Each operation below carries the questions people ask an LLM about it and the instructions they give an agent to run it. Generated by API Evangelistoverlay
GET/agent-os/deployments/{deployment_id}/memoryList a deployment's governed memory items · List deployment-scoped governed memory#
Ask an LLM
“What memories has my deployed agent stored so far?”
“Can I list only memory of a certain type or under a path prefix?”
Tell an agent
List the governed memory items for deployment {deployment_id}.
Show {status} memory items of type {type} on branch {branch} for deployment {deployment_id}.
POST/agent-os/deployments/{deployment_id}/memoryCreate a versioned memory item · Create versioned governed memory#
Ask an LLM
“How do I save a new fact into my agent's memory with an initial commit?”
“Can I write a memory item directly at a semantic path on a branch?”
Tell an agent
Create a {type} memory at {path} in deployment {deployment_id} with content {content}.
Commit a new memory item summarized as {summary} on branch {branch} of deployment {deployment_id}.
POST/agent-os/deployments/{deployment_id}/memory/candidatesPropose a memory candidate for review · Create a governed memory candidate#
Ask an LLM
“Can my agent propose a memory that a human approves before it is used?”
“Which kinds of memory stay approval-gated instead of auto-written?”
Tell an agent
Propose a memory candidate for deployment {deployment_id} with summary {summary}.
Submit a {sensitivity} {type} memory candidate for review in deployment {deployment_id}, citing {source_refs}.
GET/agent-os/deployments/{deployment_id}/memory/branchesList a deployment's memory branches · List memory branches#
Ask an LLM
“Which memory branches exist for my deployment and what are their head commits?”
“Can I see only active memory branches?”
Tell an agent
List memory branches for deployment {deployment_id}.
Show up to {limit} memory branches with status {status} in deployment {deployment_id}.
POST/agent-os/deployments/{deployment_id}/memory/branchesCreate a memory branch for isolation · Create memory branch#
Ask an LLM
“Can I branch my agent's memory for an experiment without touching the main line?”
“How do I make a separate memory branch for a codebase worktree?”
Tell an agent
Create memory branch {branch} in deployment {deployment_id}.
Branch {branch} off {parent_branch} with scope {scope} for deployment {deployment_id}.
GET/agent-os/deployments/{deployment_id}/memory/commitsList memory commit history · List memory commits#
Ask an LLM
“What is the commit history of my agent's memory?”
“Can I see every commit that touched one memory item?”
Tell an agent
List memory commits for deployment {deployment_id}.
Show the commits for memory item {memory_id} on branch {branch} in deployment {deployment_id}.
GET/agent-os/deployments/{deployment_id}/memory/commits/{commit_id}Read one memory commit and its snapshot · Get memory commit#
Ask an LLM
“What exactly changed in a specific memory commit?”
“Can I read the stored snapshot attached to a single commit?”
Tell an agent
Get memory commit {commit_id} in deployment {deployment_id}.
Show the snapshot stored with commit {commit_id} for deployment {deployment_id}.
POST/agent-os/deployments/{deployment_id}/memory/checkoutCheck out a read-only memory snapshot · Checkout memory snapshot#
Ask an LLM
“Can I view my agent's memory as it was at an earlier commit without changing the branch?”
“Does checking out an old memory commit modify current memory?”
Tell an agent
Check out memory as of commit {commit_id} for deployment {deployment_id}.
Give me a read-only view of deployment {deployment_id} memory at commit {commit_id}.
POST/agent-os/deployments/{deployment_id}/memory/revertRevert a memory commit · Revert memory commit#
Ask an LLM
“How do I undo a bad memory change my agent committed?”
“Does reverting a memory commit erase the audit history?”
Tell an agent
Revert memory commit {commit_id} in deployment {deployment_id}.
Roll back commit {commit_id} on deployment {deployment_id} because {reason}.
GET/agent-os/deployments/{deployment_id}/memory/blameFind who last changed a memory path · Blame memory path#
Ask an LLM
“Which commit and actor last wrote a particular memory path?”
