buffer

name	buffer
description	This skill should be used when interfacing with the Buffer social media scheduling API. It handles scheduling social media posts, checking the queue, listing channels, creating ideas, and managing Buffer accounts.
allowed-tools	Bash(curl:*), Bash(cat:*), Bash(jq:*)

Buffer Social Media Scheduling

Setup

Requires a BUFFER_API_TOKEN environment variable. Generate one at: https://publish.buffer.com/settings/api

Verify the token is set before making any API calls:

if [ -z "$BUFFER_API_TOKEN" ]; then
  echo "Error: BUFFER_API_TOKEN is not set. Get your token at https://publish.buffer.com/settings/api"
  exit 1
fi

API Configuration

• Endpoint: https://api.buffer.com
• Method: POST (all requests are GraphQL)
• Content-Type: application/json
• Auth: Authorization: Bearer $BUFFER_API_TOKEN
• User-Agent: Mozilla/5.0 — required to avoid Cloudflare 403 blocks

Curl Template

All API calls follow this pattern:

curl -s -X POST https://api.buffer.com \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $BUFFER_API_TOKEN" \
  -H "User-Agent: Mozilla/5.0" \
  -d '{"query": "<GRAPHQL_QUERY>", "variables": <VARIABLES_JSON>}'

Always pipe through jq for readable output. Use jq -e to detect errors.

Important: For queries with GraphQL variables (e.g. $input), the shell may expand $input even inside single quotes. Use a temp file with -d @file to avoid this:

cat > /tmp/buffer_payload.json << 'EOF'
{"query": "<GRAPHQL_QUERY>", "variables": <VARIABLES_JSON>}
EOF

curl -s -X POST https://api.buffer.com \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $BUFFER_API_TOKEN" \
  -H "User-Agent: Mozilla/5.0" \
  -d @/tmp/buffer_payload.json | jq .

Simple queries without variables can use inline -d directly.

Mode Detection

Parse the user's intent to determine which operation to perform:

Intent	Mode	Example
List organizations	organizations	"Show my Buffer organizations"
List channels	channels	"What channels do I have?"
View posts	posts	"Show my scheduled posts"
Create a text post	create-post	"Schedule a tweet saying..."
Create a Twitter/X thread	create-thread	"Post a thread on X..."
Create an image post	create-post	"Post this image to Instagram"
Save an idea	create-idea	"Save an idea about..."
Account info	account	"Show my Buffer account"

Most operations require an organizationId. If the user hasn't specified one, fetch organizations first. If only one org exists, use it automatically.

Mode: Get Organizations

Fetch the user's organizations. This is typically the first call needed.

curl -s -X POST https://api.buffer.com \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $BUFFER_API_TOKEN" \
  -H "User-Agent: Mozilla/5.0" \
  -d '{"query": "{ account { organizations { id name ownerEmail } } }"}' | jq .

Store the organizationId from the response for subsequent calls.

Mode: Get Channels

List channels (social media accounts) for an organization.

cat > /tmp/buffer_payload.json << 'EOF'
{"query": "query GetChannels($input: ChannelsInput!) { channels(input: $input) { id name displayName service avatar isQueuePaused } }", "variables": {"input": {"organizationId": "<ORG_ID>"}}}
EOF
curl -s -X POST https://api.buffer.com \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $BUFFER_API_TOKEN" \
  -H "User-Agent: Mozilla/5.0" \
  -d @/tmp/buffer_payload.json | jq .

Display results in a readable format showing channel name, service (e.g., twitter, instagram, linkedin), and queue status.

Mode: Get Posts

Query posts with filtering, sorting, and pagination.

cat > /tmp/buffer_payload.json << 'EOF'
{"query": "query GetPosts($input: PostsInput!) { posts(input: $input) { edges { node { id text status dueAt sentAt createdAt channelService externalLink } cursor } pageInfo { hasNextPage endCursor } totalCount } }", "variables": {"input": {"organizationId": "<ORG_ID>", "filter": {"status": "scheduled"}, "sort": {"field": "dueAt", "direction": "desc"}}}}
EOF
curl -s -X POST https://api.buffer.com \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $BUFFER_API_TOKEN" \
  -H "User-Agent: Mozilla/5.0" \
  -d @/tmp/buffer_payload.json | jq .

