Common Flows - Devin Docs

This page walks through common end-to-end workflows with the Devin API. Each flow includes the full sequence of API calls with code examples. For individual endpoint details, see the relevant API reference pages.

Setup

Set these environment variables before running any example:

# Required: your service user API key (starts with cog_)
export DEVIN_API_KEY="cog_your_key_here"

# Required: your organization ID (shown at the top of Settings > Devin API)
export DEVIN_ORG_ID="your_org_id"

Getting started: API key to first session

The most common workflow — authenticate, discover your account, and create your first session.

Step 1: Verify your credentials

Organization-scoped service users can skip to Step 3 — you already know your org ID from the Settings page.

curl "https://api.devin.ai/v3/self" \
  -H "Authorization: Bearer $DEVIN_API_KEY"

Response:

{
  "principal_type": "service_user",
  "service_user_id": "service-user-abc123",
  "service_user_name": "CI Bot",
  "org_id": null
}

Step 2: List your organizations

Enterprise service users can list all organizations:

curl "https://api.devin.ai/v3/enterprise/organizations" \
  -H "Authorization: Bearer $DEVIN_API_KEY"

Organization-scoped service users already know their org ID (it’s shown at the top of the Settings > Devin API page, where the service user was provisioned).

Step 3: Create a session

curl -X POST "https://api.devin.ai/v3/organizations/$DEVIN_ORG_ID/sessions" \
  -H "Authorization: Bearer $DEVIN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"prompt": "Create a Python script that analyzes CSV data"}'

Response:

{
  "session_id": "devin-abc123",
  "url": "https://app.devin.ai/sessions/devin-abc123",
  "status": "running"
}

Step 4: Poll for events

Monitor the session by polling for messages:

curl "https://api.devin.ai/v3/organizations/$DEVIN_ORG_ID/sessions/devin-abc123/messages" \
  -H "Authorization: Bearer $DEVIN_API_KEY"

Full Python example

import os
import time
import requests

API_KEY = os.environ["DEVIN_API_KEY"]
ORG_ID = os.environ["DEVIN_ORG_ID"]
BASE = "https://api.devin.ai/v3"
HEADERS = {"Authorization": f"Bearer {API_KEY}"}

# 1. Verify credentials
me = requests.get(f"{BASE}/self", headers=HEADERS)
me.raise_for_status()
data = me.json()
print(f"Authenticated as: {data.get('service_user_name') or data.get('user_name')}")

# 2. Create a session
session = requests.post(
    f"{BASE}/organizations/{ORG_ID}/sessions",
    headers={**HEADERS, "Content-Type": "application/json"},
    json={"prompt": "Create a Python script that analyzes CSV data"}
).json()
print(f"Session: {session['url']}")

# 3. Poll until complete
while True:
    status = requests.get(
        f"{BASE}/organizations/{ORG_ID}/sessions/{session['session_id']}",
        headers=HEADERS
    ).json()["status"]
    print(f"Status: {status}")
    if status in ("exit", "error", "suspended"):
        break
    time.sleep(10)

Downloading session attachments

Retrieve files produced by a session (logs, screenshots, generated code, etc.).

Step 1: Get the session

curl "https://api.devin.ai/v3/organizations/$DEVIN_ORG_ID/sessions/$SESSION_ID" \
  -H "Authorization: Bearer $DEVIN_API_KEY"

Step 2: List attachments

curl "https://api.devin.ai/v3/organizations/$DEVIN_ORG_ID/sessions/$SESSION_ID/attachments" \
  -H "Authorization: Bearer $DEVIN_API_KEY"

Response:

{
  "items": [
    {
      "attachment_id": "att_123",
      "name": "output.py",
      "url": "https://..."
    }
  ]
}

Step 3: Download

Use the url from the attachment response to download the file directly.

Full Python example

import os
import requests

API_KEY = os.environ["DEVIN_API_KEY"]
ORG_ID = os.environ["DEVIN_ORG_ID"]
SESSION_ID = os.environ["SESSION_ID"]
BASE = "https://api.devin.ai/v3"
HEADERS = {"Authorization": f"Bearer {API_KEY}"}

# List attachments
attachments = requests.get(
    f"{BASE}/organizations/{ORG_ID}/sessions/{SESSION_ID}/attachments",
    headers=HEADERS
).json()["items"]

# Download each attachment
for att in attachments:
    print(f"Downloading {att['name']}...")
    content = requests.get(att["url"]).content
    with open(att["name"], "wb") as f:
        f.write(content)
    print(f"  Saved {att['name']} ({len(content)} bytes)")

Knowledge & playbook management

Manage the context and instructions that Devin uses across sessions.

