getcompanion-ai/co-mono
1
0
Fork 0
mirror of https://github.com/getcompanion-ai/co-mono.git synced 2026-04-17 05:00:16 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

docs(agent): update steering and options

Browse source
This commit is contained in:
Mario Zechner 2026-01-18 19:21:52 +01:00
parent ad337e30be
commit c63f33d83b
1 changed files with 54 additions and 8 deletions
Download patch file Download diff file Expand all files Collapse all files

62
packages/agent/README.md
View file

@ -144,14 +144,28 @@ const agent = new Agent({
// Transform context before convertToLlm (for pruning, compaction) // Transform context before convertToLlm (for pruning, compaction)
transformContext: async (messages, signal) => pruneOldMessages(messages), transformContext: async (messages, signal) => pruneOldMessages(messages),
// How to handle queued messages: "one-at-a-time" (default) or "all" // Steering mode: "one-at-a-time" (default) or "all"
queueMode: "one-at-a-time", steeringMode: "one-at-a-time",
// Follow-up mode: "one-at-a-time" (default) or "all"
followUpMode: "one-at-a-time",
// Custom stream function (for proxy backends) // Custom stream function (for proxy backends)
streamFn: streamProxy, streamFn: streamProxy,
// Session ID for provider caching
sessionId: "session-123",
// Dynamic API key resolution (for expiring OAuth tokens) // Dynamic API key resolution (for expiring OAuth tokens)
getApiKey: async (provider) => refreshToken(), getApiKey: async (provider) => refreshToken(),
// Custom thinking budgets for token-based providers
thinkingBudgets: {
minimal: 128,
low: 512,
medium: 1024,
high: 2048,
},
}); });
``` ```
@ -206,6 +220,19 @@ agent.clearMessages();
agent.reset(); // Clear everything agent.reset(); // Clear everything
``` ```
### Session and Thinking Budgets
```typescript
agent.sessionId = "session-123";
agent.thinkingBudgets = {
minimal: 128,
low: 512,
medium: 1024,
high: 2048,
};
```
### Control ### Control
```typescript ```typescript
@ -222,26 +249,45 @@ const unsubscribe = agent.subscribe((event) => {
unsubscribe(); unsubscribe();
``` ```
## Message Queue ## Steering and Follow-up
Queue messages to inject during tool execution (for user interruptions): Steering messages let you interrupt the agent while tools are running. Follow-up messages let you queue work after the agent would otherwise stop.
```typescript ```typescript
agent.setQueueMode("one-at-a-time"); agent.setSteeringMode("one-at-a-time");
agent.setFollowUpMode("one-at-a-time");
// While agent is running tools // While agent is running tools
agent.queueMessage({ agent.steer({
role: "user", role: "user",
content: "Stop! Do this instead.", content: "Stop! Do this instead.",
timestamp: Date.now(), timestamp: Date.now(),
}); });
// After the agent finishes its current work
agent.followUp({
role: "user",
content: "Also summarize the result.",
timestamp: Date.now(),
});
const steeringMode = agent.getSteeringMode();
const followUpMode = agent.getFollowUpMode();
agent.clearSteeringQueue();
agent.clearFollowUpQueue();
agent.clearAllQueues();
``` ```
When queued messages are detected after a tool completes: Use clearSteeringQueue, clearFollowUpQueue, or clearAllQueues to drop queued messages.
When steering messages are detected after a tool completes:
1. Remaining tools are skipped with error results 1. Remaining tools are skipped with error results
2. Queued message is injected 2. Steering messages are injected
3. LLM responds to the interruption 3. LLM responds to the interruption
Follow-up messages are checked only when there are no more tool calls and no steering messages. If any are queued, they are injected and another turn runs.
## Custom Message Types ## Custom Message Types
Extend `AgentMessage` via declaration merging: Extend `AgentMessage` via declaration merging: