content large_stringlengths 3 20.5k | url large_stringlengths 53 192 ⌀ | branch large_stringclasses 4
values | source large_stringclasses 51
values | embeddings listlengths 384 384 | score float64 -0.21 0.65 |
|---|---|---|---|---|---|
# Usage The Agents SDK automatically tracks token usage for every run. You can access it from the run context and use it to monitor costs, enforce limits, or record analytics. ## What is tracked - \*\*requests\*\*: number of LLM API calls made - \*\*input\_tokens\*\*: total input tokens sent - \*\*output\_tokens\*\*: t... | https://github.com/openai/openai-agents-python/blob/main/docs/usage.md | main | openai-agents | [
0.006352993194013834,
0.008699042722582817,
-0.09363964945077896,
0.06534525007009506,
0.016937164589762688,
-0.07336359471082687,
0.0675513744354248,
0.05519678443670273,
0.04504784941673279,
0.02216021530330181,
-0.05565856024622917,
-0.11537636071443558,
0.025125376880168915,
-0.0298680... | 0.159664 |
# Quickstart ## Create a project and virtual environment You'll only need to do this once. ```bash mkdir my\_project cd my\_project python -m venv .venv ``` ### Activate the virtual environment Do this every time you start a new terminal session. On macOS or Linux: ```bash source .venv/bin/activate ``` On Windows: ```c... | https://github.com/openai/openai-agents-python/blob/main/docs/quickstart.md | main | openai-agents | [
0.045765507966279984,
-0.05516710877418518,
-0.10666318237781525,
-0.046390969306230545,
0.0471041277050972,
-0.004988941829651594,
0.009592940099537373,
0.07244452834129333,
0.03815080597996712,
0.0407959520816803,
0.022062180563807487,
-0.05668669939041138,
0.06018516793847084,
0.0709864... | 0.002379 |
agent for math questions", instructions="You explain math step by step and include worked examples.", ) ``` ## Define your handoffs On an agent, you can define an inventory of outgoing handoff options that it can choose from while solving the task. ```python triage\_agent = Agent( name="Triage Agent", instructions="Rou... | https://github.com/openai/openai-agents-python/blob/main/docs/quickstart.md | main | openai-agents | [
-0.056462232023477554,
0.04006600007414818,
-0.10706181079149246,
0.05500137433409691,
-0.08192071318626404,
-0.056600771844387054,
-0.023676730692386627,
0.05692244693636894,
-0.09907156974077225,
-0.05424707755446434,
-0.0009228395647369325,
-0.03462538123130798,
0.003973111510276794,
0.... | 0.088701 |
# Handoffs Handoffs allow an agent to delegate tasks to another agent. This is particularly useful in scenarios where different agents specialize in distinct areas. For example, a customer support app might have agents that each specifically handle tasks like order status, refunds, FAQs, etc. Handoffs are represented a... | https://github.com/openai/openai-agents-python/blob/main/docs/handoffs.md | main | openai-agents | [
-0.0909150093793869,
-0.031978242099285126,
-0.03272769972681999,
0.07912160456180573,
-0.03382987529039383,
-0.013685553334653378,
0.04761238768696785,
0.007768602576106787,
0.03309984877705574,
-0.014093424193561077,
0.0043170335702598095,
0.04982192814350128,
-0.006122915539890528,
0.02... | 0.088987 |
with reason: {input\_data.reason}") agent = Agent(name="Escalation agent") handoff\_obj = handoff( agent=agent, on\_handoff=on\_handoff, input\_type=EscalationData, ) ``` `input\_type` describes the arguments for the handoff tool call itself. The SDK exposes that schema to the model as the handoff tool's `parameters`, ... | https://github.com/openai/openai-agents-python/blob/main/docs/handoffs.md | main | openai-agents | [
-0.04902251437306404,
0.012719828635454178,
-0.0063070314936339855,
0.026512393727898598,
0.023090187460184097,
-0.046455029398202896,
0.04938521981239319,
0.01927901804447174,
0.05023624375462532,
-0.0773608461022377,
0.00802350603044033,
-0.03082359954714775,
-0.04079494625329971,
0.0270... | 0.125617 |
