pdm-python
+.pdm-build/
+
+# pixi
+#   Similar to Pipfile.lock, it is generally recommended to include pixi.lock in version control.
+# pixi.lock
+#   Pixi creates a virtual environment in the .pixi directory, just like venv module creates one
+#   in the .venv directory. It is recommended not to include this directory in version control.
+.pixi
+
+# PEP 582; used by e.g. github.com/David-OConnor/pyflow and github.com/pdm-project/pdm
+__pypackages__/
+
+# Celery stuff
+celerybeat-schedule
+celerybeat.pid
+
+# Redis
+*.rdb
+*.aof
+*.pid
+
+# RabbitMQ
+mnesia/
+rabbitmq/
+rabbitmq-data/
+
+# ActiveMQ
+activemq-data/
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.envrc
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+# pytype static type analyzer
+.pytype/
+
+# Cython debug symbols
+cython_debug/
+
+# PyCharm
+#   JetBrains specific template is maintained in a separate JetBrains.gitignore that can
+#   be found at https://github.com/github/gitignore/blob/main/Global/JetBrains.gitignore
+#   and can be added to the global gitignore or merged into this file.  For a more nuclear
+#   option (not recommended) you can uncomment the following to ignore the entire idea folder.
+# .idea/
+
+# Abstra
+#   Abstra is an AI-powered process automation framework.
+#   Ignore directories containing user credentials, local state, and settings.
+#   Learn more at https://abstra.io/docs
+.abstra/
+
+# Visual Studio Code
+#   Visual Studio Code specific template is maintained in a separate VisualStudioCode.gitignore 
+#   that can be found at https://github.com/github/gitignore/blob/main/Global/VisualStudioCode.gitignore
+#   and can be added to the global gitignore or merged into this file. However, if you prefer, 
+#   you could uncomment the following to ignore the entire vscode folder
+# .vscode/
+
+# Ruff stuff:
+.ruff_cache/
+
+# PyPI configuration file
+.pypirc
+
+# Marimo
+marimo/_static/
+marimo/_lsp/
+__marimo__/
+
+# Streamlit
+.streamlit/secrets.toml
+
+# Node.js (client/)
+
+# Outbound sales example: eval run artifacts are data, not source
+test-runs/
+
+# Downloaded call recordings
+recordings/

outbound-sales/CLAUDE.md

@@ -0,0 +1,37 @@
+# CLAUDE.md - outbound-sales
+
+Hailey, an outbound sales voice agent built with Pipecat. The dialer reads `leads.csv`, server.py creates a Daily room with PSTN dial-out and starts a bot, and Hailey tries to collect the IT decision maker's contact info on the call.
+
+## Layout
+
+All code lives in `server/`. Run every command from there with `uv run`.
+
+- `bot.py`: Hailey's pipeline, tools (`save_contact_info`, `end_call`), dial-out retry logic, and the eval entry point
+- `server.py`: FastAPI webhook server (port 8080); `/dialout` starts calls, `/call_result` records outcomes, `/results` is polled by the dialer
+- `server_utils.py`: data models, Daily room creation, bot starting, `report_result`
+- `dialer.py`: batch dialer, 5 calls at a time
+- `scenarios/` + `evals.yaml`: text-mode behavioral evals
+
+## Dev loop
+
+Prefer evals over real calls. They run the same bot in text mode with no telephony:
+
+```bash
+uv run pipecat eval suite evals.yaml          # whole suite
+uv run bot.py -t eval                          # then: uv run pipecat eval run scenarios/happy_path.yaml -v
+```
+
+The bot exits when Hailey hangs up, so restart `bot.py -t eval` between single-scenario runs.
+
+Real calls need two terminals: `uv run server.py` (port 8080) and `uv run bot.py -t daily` (port 7860), plus a purchased Daily phone number and dial-out enablement.
+
+## Rules and gotchas
+
+- **This is a demo: results are NOT saved to files.** Call outcomes are logged to the terminal and held in server.py's memory (`CALL_RESULTS`). A real production app would write them to a database in `/call_result`. Do not add file or CSV persistence.
+- The outcome report from the bot doubles as the dialer's "call finished" signal. If you touch the shutdown path in `run_bot`, keep the `report_result` call in the `finally` block.
+- In `end_call`, `worker.flush_pipeline()` must run before pushing `EndWorkerFrame`. The eval websocket server closes as soon as the EndFrame passes the input transport, so anything still queued would be lost.
+- The first reply is canned (`CannedGreetingGate`), skipping the LLM round-trip. Eval runs push it as LLM response frames because text-mode evals never see TTS output.
+- Smart Turn's silence fallback is capped at 1s (`stop_secs=1.0`) on purpose; don't raise it back to the 3s default.
+- Calls are recorded with Daily cloud recording, started from the bot's meeting token. There is no local recording code.
+- **Before deploying to Pipecat Cloud**, change the fields in `pcc-deploy.toml`: `agent_name`, `image` (it points at the example author's Docker Hub repo), and `secret_set` are all account-specific.
+- Python deps come from `pyproject.toml` via `uv sync`; `pipecat-ai` installs from the GitHub `main` branch until 1.4.0 ships on PyPI.