Update an agent
Updates the specified agent. All request fields are optional; omitted fields are left unchanged.
Merge semantics follow PATCH conventions:
- Top-level scalars (
name,description) merge field-by-field — sending one does not affect the other. - Nested objects (
runtime,permissions,tools,subagents,skills,extras) are replaced in full when any of their fields is provided. To preserve existing nested values, include them in your PATCH. - Providing any of
system_prompt,tools,subagents,skills, orfilestriggers a new file tree commit.instructionsis a deprecated alias forsystem_promptand is still accepted;system_promptwins when both are sent.
Entries under files are created or replaced by path. The call is
atomic per request: if any entry fails validation, no changes are
applied.
Documentation Index
Fetch the complete documentation index at: https://langchain-5e9cc07a-preview-cbclea-1779438149-9d7c578.mintlify.app/llms.txt
Use this file to discover all available pages before exploring further.
Authorizations
Path Parameters
Agent ID
Query Parameters
Echo the agent's file tree in the response
Body
Caller-defined key/value tracking merged into the agent's runtime configuration. Typed runtime fields win on key collision.
Paths to remove from the file tree. Relative paths only.
Caller-defined key/value tracking attached to the agent's metadata.
Raw file entries created or replaced by path. Triggers a new commit.
Deprecated: use system_prompt. Accepted for backwards compatibility;
system_prompt takes precedence when both are set.
Skills. Each entry is written to skills/<name>/SKILL.md and supporting files. Triggers a new commit.
Subagents. Each entry is written to subagents/<name>/AGENTS.md and subagents/<name>/tools.json. Triggers a new commit.
Agent system prompt. Written to AGENTS.md. Triggers a new commit.
"You are a helpful agent. Use the available tools to answer the user's question."
Tool config. Written to tools.json. Triggers a new commit.
Response
OK
READ, RUN, WRITE Raw file map. Returned only when include_files=true.
Deprecated: use system_prompt. Echoed alongside system_prompt with
the same value for backwards compatibility.
Skills parsed from skills/<name>/SKILL.md and supporting files.
Subagents parsed from subagents/<name>/AGENTS.md + subagents/<name>/tools.json.
Agent system prompt parsed from AGENTS.md.
Tool config parsed from tools.json.