both the handoff and the active [`RunConfig.handoff\_input\_filter`][agents.run.RunConfig.handoff\_input\_filter] define a filter, the per-handoff [`input\_filter`][agents.handoffs.Handoff.input\_filter] takes precedence for that specific handoff. !!! note Handoffs stay within a single run. Input guardrails still apply... | https://github.com/openai/openai-agents-python/blob/main/docs/handoffs.md | main | openai-agents | [
-0.028211915865540504,
0.04517893120646477,
-0.024572601541876793,
0.0022391409147530794,
0.017149826511740685,
-0.03929300606250763,
0.06273575872182846,
-0.02375633269548416,
0.017387142404913902,
-0.07067327201366425,
0.009444375522434711,
0.029058421030640602,
0.021590478718280792,
-0.... | 0.023704 |
# Streaming Streaming lets you subscribe to updates of the agent run as it proceeds. This can be useful for showing the end-user progress updates and partial responses. To stream, you can call [`Runner.run\_streamed()`][agents.run.Runner.run\_streamed], which will give you a [`RunResultStreaming`][agents.result.RunResu... | https://github.com/openai/openai-agents-python/blob/main/docs/streaming.md | main | openai-agents | [
-0.03368927538394928,
-0.014010405167937279,
-0.09237625449895859,
0.02239937335252762,
0.02029244601726532,
-0.021430138498544693,
0.004978630226105452,
-0.0022141060326248407,
0.016505485400557518,
-0.0009547812514938414,
-0.049041252583265305,
-0.01295494195073843,
-0.042743731290102005,
... | 0.101508 |
higher level events. They inform you when an item has been fully generated. This allows you to push progress updates at the level of "message generated", "tool ran", etc, instead of each token. Similarly, [`AgentUpdatedStreamEvent`][agents.stream\_events.AgentUpdatedStreamEvent] gives you updates when the current agent... | https://github.com/openai/openai-agents-python/blob/main/docs/streaming.md | main | openai-agents | [
-0.057004474103450775,
-0.03173544630408287,
-0.04069133102893829,
0.06229688972234726,
0.049955982714891434,
-0.04764678329229355,
0.000982882222160697,
0.018560683354735374,
0.11455424129962921,
0.013108954764902592,
-0.014727914705872536,
-0.002861842280253768,
-0.035982079803943634,
-0... | 0.202447 |
# Running agents You can run agents via the [`Runner`][agents.run.Runner] class. You have 3 options: 1. [`Runner.run()`][agents.run.Runner.run], which runs async and returns a [`RunResult`][agents.result.RunResult]. 2. [`Runner.run\_sync()`][agents.run.Runner.run\_sync], which is a sync method and just runs `.run()` un... | https://github.com/openai/openai-agents-python/blob/main/docs/running_agents.md | main | openai-agents | [
-0.01114762481302023,
-0.0880575031042099,
-0.105443075299263,
-0.010232185944914818,
0.0023954608477652073,
-0.023015346378087997,
-0.04709898307919502,
-0.021329771727323532,
0.0034367945045232773,
-0.050457023084163666,
-0.025492044165730476,
0.009725448675453663,
0.015110386535525322,
... | 0.102053 |
instructions="Be concise.") async with responses\_websocket\_session( responses\_websocket\_options={"ping\_interval": 20.0, "ping\_timeout": 60.0}, ) as ws: first = ws.run\_streamed(agent, "Say hello in one short sentence.") async for \_event in first.stream\_events(): pass second = ws.run\_streamed( agent, "Now say g... | https://github.com/openai/openai-agents-python/blob/main/docs/running_agents.md | main | openai-agents | [
-0.029827214777469635,
0.03157433494925499,
-0.06237661466002464,
0.08138804137706757,
-0.013009706512093544,
-0.0758177638053894,
0.011462114751338959,
-0.006228509824723005,
0.0040968721732497215,
-0.05904047191143036,
-0.032373346388339996,
-0.024647805839776993,
-0.05474657937884331,
-... | 0.093709 |
across multiple runs. - [`trace\_metadata`][agents.run.RunConfig.trace\_metadata]: Metadata to include on all traces. ##### Tool execution, approval, and tool error behavior - [`tool\_execution`][agents.run.RunConfig.tool\_execution]: Configure SDK-side execution behavior for local tool calls, such as limiting how many... | https://github.com/openai/openai-agents-python/blob/main/docs/running_agents.md | main | openai-agents | [