Create a knowledge note

curl -X POST "https://api.devin.ai/v3/organizations/$DEVIN_ORG_ID/knowledge/notes" \
  -H "Authorization: Bearer $DEVIN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "Coding standards",
    "trigger": "When writing code in any repository",
    "body": "Use TypeScript strict mode. Follow existing code style. Run lint before committing."
  }'

List knowledge notes

curl "https://api.devin.ai/v3/organizations/$DEVIN_ORG_ID/knowledge/notes" \
  -H "Authorization: Bearer $DEVIN_API_KEY"

Update a knowledge note

curl -X PUT "https://api.devin.ai/v3/organizations/$DEVIN_ORG_ID/knowledge/notes/$NOTE_ID" \
  -H "Authorization: Bearer $DEVIN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "Coding standards (updated)",
    "trigger": "When writing code in any repository",
    "body": "Use TypeScript strict mode. Follow existing code style. Run lint and type-check before committing."
  }'

Delete a knowledge note

curl -X DELETE "https://api.devin.ai/v3/organizations/$DEVIN_ORG_ID/knowledge/notes/$NOTE_ID" \
  -H "Authorization: Bearer $DEVIN_API_KEY"

Create a playbook

curl -X POST "https://api.devin.ai/v3/organizations/$DEVIN_ORG_ID/playbooks" \
  -H "Authorization: Bearer $DEVIN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "title": "PR Review",
    "body": "Review the PR for bugs, security issues, and style violations. Leave inline comments."
  }'

Full Python example

import os
import requests

API_KEY = os.environ["DEVIN_API_KEY"]
ORG_ID = os.environ["DEVIN_ORG_ID"]
BASE = "https://api.devin.ai/v3"
HEADERS = {
    "Authorization": f"Bearer {API_KEY}",
    "Content-Type": "application/json"
}

# Create knowledge note
note = requests.post(
    f"{BASE}/organizations/{ORG_ID}/knowledge/notes",
    headers=HEADERS,
    json={
        "name": "Coding standards",
        "trigger": "When writing code in any repository",
        "body": "Use TypeScript strict mode. Follow existing code style."
    }
).json()
print(f"Created note: {note['note_id']}")

# List all notes
notes = requests.get(
    f"{BASE}/organizations/{ORG_ID}/knowledge/notes",
    headers={"Authorization": f"Bearer {API_KEY}"}
).json()["items"]
print(f"Total notes: {len(notes)}")

# Create playbook
playbook = requests.post(
    f"{BASE}/organizations/{ORG_ID}/playbooks",
    headers=HEADERS,
    json={
        "title": "PR Review",
        "body": "Review the PR for bugs, security issues, and style violations."
    }
).json()
print(f"Created playbook: {playbook['playbook_id']}")

Scheduling automated sessions

Create recurring sessions with an automation that has a schedule:recurring trigger. The trigger takes one rrule condition: an iCalendar RRULE evaluated in UTC (for example FREQ=WEEKLY;BYDAY=MO,TU,WE,TH,FR;BYHOUR=13;BYMINUTE=0 for weekdays at 9 AM Eastern during daylight time). Requires the ManageOrgAutomations permission.

The older /schedules endpoints are deprecated. Starting September 24, 2026, POST /v3/organizations/{org_id}/schedules returns 403 for organizations migrated to Automations. See the API release notes.

Create a scheduled automation

curl -X POST "https://api.devin.ai/v3/organizations/$DEVIN_ORG_ID/automations" \
  -H "Authorization: Bearer $DEVIN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "Weekday test run",
    "run_as": {"type": "organization"},
    "triggers": [{
      "event_type": "schedule:recurring",
      "conditions": {"any": [{"all": [{
        "field": "rrule",
        "operator": "recurrence",
        "value": "FREQ=WEEKLY;BYDAY=MO,TU,WE,TH,FR;BYHOUR=13;BYMINUTE=0"
      }]}]}
    }],
    "actions": [{
      "type": "start_session",
      "prompt": "Run the test suite and report any failures"
    }]
  }'

For a one-time run, use a DTSTART with COUNT=1, for example DTSTART:20260915T170000Z\nRRULE:FREQ=DAILY;COUNT=1. The automation fires once and then disables itself.

List, update, and delete

# List automations
curl "https://api.devin.ai/v3/organizations/$DEVIN_ORG_ID/automations" \
  -H "Authorization: Bearer $DEVIN_API_KEY"

# Disable an automation
curl -X PATCH "https://api.devin.ai/v3/organizations/$DEVIN_ORG_ID/automations/$AUTOMATION_ID" \
  -H "Authorization: Bearer $DEVIN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"enabled": false}'

# Delete an automation
curl -X DELETE "https://api.devin.ai/v3/organizations/$DEVIN_ORG_ID/automations/$AUTOMATION_ID" \
  -H "Authorization: Bearer $DEVIN_API_KEY"

Full Python example

import os
import requests

API_KEY = os.environ["DEVIN_API_KEY"]
ORG_ID = os.environ["DEVIN_ORG_ID"]
BASE = "https://api.devin.ai/v3"
HEADERS = {
    "Authorization": f"Bearer {API_KEY}",
    "Content-Type": "application/json"
}

