Files

T

History

Lee moon soo a8b030b786 fix(runner): keep native harness alive during terminal activity (#5212 )

Signed-off-by: Lee moon soo <moonsoo.lee@databricks.com>

2026-08-21 12:35:22 -07:00

credentials

fix(credentials): stop mislabeling OAuth Databricks profiles as malformed (#3059 )

2026-07-22 16:21:27 +08:00

harnesses

fix(runner): keep native harness alive during terminal activity (#5212 )

2026-08-21 12:35:22 -07:00

policies

perf(policies): load the conversation and spawn tree once per engine build (#4320 )

2026-08-07 15:57:31 +09:00

__init__.py

perf(terminal): attach web terminals over loopback when the runner is local (#4763 )

2026-08-13 16:42:54 -07:00

_globals.py

perf(terminal): attach web terminals over loopback when the runner is local (#4763 )

2026-08-13 16:42:54 -07:00

agent_cache.py

fix(spec): gracefully drop unsupported sub-agents on the execution path (#1145 )

2026-06-24 18:29:53 +00:00

caps.py

Smart Routing MVP: per-task model and harness routing (#4074 )

2026-08-05 15:34:54 -07:00

compaction.py

Fix claude native replay for screenshots from MCP results (#4659 )

2026-08-12 14:24:13 -07:00

content_resolver.py

fix(policies): scan text attachments for PII at the request gate (#2927 )

2026-07-27 10:19:38 +00:00

filesystem_registry.py

⚡ perf(runner): Defer untracked cache setup (#2976 )

2026-07-21 15:32:01 +08:00

inflight_text.py

fix(claude-native): supersede stranded in-flight text on later commits (#4656 )

2026-08-12 15:32:35 +08:00

llm_retry.py

fix: sum input+max_tokens for Anthropic context overflow actual_tokens (#33 )

2026-06-14 19:06:03 +00:00

pending_elicitations.py

Add cross-replica live-state mirror for the session sidebar (#2574 )

2026-07-16 12:01:13 -07:00

pending_inputs.py

feat(kiro): add native CLI harness (#899 )

2026-06-25 00:55:25 +00:00

prompt.py

Fix claude native replay for screenshots from MCP results (#4659 )

2026-08-12 14:24:13 -07:00

README.md

Welcome to omnigent

2026-06-13 00:37:34 +00:00

session_stream.py

refactor(native): simplify streamed text reconciliation (#4593 )

2026-08-11 19:14:35 +08:00

subagent_block_notifier.py

Welcome to omnigent

2026-06-13 00:37:34 +00:00

telemetry.py

refactor(telemetry): type OTLP exporters (#3719 )

2026-07-31 13:15:13 +00:00

tool_output.py

Welcome to omnigent

2026-06-13 00:37:34 +00:00

tool_result_replay.py

Fix claude native replay for screenshots from MCP results (#4659 )

2026-08-12 14:24:13 -07:00

tool_retry.py

Welcome to omnigent

2026-06-13 00:37:34 +00:00

user_session_stream.py

Welcome to omnigent

2026-06-13 00:37:34 +00:00

workflow.py

fix(openai-agents): preserve reasoning item IDs by default (#5065 )

2026-08-20 14:57:57 -07:00

README.md

runtime

The execution engine — how an agent runs. Given a spec and user input, it drives the agent's reasoning loop: invoking LLMs, calling tools, managing skills, and producing responses.

The runtime is a library, not a service. The server is its primary host, but it can also be used directly for local development, embedded in other applications, or invoked from tests.