-0.03065570257604122,
-0.004600216634571552,
-0.007343471981585026,
0.05676833167672157,
0.041323862969875336,
-0.03866979479789734,
-0.01611749641597271,
0.0061215609312057495,
0.04978703707456589,
-0.01468653604388237,
-0.018461380153894424,
-0.04397138953208923,
0.006496733520179987,
-0... | 0.010973 |
but strips the reasoning item `id`, which avoids triggering that API invariant in SDK-generated follow-up inputs. Scope notes: - This only changes reasoning items generated/forwarded by the SDK when it builds follow-up input. - It does not rewrite user-supplied initial input items. - `call\_model\_input\_filter` can st... | https://github.com/openai/openai-agents-python/blob/main/docs/running_agents.md | main | openai-agents | [
-0.06458991765975952,
0.03604399785399437,
-0.035672109574079514,
0.061386194080114365,
0.021827593445777893,
0.0077279433608055115,
0.11381708085536957,
0.008860999718308449,
0.05205784738063812,
-0.0010672260541468859,
-0.030923619866371155,
-0.025966210290789604,
0.0177152156829834,
-0.... | 0.163374 |
First turn result = await Runner.run(agent, "What city is the Golden Gate Bridge in?", session=session) print(result.final\_output) # San Francisco # Second turn - agent automatically remembers previous context result = await Runner.run(agent, "What state is it in?", session=session) print(result.final\_output) # Calif... | https://github.com/openai/openai-agents-python/blob/main/docs/running_agents.md | main | openai-agents | [
0.028356419876217842,
0.009695482440292835,
-0.02494031749665737,
0.08519340306520462,
-0.014727823436260223,
-0.007852677255868912,
0.04091902822256088,
-0.04625144600868225,
0.059310514479875565,
-0.03803441673517227,
-0.07558712363243103,
-0.0007987641729414463,
-0.023924831300973892,
0... | 0.09171 |
# Keep only the last 5 items and preserve existing instructions. trimmed = data.model\_data.input[-5:] return ModelInputData(input=trimmed, instructions=data.model\_data.instructions) agent = Agent(name="Assistant", instructions="Answer concisely.") result = Runner.run\_sync( agent, "Explain quines", run\_config=RunCon... | https://github.com/openai/openai-agents-python/blob/main/docs/running_agents.md | main | openai-agents | [
-0.02496510185301304,
0.06502281874418259,
0.03180248290300369,
0.02187821827828884,
-0.04757130146026611,
-0.025802170857787132,
0.06179623305797577,
0.04258288815617561,
-0.02001943066716194,
-0.0007153898477554321,
0.026357538998126984,
0.033927008509635925,
-0.013079123571515083,
-0.04... | 0.053253 |
integration requires only a SQLite or Postgres database. View the integration [repo](https://github.com/dbos-inc/dbos-openai-agents) and the [docs](https://docs.dbos.dev/integrations/openai-agents) for more details. ## Exceptions The SDK raises exceptions in certain cases. The full list is in [`agents.exceptions`][]. A... | https://github.com/openai/openai-agents-python/blob/main/docs/running_agents.md | main | openai-agents | [
0.03887604549527168,
-0.0237258430570364,
-0.08112867176532745,
-0.005096824374049902,
0.02404717728495598,
-0.020497391000390053,
0.026269370689988136,
0.043880395591259,
-0.03307506814599037,
0.006816408131271601,
0.023740645498037338,
-0.004371616058051586,
0.050833169370889664,
-0.0567... | 0.142303 |
# Configuration This page covers SDK-wide defaults that you usually set once during application startup, such as the default OpenAI key or client, the default OpenAI API shape, tracing export defaults, and logging behavior. These defaults still apply to sandbox-based workflows, but sandbox workspaces, sandbox clients, ... | https://github.com/openai/openai-agents-python/blob/main/docs/config.md | main | openai-agents | [
0.027462974190711975,
-0.04257188364863396,
-0.08362365514039993,
0.007470843847841024,
0.0382082536816597,
-0.021669235080480576,
0.02807624265551567,
-0.001112352591007948,
