Conversation Participants & Lenses

Invite one or more AI agents into a conversation "room", address a specific one with an @mention, import public LenserFight agents, and call a Lens inside a message.

Endpoints

Method	Path	Description	Auth
GET	/workspaces/:workspaceId/ai/sessions/:sessionId/participants	List invited agents	JWT + Entitlement
POST	/workspaces/:workspaceId/ai/sessions/:sessionId/participants	Invite an agent	JWT + Entitlement
DELETE	/workspaces/:workspaceId/ai/sessions/:sessionId/participants/:participantId	Remove an agent	JWT + Entitlement
GET	/workspaces/:workspaceId/ai/agents/discover?q=	Discover Chainabit + LenserFight agents	JWT + Entitlement

Invite an agent

Provide either a native agentId, or a lenserfightAgentId to mirror-import a public LenserFight agent and invite it in one call. mentionHandle is optional — it defaults to a slug of the agent's name.

The first agent invited becomes the default responder. Re-inviting a previously removed agent reactivates it.

Code Examples

cURL

curl -X POST https://api.chainabit.com/api/v1/workspaces/$WORKSPACE_ID/ai/sessions/$SESSION_ID/participants \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{ "agentId": "a1b2c3", "mentionHandle": "researcher" }'

LenserFight import

curl -X POST https://api.chainabit.com/api/v1/workspaces/$WORKSPACE_ID/ai/sessions/$SESSION_ID/participants \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{ "lenserfightAgentId": "lf_agent_123" }'

Address an agent with @mention

Once multiple agents are in the room, mention one to make only that agent answer:

@researcher summarize the latest findings

• A message addressed with @handle triggers only that agent.
• An unaddressed message goes to the default participant.
• Exactly one agent answers per message — there is no fan-out.

Discover & import LenserFight agents

Response

Returns { local: [...], lenserfight: [...] }. Invite a lenserfight result by passing its lenserfightAgentId to the invite endpoint above.

Code Example

curl "https://api.chainabit.com/api/v1/workspaces/$WORKSPACE_ID/ai/agents/discover?q=research" \
  -H "Authorization: Bearer $TOKEN"

Call a Lens in a message

Call a LenserFight Lens directly inside a message with a /lens command:

/lens summarize topic=AI tone=formal

• If the Lens has missing required parameters, the agent asks you for them. Reply with the value and the Lens runs automatically.
• Once complete, the resolved Lens template augments that single turn — Lenses are applied per message and are never added as a conversation participant.

You:   /lens summarize
Agent: Which topic would you like me to summarize?
You:   AI agents
Agent: <answers through the resolved Lens>

Supported tokens: version=<versionId> selects a specific Lens version; all other key=value pairs (quoted values allowed) are Lens parameters.

Human Participants

Add another workspace member to a session as a full read+write participant — distinct from inviting an AI agent above, and from a public share link, which is read-only and unauthenticated.

Endpoints

Method	Path	Description	Auth
POST	/workspaces/:workspaceId/ai/sessions/:sessionId/participants/humans	Invite a human participant	JWT + Entitlement

Invite a human participant

Provide the chainerId of an existing member of the session's workspace. There is no accept step — the target becomes a full participant immediately, unlike the Chainy Collaborators invite flow. The target must be an active member of the workspace the session belongs to; otherwise the request fails.

Request

Field	Type	Required	Description
chainerId	string	Yes	ID of the workspace member to add as a participant

Code Examples

curl -X POST https://api.chainabit.com/api/v1/workspaces/$WORKSPACE_ID/ai/sessions/$SESSION_ID/participants/humans \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{ "chainerId": "b2c3d4e5-f6a7-8901-bcde-f12345678901" }'

Notes

• All endpoints require membership in the target workspace.
• Only public LenserFight agents can be imported.
• Running a LenserFight workflow from chat is approval-gated because it spends external credits.