Skip to content

synth-ai 0.11.2

Choose a tag to compare

View all tags
@JoshuaPurtell JoshuaPurtell released this 16 Jun 20:43
v0.11.2
38ce95c
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

0.11.2 — 2026-06-16

Added

  • SynthClient().research — public Managed Research + Research Factory namespace (ResearchClient).
  • MCP entrypoint synth-ai-managed-research-mcp (stdio) alongside hosted HTTP MCP at https://api.usesynth.ai/mcp.
  • Optional extra synth-ai[research] for Research/MCP dependencies.

Changed

  • Canonical install: pip install "synth-ai[research]==0.11.2" (or uv add "synth-ai[research]").
  • Canonical Python entry: from synth_ai import SynthClient → client.research.
  • MCP tool enumeration and SDK namespaces reconciled to the shipped backend OpenAPI contract.

Removed (breaking vs 0.11.1)

The following public SDK/MCP surfaces were removed because they are not part of the launch contract. Use the replacements below.

Removed area Replacement
Objective / milestone / experiment CRUD helpers Objectives are runtime-managed during runs; inspect with smr_get_run_work_graph, smr_objectives (read), run observability
Project economics convenience wrappers on deprecated paths smr_get_project_usage, smr_get_run_usage, resource-limit progress tools, billing entitlements
Connect-git / legacy GitHub setup SDK modules GitHub OAuth via MCP smr_setup_github_* where enabled; attach repos via console or smr_work_repos_*
Standalone managed-research PyPI package as the primary story synth-ai[research] only; legacy ManagedResearchClient import remains under synth_ai for advanced/low-level callers
Data-factory trigger/publish MCP tools (smr_trigger_data_factory, …) Factory programs via hosted control plane + documented factory MCP tools

Migration (0.11.1 → 0.11.2)

uv add "synth-ai[research]==0.11.2"
export SYNTH_API_KEY="sk_..."
# Before (0.11.1 docs / standalone package)
# from managed_research import ManagedResearchClient
# client = ManagedResearchClient()

# After
from synth_ai import SynthClient

research = SynthClient().research
projects = research.projects.list()
run = research.runs.start(
    "Inspect the repo and leave evidence.",
    host_kind="daytona",
    work_mode="directed_effort",
    providers=[{"provider": "openrouter"}],
    runbook="lite",
)
  • MCP stdio: uv tool install "synth-ai[research]" then run synth-ai-managed-research-mcp.
  • Errors: launch denials return typed exceptions / MCP tool errors with 402 (insufficient credits) and 429 (limits). See Preflight and Errors.
Assets 2
Loading