0.022335438057780266,
-0.03443092480301857,
0.01618630439043045,
-0.009930701926350594,
0.014839998446404934,
-0.02... | 0.177159 |
run\_config=RunConfig(trace\_include\_sensitive\_data=False), ) ``` You can also change the default without code by setting this environment variable before your app starts: ```bash export OPENAI\_AGENTS\_TRACE\_INCLUDE\_SENSITIVE\_DATA=0 ``` For full tracing controls, see the [tracing guide](tracing.md). ## Debug logg... | https://github.com/openai/openai-agents-python/blob/main/docs/config.md | main | openai-agents | [
0.09932953119277954,
-0.02491641603410244,
-0.08079086244106293,
0.06451047211885452,
0.07946053892374039,
-0.12196429073810577,
0.02259979024529457,
-0.003268955508247018,
0.03571024909615517,
-0.0024750400334596634,
0.052407752722501755,
-0.044272445142269135,
0.03570760414004326,
0.0602... | 0.075547 |
# Release process/changelog The project follows a slightly modified version of semantic versioning using the form `0.Y.Z`. The leading `0` indicates the SDK is still evolving rapidly. Increment the components as follows: ## Minor (`Y`) versions We will increase minor versions `Y` for \*\*breaking changes\*\* to any pub... | https://github.com/openai/openai-agents-python/blob/main/docs/release.md | main | openai-agents | [
0.004922463092952967,
0.03539624810218811,
0.08820141106843948,
-0.04649090766906738,
0.13255177438259125,
-0.007466678973287344,
-0.02356601320207119,
-0.046634092926979065,
-0.048260610550642014,
0.018531762063503265,
0.027310922741889954,
0.0011997695546597242,
-0.058975979685783386,
-0... | 0.091616 |
```python result = Runner.run\_sync( agent, input, error\_handlers={"model\_refusal": lambda data: data.error.refusal}, ) ``` For structured-output agents, the handler can return a value matching the agent's output schema, and the SDK will validate it like other run error handler final outputs. ### 0.14.0 This minor re... | https://github.com/openai/openai-agents-python/blob/main/docs/release.md | main | openai-agents | [
-0.010080637410283089,
-0.012008675374090672,
-0.023983143270015717,
0.03421830013394356,
0.04547335207462311,
-0.05219271406531334,
-0.08259554207324982,
-0.01399222295731306,
-0.03281905874609947,
-0.02984704077243805,
0.016173643991351128,
-0.04663839563727379,
0.06729177385568619,
-0.0... | 0.120981 |
from `Tool` to `FunctionTool`. This change should not usually cause breaking issues, but if your code relies on the broader union type, you may need to make some adjustments on your side. ### 0.8.0 In this version, two runtime behavior changes may require migration work: - Function tools wrapping \*\*synchronous\*\* Py... | https://github.com/openai/openai-agents-python/blob/main/docs/release.md | main | openai-agents | [
-0.050318643450737,
-0.046276673674583435,
0.03912973403930664,
0.0812763199210167,
-0.05297337472438812,
-0.10760458558797836,
-0.08352838456630707,
-0.004888579249382019,
-0.024418413639068604,
-0.039383478462696075,
-0.019132932648062706,
-0.04104228317737579,
-0.019638339057564735,
-0.... | -0.016255 |
# Quickstart !!! warning "Beta feature" Sandbox agents are in beta. Expect details of the API, defaults, and supported capabilities to change before general availability, and expect more advanced features over time. Modern agents work best when they can operate on real files in a filesystem. \*\*Sandbox Agents\*\* in t... | https://github.com/openai/openai-agents-python/blob/main/docs/sandbox_agents.md | main | openai-agents | [
-0.0013189627788960934,
-0.024418465793132782,
-0.042663849890232086,
0.02698478102684021,
0.04298470914363861,
-0.011441227979958057,
-0.09648334234952927,
0.019873570650815964,
-0.03374539688229561,
0.004483992233872414,
0.03562355786561966,
0.0025687487795948982,
0.050486013293266296,
-... | 0.121165 |
sandbox client choice, or sandbox-session resume behavior are part of the design. | https://github.com/openai/openai-agents-python/blob/main/docs/sandbox_agents.md | main | openai-agents | [
