Agent

Assemble the agent boot context pack

Assembles the identity + resume + facts + lessons + (optional) goal-graph brief + procedural skill suggestions + (optional) repo map into one bounded, token-budgeted response. The primary entry point of the agent memory layer — see agent_boot in CLAUDE.md.

post/api/agent/boot

Request

  • Base URL: http://localhost:38200
  • URL: http://localhost:38200/api/agent/boot
  • Auth: one of:
    • HTTP bearer
    • API key in cookie token

Request body

agentstring required
agent_namestring
goalstring
workspacestring
token_budgetinteger
session_idinteger
project_idinteger

When present

include_globalboolean

Also read the GLOBAL workspace (slug=global) and union the results. Still tenant-filtered.

include_repo_mapboolean

Assemble a token-budgeted repo map from code-indexed contexts and include it as repo_map.

repo_map_token_budgetinteger

Token cap for the repo_map slot (default ~2000). Ignored unless include_repo_map=true.

epistemic_min'observed' | 'told' | 'inferred' | 'assumed'

Epistemic floor for the facts slot; omitted = no floor.

fullboolean

Bypass the skills diff-since-last-boot behavior and return the full current suggestion set.

Response

Boot context pack.

resumeobject nullable
handoffobject nullable
factsobject[]
briefobject nullable

Goal-graph situation brief; present only when project_id was supplied.

skillsobject[]
skills_fullboolean
pushobject nullable
repo_mapobject nullable

Changes

No recorded changes to this operation across all 1 revision of this API.