Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.litellm-agent-platform.ai/llms.txt

Use this file to discover all available pages before exploring further.

A skill is a named block of Markdown instructions that gets appended to an agent’s system prompt. Skills are stored in a library and can be reused across agents. When a session starts, every attached skill is materialized as ~/.claude/skills/<slug>/SKILL.md inside the sandbox so the agent can read it directly.

​
Create a skill

Send a POST to /api/v1/skills with a name, optional description, and content (the Markdown body): 
curl -X POST https://your-lap-deployment/api/v1/skills \
  -H "Authorization: Bearer $MASTER_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "PR reviewer",
    "description": "Guidelines for reviewing pull requests",
    "content": "## PR review checklist\n\n- Check for test coverage\n- Verify no secrets are committed\n- Confirm the PR description explains the why"
  }'
The response is an ApiSkill object: 
{
  "id": "sk_01abc...",
  "name": "PR reviewer",
  "description": "Guidelines for reviewing pull requests",
  "content": "## PR review checklist\n\n...",
  "created_at": "2024-01-15T10:00:00.000Z"
}
Save the id — you’ll use it to attach the skill to agents.

​
Skill fields

FieldTypeRequiredDescription
namestringYesDisplay name shown in the library
descriptionstringNoShort summary of what the skill does
contentstringYesMarkdown instructions appended to the agent’s system prompt

​
List skills

curl https://your-lap-deployment/api/v1/skills \
  -H "Authorization: Bearer $MASTER_KEY"
Returns { "data": [...] } with all skills you own, ordered newest first.

​
Attach a skill to an agent

​
At agent create time

Pass a skill_ids array when creating an agent. Every ID in the list must exist in your library: 
curl -X POST https://your-lap-deployment/api/v1/managed_agents/agents \
  -H "Authorization: Bearer $MASTER_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "pr-review-bot",
    "model": "anthropic/claude-sonnet-4-6",
    "harness_id": "claude-agent-sdk",
    "prompt": "You are a senior engineer.",
    "skill_ids": ["sk_01abc..."]
  }'

​
To an existing agent

Use POST /api/v1/managed_agents/agents/{id}/skill. You can reference a library skill by ID or provide the content inline. By library ID: 
curl -X POST https://your-lap-deployment/api/v1/managed_agents/agents/{agent_id}/skill \
  -H "Authorization: Bearer $MASTER_KEY" \
  -H "Content-Type: application/json" \
  -d '{ "skill_id": "sk_01abc..." }'
Inline (saves to library automatically): 
curl -X POST https://your-lap-deployment/api/v1/managed_agents/agents/{agent_id}/skill \
  -H "Authorization: Bearer $MASTER_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "Security scanner",
    "content": "## Security guidelines\n\nNever log secrets. Sanitize all inputs."
  }'
Inline content is saved to the skill library by default so it can be reattached to other agents. To skip saving, pass "save_to_library": false — the skill will only exist on this agent’s prompt. Multiple skills can be stacked on one agent. Each is appended as a distinct block, ordered by attach time.

​
Detach a skill

Remove a specific skill from an agent’s prompt: 
curl -X DELETE \
  "https://your-lap-deployment/api/v1/managed_agents/agents/{agent_id}/skill?skill_id=sk_01abc..." \
  -H "Authorization: Bearer $MASTER_KEY"
To detach all skills at once, omit the skill_id query parameter: 
curl -X DELETE \
  "https://your-lap-deployment/api/v1/managed_agents/agents/{agent_id}/skill" \
  -H "Authorization: Bearer $MASTER_KEY"

​
Update a skill

curl -X PATCH https://your-lap-deployment/api/v1/skills/{skill_id} \
  -H "Authorization: Bearer $MASTER_KEY" \
  -H "Content-Type: application/json" \
  -d '{ "content": "Updated checklist content here." }'
You can update name, description, and content independently — only the fields you send are changed.
Updating a skill in the library does not automatically update agents that already have it attached. Detach and reattach the skill to pick up the new content.

​
Delete a skill

curl -X DELETE https://your-lap-deployment/api/v1/skills/{skill_id} \
  -H "Authorization: Bearer $MASTER_KEY"
Returns 204 No Content on success. Deleting a library skill does not remove its content from agents that already have it attached — the Markdown is embedded in the agent’s prompt at attach time.

​
Example: “PR reviewer” end to end

BASE="https://your-lap-deployment"
KEY="$MASTER_KEY"

# 1. Create the skill
SKILL_ID=$(curl -s -X POST "$BASE/api/v1/skills" \
  -H "Authorization: Bearer $KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "PR reviewer",
    "description": "Code review guidelines",
    "content": "## PR review checklist\n\n- Confirm test coverage\n- Check for hardcoded secrets\n- Verify the PR description explains the change"
  }' | jq -r '.id')

# 2. Create an agent with the skill attached
AGENT_ID=$(curl -s -X POST "$BASE/api/v1/managed_agents/agents" \
  -H "Authorization: Bearer $KEY" \
  -H "Content-Type: application/json" \
  -d "{
    \"name\": \"pr-review-bot\",
    \"model\": \"anthropic/claude-sonnet-4-6\",
    \"harness_id\": \"claude-agent-sdk\",
    \"prompt\": \"You are a senior engineer.\",
    \"skill_ids\": [\"$SKILL_ID\"]
  }" | jq -r '.id')

echo "Agent $AGENT_ID created with skill $SKILL_ID"