-0.060189761221408844,
0.010593163780868053,
-0.0020926662255078554,
-0.0069209011271595955,
0.06268356740474701,
0.05749383196234703,
-0.016170237213373184,
-0.03679570183157921,
0.0011776058236137033,
-0.02738766185939312,
-0.04871378466486931,
0.11879594624042511,
0.001581101561896503,
... | 0.133378 |
# Results When you call the `Runner.run` methods, you receive one of two result types: - [`RunResult`][agents.result.RunResult] from `Runner.run(...)` or `Runner.run\_sync(...)` - [`RunResultStreaming`][agents.result.RunResultStreaming] from `Runner.run\_streamed(...)` Both inherit from [`RunResultBase`][agents.result.... | https://github.com/openai/openai-agents-python/blob/main/docs/results.md | main | openai-agents | [
-0.02317347377538681,
-0.022477520629763603,
-0.08800994604825974,
0.07400224357843399,
0.023256808519363403,
-0.005754624493420124,
-0.041775893419981,
0.013393213972449303,
-0.010034752078354359,
-0.004682997707277536,
-0.012682801112532616,
-0.007152554579079151,
-0.025773392990231514,
... | 0.105881 |
Unlike the JavaScript SDK, Python does not expose a separate `output` property for the model-shaped delta only. Use `new\_items` when you need SDK metadata, or inspect `raw\_responses` when you need the raw model payloads. Computer-tool replay follows the raw Responses payload shape. Preview-model `computer\_call` item... | https://github.com/openai/openai-agents-python/blob/main/docs/results.md | main | openai-agents | [
-0.10015098005533218,
-0.052523545920848846,
-0.05865921452641487,
0.03158325329422951,
-0.025039376690983772,
0.0019509319681674242,
0.0099667152389884,
-0.011453229933977127,
0.02353649213910103,
-0.03782057389616966,
-0.0549483522772789,
0.003381311194971204,
-0.014572897925972939,
-0.0... | 0.065667 |
[`RunResultStreaming`][agents.result.RunResultStreaming] inherits the same result surfaces above, but adds streaming-specific controls: - [`stream\_events()`][agents.result.RunResultStreaming.stream\_events] to consume semantic stream events - [`current\_agent`][agents.result.RunResultStreaming.current\_agent] to track... | https://github.com/openai/openai-agents-python/blob/main/docs/results.md | main | openai-agents | [
-0.018695276230573654,
-0.012930776923894882,
-0.04718250036239624,
0.03984207659959793,
0.05555032193660736,
-0.013219732791185379,
0.016758829355239868,
-0.013528449460864067,
0.0665358304977417,
0.021737512201070786,
-0.03539585322141647,
-0.004915095865726471,
-0.06801315397024155,
-0.... | 0.099644 |
# OpenAI Agents SDK The [OpenAI Agents SDK](https://github.com/openai/openai-agents-python) enables you to build agentic AI apps in a lightweight, easy-to-use package with very few abstractions. It's a production-ready upgrade of our previous experimentation for agents, [Swarm](https://github.com/openai/swarm/tree/main... | https://github.com/openai/openai-agents-python/blob/main/docs/index.md | main | openai-agents | [
-0.024967657402157784,
-0.0324515700340271,
-0.06958295404911041,
0.019490079954266548,
0.07729653269052505,
-0.0883282721042633,
-0.019101954996585846,
-0.0023673339746892452,
0.004604883026331663,
0.050354644656181335,
-0.029794424772262573,
-0.0725754052400589,
0.048819441348314285,
0.0... | 0.265097 |
in programming.") print(result.final\_output) # Code within the code, # Functions calling themselves, # Infinite loop's dance. ``` (\_If running this, ensure you set the `OPENAI\_API\_KEY` environment variable\_) ```bash export OPENAI\_API\_KEY=sk-... ``` ## Start here - Build your first text-based agent with the [Quic... | https://github.com/openai/openai-agents-python/blob/main/docs/index.md | main | openai-agents | [
0.00608024513348937,
-0.02409612014889717,
-0.14309240877628326,
0.003933498170226812,
0.008041534572839737,
-0.04245614632964134,
0.0004649637849070132,
0.01538006030023098,
0.004985414445400238,
-0.005191073287278414,
-0.005090302787721157,
0.010089666582643986,