For available filter fields (status, channelIds, date range), sort options, and pagination details, refer to references/api-reference.md.

Mode: Create Text Post

Create and schedule a text post.

cat > /tmp/buffer_payload.json << 'EOF'
{"query": "mutation CreatePost($input: CreatePostInput!) { createPost(input: $input) { ... on PostActionSuccess { post { id text status dueAt } } ... on MutationError { message } } }", "variables": {"input": {"channelId": "<CHANNEL_ID>", "text": "<POST_CONTENT>", "schedulingType": "automatic", "mode": "addToQueue"}}}
EOF
curl -s -X POST https://api.buffer.com \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $BUFFER_API_TOKEN" \
  -H "User-Agent: Mozilla/5.0" \
  -d @/tmp/buffer_payload.json | jq .

Required Fields

• channelId (String!): The channel to post to
• text (String!): The post content

Scheduling Options

• schedulingType: automatic (Buffer picks time) or notification (sends reminder)
• mode: Controls when the post is published (e.g., addToQueue, shareNow, shareNext, customSchedule, recommendedTime)
• dueAt (String): ISO 8601 datetime, required when mode is customSchedule. Example: "2026-03-15T14:00:00Z"

For full field definitions and enum values, refer to references/api-reference.md.

Response Handling

The mutation returns a union type. Always check for both success and error:

# Check if the response contains an error
result=$(curl -s -X POST https://api.buffer.com -H "User-Agent: Mozilla/5.0" ...)
echo "$result" | jq -e '.data.createPost.message' > /dev/null 2>&1 && {
  echo "Error: $(echo "$result" | jq -r '.data.createPost.message')"
} || {
  echo "$result" | jq '.data.createPost.post'
}

Mode: Create Twitter/X Thread

Twitter/X threads are created using the metadata.twitter.thread field on the createPost mutation. The thread is an array of ThreadedPostInput objects, each with a text field (and optional assets).

Important: The first tweet in the thread must be included in the thread array. The top-level text field is also required but Buffer uses the thread array to render all tweets in its UI. Include the same text in both places for the first tweet.

cat > /tmp/buffer_payload.json << 'EOF'
{"query": "mutation CreatePost($input: CreatePostInput!) { createPost(input: $input) { ... on PostActionSuccess { post { id text status dueAt } } ... on MutationError { message } } }", "variables": {"input": {"channelId": "<CHANNEL_ID>", "text": "<FIRST_TWEET_TEXT>", "schedulingType": "automatic", "mode": "addToQueue", "metadata": {"twitter": {"thread": [{"text": "<FIRST_TWEET_TEXT>"}, {"text": "<SECOND_TWEET_TEXT>"}, {"text": "<THIRD_TWEET_TEXT>"}]}}}}}
EOF
curl -s -X POST https://api.buffer.com \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $BUFFER_API_TOKEN" \
  -H "User-Agent: Mozilla/5.0" \
  -d @/tmp/buffer_payload.json | jq .

Thread Structure

• Top-level text: Required. Use the first tweet's text.
• metadata.twitter.thread: Array of ThreadedPostInput objects. Each has:
  • text (String): The tweet content
  • assets (AssetsInput, optional): Images for that specific tweet
• The first entry in the thread array should match the top-level text (this is what Buffer displays as tweet 1).
• All subsequent entries become reply tweets in the thread.

Thread Tips

• Each tweet in the thread must respect Twitter's character limit (280 characters).
• Threads are scheduled as a single unit and posted all at once.
• Images can be attached to individual tweets via the assets field on each ThreadedPostInput.

Mode: Create Image Post

Same as text post, but include an assets field with image URLs.

cat > /tmp/buffer_payload.json << 'EOF'
{"query": "mutation CreatePost($input: CreatePostInput!) { createPost(input: $input) { ... on PostActionSuccess { post { id text status dueAt } } ... on MutationError { message } } }", "variables": {"input": {"channelId": "<CHANNEL_ID>", "text": "<POST_CONTENT>", "schedulingType": "automatic", "mode": "addToQueue", "assets": {"images": [{"url": "<IMAGE_URL>"}]}}}}
EOF
curl -s -X POST https://api.buffer.com \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $BUFFER_API_TOKEN" \
  -H "User-Agent: Mozilla/5.0" \
  -d @/tmp/buffer_payload.json | jq .