“Where did the information at this memory path come from?”
Tell an agent
Blame memory path {path} in deployment {deployment_id}.
Show the source refs and author of {path} on branch {branch} for deployment {deployment_id}.
GET/agent-os/deployments/{deployment_id}/memory/diffDiff two memory commits · Diff memory commits#
Ask an LLM
“What is different between two versions of my agent's memory?”
“Can I compare a commit against a chosen base commit?”
Tell an agent
Diff memory commit {target_commit_id} in deployment {deployment_id}.
Compare commit {target_commit_id} against {base_commit_id} for deployment {deployment_id}.
POST/agent-os/deployments/{deployment_id}/memory/searchSearch a deployment's approved memory · Search approved deployment memory#
Ask an LLM
“Can my agent search only approved memory and skip unreviewed candidates?”
“Is it possible to exclude memory older than a certain number of days from a search?”
Tell an agent
Search deployment {deployment_id} memory for {query}.
Find approved memories about {query} from the last {max_age_days} days in deployment {deployment_id}.
POST/agent-os/deployments/{deployment_id}/memory/reconcileTurn reconciliation output into a memory proposal · Create post-action memory from reconciliation#
Ask an LLM
“Can my agent learn from the gap between what it expected and what actually happened?”
“Does a reconciliation memory proposal change pricing or trust automatically?”
Tell an agent
Create a post-action memory proposal for deployment {deployment_id} from reconciliation {reconciliation}.
Record the actual outcome {actual_outcome} versus the pre-action result {pre_action_result} as a memory candidate on deployment {deployment_id}.
GET/agent-os/deployments/{deployment_id}/memory/policyRead a deployment's memory policy · Read deployment memory policy#
Ask an LLM
“Which memory types are allowed, blocked or auto-written for my deployment?”
“Is cross-agent memory sharing turned on for this deployment?”
Tell an agent
Show the memory policy for deployment {deployment_id}.
Tell me the receipt-evidence rules in deployment {deployment_id}'s memory policy.
PATCH/agent-os/deployments/{deployment_id}/memory/policyUpdate a deployment's memory policy · Update deployment memory policy#
Ask an LLM
“How do I change which memory types my agent may auto-write?”
“Can I enable public or cross-agent sharing of deployment memory?”
Tell an agent
Update the memory policy on deployment {deployment_id}.
Change deployment {deployment_id}'s memory policy to block pricing memory.
POST/agent-os/deployments/{deployment_id}/memory/{memory_id}/approveApprove a memory candidate · Approve a memory candidate#
Ask an LLM
“How do I approve a memory my agent proposed so it can be retrieved?”
“Why would approving a provider-trust memory candidate fail?”
Tell an agent
Approve memory candidate {memory_id} in deployment {deployment_id}.
Accept the proposed memory {memory_id} for deployment {deployment_id} into approved memory.
POST/agent-os/deployments/{deployment_id}/memory/{memory_id}/rejectReject a memory candidate · Reject a memory candidate#
Ask an LLM
“Can I turn down a memory my agent proposed so it's never used?”
“What happens to a rejected memory candidate in future searches?”
Tell an agent
Reject memory candidate {memory_id} in deployment {deployment_id}.
Decline proposed memory {memory_id} on deployment {deployment_id}.
DELETE/agent-os/deployments/{deployment_id}/memory/{memory_id}Delete or redact a memory item · Delete or redact a memory item#
Ask an LLM
“How do I remove a memory item that should never have been stored?”
“Can I redact only certain fields of a memory while keeping the audit trail?”
Tell an agentdestructive · confirm first
Delete memory item {memory_id} from deployment {deployment_id}.
Redact fields {fields} of memory {memory_id} in deployment {deployment_id} because {reason}.
Calls without X-API-Key need no key. Calls with it need a paid plan — get a key.
Get an API key
Free tier, no form to fill in. Signing in shares your email address with us — we
store it to create your key and to recognise you if you sign in with another
provider. See our Privacy Policy and
Terms.