0.03843499720096588,
-0.0... | 0.122905 |
# Tracing The Agents SDK includes built-in tracing, collecting a comprehensive record of events during an agent run: LLM generations, tool calls, handoffs, guardrails, and even custom events that occur. Using the [Traces dashboard](https://platform.openai.com/traces), you can debug, visualize, and monitor your workflow... | https://github.com/openai/openai-agents-python/blob/main/docs/tracing.md | main | openai-agents | [
0.019490500912070274,
-0.01080929022282362,
-0.08537338674068451,
0.04139726981520653,
0.09703563153743744,
-0.07433729618787766,
0.011550666764378548,
-0.05135092884302139,
0.03206559643149376,
-0.005777894053608179,
0.024869833141565323,
0.015382999554276466,
0.01729300618171692,
-0.0304... | 0.126621 |
-> None: try: with trace("background\_job"): Runner.run\_sync(agent, prompt) finally: flush\_traces() @app.post("/run") async def run(prompt: str, background\_tasks: BackgroundTasks): background\_tasks.add\_task(process\_in\_background, prompt) return {"status": "queued"} ``` [`flush\_traces()`][agents.tracing.flush\_t... | https://github.com/openai/openai-agents-python/blob/main/docs/tracing.md | main | openai-agents | [
-0.021799292415380478,
-0.03803158551454544,
-0.08739941567182541,
0.023589102551341057,
0.014691363088786602,
-0.10477624088525772,
-0.018584106117486954,
-0.04285678640007973,
-0.005462566390633583,
-0.037123147398233414,
-0.021068457514047623,
0.005327697843313217,
0.00577252171933651,
... | 0.014065 |
You can use an OpenAI API key with non-OpenAI models to enable free tracing in the OpenAI Traces dashboard without needing to disable tracing. See the [Third-party adapters](models/index.md#third-party-adapters) section in the Models guide for adapter selection and setup caveats. ```python import os from agents import ... | https://github.com/openai/openai-agents-python/blob/main/docs/tracing.md | main | openai-agents | [
-0.003545150626450777,
-0.073471799492836,
-0.05179986730217934,
0.06351680308580399,
0.10449713468551636,
-0.045464012771844864,
-0.03304215893149376,
0.00854197796434164,
0.029764603823423386,
-0.0002618956787046045,
0.024094324558973312,
-0.0651063621044159,
0.03438840061426163,
-0.0188... | 0.02918 |
# Human-in-the-loop Use the human-in-the-loop (HITL) flow to pause agent execution until a person approves or rejects sensitive tool calls. Tools declare when they need approval, run results surface pending approvals as interruptions, and `RunState` lets you serialize and resume runs after decisions are made. That appr... | https://github.com/openai/openai-agents-python/blob/main/docs/human_in_the_loop.md | main | openai-agents | [
-0.08411708474159241,
-0.010712023824453354,
-0.039349742233753204,
0.06900519877672195,
-0.011893642134964466,
-0.07515248656272888,
0.008956637233495712,
-0.005702519323676825,
-0.014513415284454823,
0.0007358422735705972,
0.011842326261103153,
0.000028712414859910496,
-0.00092771701747551... | 0.058592 |
function tools, hosted MCP approvals, and nested `Agent.as\_tool()` approvals. If you rerun after approving or rejecting only some items, those resolved calls can continue while unresolved ones remain in `interruptions` and pause the run again. ## Custom rejection messages By default, a rejected tool call returns the S... | https://github.com/openai/openai-agents-python/blob/main/docs/human_in_the_loop.md | main | openai-agents | [
-0.04776785150170326,
0.03572225943207741,
0.05820273235440254,
0.08299005031585693,
0.04352935403585434,
-0.06626363098621368,
-0.021590974181890488,
0.0069921426475048065,
0.0656336098909378,
-0.03444661572575569,
0.024480724707245827,
-0.04452599212527275,
0.07001180946826935,
-0.064513... | 0.035357 |
await Runner.run(agent, state) print(result.final\_output) if \_\_name\_\_ == "\_\_main\_\_": asyncio.run(main()) ``` In this example, `prompt\_approval` is synchronous because it uses `input()` and is executed with `run\_in\_executor(...)`. If your approval source is already asynchronous (for example, an HTTP request ... | https://github.com/openai/openai-agents-python/blob/main/docs/human_in_the_loop.md | main | openai-agents | [