Images must be publicly accessible URLs. Multiple images can be included in the images array.

Mode: Create Idea

Save an idea for later use.

cat > /tmp/buffer_payload.json << 'EOF'
{"query": "mutation CreateIdea($input: CreateIdeaInput!) { createIdea(input: $input) { ... on Idea { id content { title text services } } } }", "variables": {"input": {"organizationId": "<ORG_ID>", "content": {"title": "<IDEA_TITLE>", "text": "<IDEA_BODY>", "services": ["twitter", "linkedin"]}}}}
EOF
curl -s -X POST https://api.buffer.com \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $BUFFER_API_TOKEN" \
  -H "User-Agent: Mozilla/5.0" \
  -d @/tmp/buffer_payload.json | jq .

For IdeaContentInput field details (title, text, services, tags), refer to references/api-reference.md.

Note: Tags require existing tag objects with id and color fields — fetch existing tags first before using.

Mode: Get Account

Note: The full account query (id, name, email, timezone) fails with a standard API token scope. Use Get Organizations instead to confirm account identity.

curl -s -X POST https://api.buffer.com \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $BUFFER_API_TOKEN" \
  -H "User-Agent: Mozilla/5.0" \
  -d '{"query": "{ account { organizations { id name ownerEmail } } }"}' | jq .

Common Workflows

Schedule a Post

Multi-step workflow when the user wants to schedule a post:

1. Get organizations → extract organizationId
2. Get channels → show available channels, let user pick (or match by service name)
3. Create post → use selected channelId with the post content

If the user specifies a service (e.g., "post to Twitter"), match it against the channel's service field.

Check the Queue

1. Get organizations → extract organizationId
2. Get posts with filter: { status: "scheduled" } and sort: { field: "dueAt", direction: "asc" }
3. Display posts grouped by date with channel info

Brainstorm and Save Ideas

1. Get organizations → extract organizationId
2. Help the user draft idea content
3. Create idea with title, text, tags, and target services

Schema Introspection

Before attempting a mutation or querying a field you're unsure about, introspect the schema first. Don't guess — the API surface is limited.

List all available mutations:

curl -s -X POST https://api.buffer.com \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $BUFFER_API_TOKEN" \
  -H "User-Agent: Mozilla/5.0" \
  -d '{"query": "{ __schema { mutationType { fields { name } } } }"}' | jq '[.data.__schema.mutationType.fields[].name]'

List all available queries:

curl -s -X POST https://api.buffer.com \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $BUFFER_API_TOKEN" \
  -H "User-Agent: Mozilla/5.0" \
  -d '{"query": "{ __schema { queryType { fields { name } } } }"}' | jq '[.data.__schema.queryType.fields[].name]'

Inspect fields on a specific type:

curl -s -X POST https://api.buffer.com \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $BUFFER_API_TOKEN" \
  -H "User-Agent: Mozilla/5.0" \
  -d '{"query": "{ __type(name: \"<TYPE_NAME>\") { fields { name type { name kind } } } }"}' | jq '.data.__type.fields[]'

Error Handling

GraphQL Errors

Check for the top-level errors array in every response:

result=$(curl -s -X POST https://api.buffer.com ...)
errors=$(echo "$result" | jq -r '.errors // empty')
if [ -n "$errors" ]; then
  echo "GraphQL Error:"
  echo "$result" | jq '.errors[].message'
  exit 1
fi

Mutation Errors

Mutations return union types. Always handle the MutationError variant:

{
  "data": {
    "createPost": {
      "message": "Validation failed: text is required"
    }
  }
}

Common Issues

• 403 Forbidden (error code 1010): Cloudflare is blocking the request. Always include -H "User-Agent: Mozilla/5.0" in every curl call.
• 401 Unauthorized: Token is invalid or expired. Regenerate at https://publish.buffer.com/settings/api
• Missing organizationId: Most queries require an org ID. Fetch organizations first.
• Invalid channelId: Verify the channel exists and belongs to the current organization.
• Past dueAt: When using customSchedule, the dueAt must be in the future.