# Create a daily health check automation (09:00 UTC every day)
automation = requests.post(
    f"{BASE}/organizations/{ORG_ID}/automations",
    headers=HEADERS,
    json={
        "name": "Daily health check",
        "run_as": {"type": "organization"},
        "triggers": [{
            "event_type": "schedule:recurring",
            "conditions": {"any": [{"all": [{
                "field": "rrule",
                "operator": "recurrence",
                "value": "FREQ=DAILY;BYHOUR=9;BYMINUTE=0"
            }]}]}
        }],
        "actions": [{
            "type": "start_session",
            "prompt": "Run the test suite and report any failures"
        }]
    }
).json()
print(f"Created automation: {automation['automation_id']}")

# List all automations (the endpoint returns up to 100 per page)
automations = []
after = None
while True:
    page = requests.get(
        f"{BASE}/organizations/{ORG_ID}/automations",
        headers={"Authorization": f"Bearer {API_KEY}"},
        params={"after": after},
    ).json()
    automations.extend(page["items"])
    if not page["has_next_page"]:
        break
    after = page["end_cursor"]

for a in automations:
    status = "enabled" if a.get("enabled") else "disabled"
    print(f"  {a['automation_id']}: {a['name']} ({status})")

Hand off a task from anywhere

Because the Sessions API creates a cloud Devin session from a single request, any tool, script, or coding agent can “hand off” work to Devin — bundling the current repo, branch, and uncommitted changes into the prompt so the cloud session picks up where you left off.

Create a session with repo context

# Build the JSON body with jq so the raw diff — which contains quotes,
# backslashes, and newlines — is escaped into a valid JSON string.
curl -X POST "https://api.devin.ai/v3/organizations/$DEVIN_ORG_ID/sessions" \
  -H "Authorization: Bearer $DEVIN_API_KEY" \
  -H "Content-Type: application/json" \
  -d "$(jq -n --arg diff "$(git diff HEAD)" \
    '{prompt: "Repo: my-org/my-repo (branch: fix-flaky-tests)\nFix the flaky integration tests in CI.\n\nUncommitted changes:\n\($diff)"}')"

The cloud session clones the repo, applies the context from your prompt, and runs in its own VM with a shell, browser, and full repo access. Track it by polling for messages or in the Devin web app.

git diff HEAD can include uncommitted secrets — API keys, tokens, or .env edits — and the prompt is uploaded to the cloud session. Review your diff and commit, stash, or remove sensitive changes before handing off.

Don’t want to build this yourself? The open-source Devin Handoff plugin wraps exactly this flow — auto-detecting repo, branch, and diff — so you can hand off from the Devin CLI, Claude Code, Codex, Cursor, or a plain shell script. See Hand off to Devin.

Error handling

All examples above should include error handling in production. Here’s a reusable pattern:

import requests

def api_request(method, url, headers, **kwargs):
    """Make an API request with standard error handling."""
    response = requests.request(method, url, headers=headers, **kwargs)
    try:
        response.raise_for_status()
        return response.json()
    except requests.exceptions.HTTPError as e:
        status = e.response.status_code
        if status == 401:
            raise Exception("Invalid or expired API key")
        elif status == 403:
            raise Exception("Service user lacks required permission")
        elif status == 404:
            raise Exception("Resource not found")
        elif status == 429:
            raise Exception("Rate limit exceeded — wait and retry")
        else:
            raise Exception(f"API error {status}: {e.response.text}")

Support

Need Help?

For questions about the API or to report issues, email support@cognition.ai

​Setup

​Getting started: API key to first session

​Step 1: Verify your credentials

​Step 2: List your organizations

​Step 3: Create a session

​Step 4: Poll for events

​Full Python example

​Downloading session attachments

​Step 1: Get the session

​Step 2: List attachments

​Step 3: Download

​Full Python example

​Knowledge & playbook management

​Create a knowledge note

​List knowledge notes

​Update a knowledge note

​Delete a knowledge note

​Create a playbook

​Full Python example

​Scheduling automated sessions

​Create a scheduled automation

​List, update, and delete

​Full Python example

​Hand off a task from anywhere

​Create a session with repo context

​Error handling

​Support

Need Help?

Setup

Getting started: API key to first session

Step 1: Verify your credentials

Step 2: List your organizations

Step 3: Create a session

Step 4: Poll for events

Full Python example

Downloading session attachments

Step 1: Get the session

Step 2: List attachments

Step 3: Download

Full Python example

Knowledge & playbook management

Create a knowledge note

List knowledge notes

Update a knowledge note

Delete a knowledge note

Create a playbook

Full Python example

Scheduling automated sessions

Create a scheduled automation

List, update, and delete

Full Python example

Hand off a task from anywhere

Create a session with repo context

Error handling

Support