-0.02934313751757145,
0.014812403358519077,
-0.07874941825866699,
0.07921181619167328,
0.02368832752108574,
-0.10000844299793243,
-0.08090975880622864,
-0.014317895285785198,
-0.010493990033864975,
-0.026340704411268234,
0.01508419867604971,
-0.06838469952344894,
-0.03514210879802704,
-0.0... | 0.056719 |
avoid incompatibilities when models, prompts, or tool definitions change. | https://github.com/openai/openai-agents-python/blob/main/docs/human_in_the_loop.md | main | openai-agents | [
-0.03611920401453972,
-0.04665961116552353,
-0.016707446426153183,
0.016534356400370598,
-0.01877373270690441,
-0.050319988280534744,
-0.06978604942560196,
0.01616285927593708,
0.012851215898990631,
-0.006502303294837475,
0.02913683094084263,
-0.061068203300237656,
0.06996577233076096,
-0.... | 0.072172 |
# Agents Agents are the core building block in your apps. An agent is a large language model (LLM) configured with instructions, tools, and optional runtime behavior such as handoffs, guardrails, and structured outputs. Use this page when you want to define or customize a single plain `Agent`. If you are deciding how m... | https://github.com/openai/openai-agents-python/blob/main/docs/agents.md | main | openai-agents | [
0.03588644415140152,
-0.057851117104291916,
-0.09628242254257202,
-0.006798116490244865,
0.04606645181775093,
-0.006456843111664057,
-0.0340401791036129,
-0.0007103862008079886,
0.07251396030187607,
-0.03888767585158348,
-0.004166399594396353,
-0.031681232154369354,
0.07476469874382019,
-0... | 0.132812 |
See [Forcing tool use](#forcing-tool-use). | ```python from agents import Agent, ModelSettings, function\_tool @function\_tool def get\_weather(city: str) -> str: """returns weather info for the specified city.""" return f"The weather in {city} is sunny" agent = Agent( name="Haiku agent", instructions="Always respond i... | https://github.com/openai/openai-agents-python/blob/main/docs/agents.md | main | openai-agents | [
0.0509549081325531,
0.014673331752419472,
-0.029559804126620293,
0.03877196088433266,
0.09034035354852676,
-0.017671149224042892,
-0.018911568447947502,
-0.03772147744894028,
-0.01852908544242382,
-0.05791395157575607,
0.0020902722608298063,
-0.11179694533348083,
0.05121234804391861,
-0.00... | 0.066266 |
booking\_agent.as\_tool( tool\_name="booking\_expert", tool\_description="Handles booking questions and requests.", ), refund\_agent.as\_tool( tool\_name="refund\_expert", tool\_description="Handles refund questions and requests.", ) ], ) ``` ### Handoffs Handoffs are sub‑agents the agent can delegate to. When a handof... | https://github.com/openai/openai-agents-python/blob/main/docs/agents.md | main | openai-agents | [
-0.05329553410410881,
0.005532151088118553,
-0.08470272272825241,
0.04276301711797714,
-0.08872997760772705,
-0.020985977724194527,
0.07402808964252472,
0.038917798548936844,
0.008198413997888565,
-0.020060207694768906,
-0.01492171548306942,
-0.01701711304485798,
-0.00444506760686636,
0.05... | 0.077839 |
`auto`, which allows the LLM to decide whether or not to use a tool. 2. `required`, which requires the LLM to use a tool (but it can intelligently decide which tool). 3. `none`, which requires the LLM to \_not\_ use a tool. 4. Setting a specific string e.g. `my\_tool`, which requires the LLM to use that specific tool. ... | https://github.com/openai/openai-agents-python/blob/main/docs/agents.md | main | openai-agents | [
-0.031799763441085815,
-0.023118359968066216,
-0.03974808007478714,
0.08338243514299393,
0.05503740534186363,
-0.08597598969936371,
0.0014864015392959118,
-0.04823995754122734,
0.018785353749990463,
-0.03851841762661934,
0.0008704225765541196,
-0.13444888591766357,
0.05551859363913536,
-0.... | 0.058787 |
# Agent visualization Agent visualization allows you to generate a structured graphical representation of agents and their relationships using \*\*Graphviz\*\*. This is useful for understanding how agents, tools, and handoffs interact within an application. ## Installation Install the optional `viz` dependency group: `... | https://github.com/openai/openai-agents-python/blob/main/docs/visualization.md | main | openai-agents | [
0.03622441738843918,
-0.0054029137827456,
-0.07031360268592834,
0.029925020411610603,
0.01397186703979969,
-0.07126850634813309,
-0.008475353009998798,
-0.03121686354279518,
-0.024026161059737206,
0.006754994858056307,
0.03758075460791588,
-0.04886382073163986,
0.08201058954000473,
0.01321... | 0.177854 |
# REPL utility The SDK provides `run\_demo\_loop` for quick, interactive testing of an agent's behavior directly in your terminal. ```python import asyncio from agents import Agent, run\_demo\_loop async def main() -> None: agent = Agent(name="Assistant", instructions="You are a helpful assistant.") await run\_demo\_lo... | https://github.com/openai/openai-agents-python/blob/main/docs/repl.md | main | openai-agents | [
-0.07656516134738922,
-0.02031262405216694,
-0.10321172326803207,
0.063954658806324,
-0.03208232298493385,
-0.09762124717235565,
-0.008446766994893551,
-0.020924611017107964,
0.013241284526884556,
-0.10103588551282883,
0.0022114471066743135,
-0.06864325702190399,
-0.03943027928471565,
-0.0... | 0.15201 |
# Tools Tools let agents take actions: things like fetching data, running code, calling external APIs, and even using a computer. The SDK supports five categories: - Hosted OpenAI tools: run alongside the model on OpenAI servers. - Local/runtime execution tools: `ComputerTool` and `ApplyPatchTool` always run in your en... | https://github.com/openai/openai-agents-python/blob/main/docs/tools.md | main | openai-agents | [
-0.05182822421193123,
-0.02278819866478443,
-0.07731623202562332,
0.055511172860860825,
0.016744187101721764,
-0.10626603662967682,
-0.07639777660369873,
0.037193868309259415,
0.027482908219099045,
0.010263281874358654,
0.0009028132189996541,
-0.03128798305988312,
0.01197018288075924,
-0.0... | 0.149026 |
namespace before using CRM tools.", tools=[\*crm\_tools, ToolSearchTool()], ) result = await Runner.run(agent, "Look up customer\_42 and list their open orders.") print(result.final\_output) ``` What to know: - Hosted tool search is available only with OpenAI Responses models. The current Python SDK support depends on ... | https://github.com/openai/openai-agents-python/blob/main/docs/tools.md | main | openai-agents | [
-0.09205784648656845,
-0.0026311425026506186,
-0.05278944969177246,
0.12405543774366379,
-0.021168645471334457,
-0.09353306144475937,
-0.04698934033513069,
-0.029831213876605034,
-0.02370557375252247,
-0.02637314610183239,
0.03563569113612175,
-0.06361650675535202,
0.0691380724310875,
-0.0... | 0.03264 |
runtime configuration below when you want commands to run in your own process. Local runtime tools require you to supply implementations: - [`ComputerTool`][agents.tool.ComputerTool]: implement the [`Computer`][agents.computer.Computer] or [`AsyncComputer`][agents.computer.AsyncComputer] interface to enable GUI/browser... | https://github.com/openai/openai-agents-python/blob/main/docs/tools.md | main | openai-agents | [
-0.007166268769651651,
0.013066740706562996,
-0.04347725585103035,
0.010045036673545837,
-0.027652733027935028,
-0.027455568313598633,
-0.03612522780895233,
0.03946540132164955,
-0.053061820566654205,
0.019915718585252762,
0.0017892781179398298,
-0.10076935589313507,
-0.005045396275818348,
... | 0.095486 |
- The schema for the function inputs is automatically created from the function's arguments - Descriptions for each input are taken from the docstring of the function, unless disabled We use Python's `inspect` module to extract the function signature, along with [`griffe`](https://mkdocstrings.github.io/griffe/) to par... | https://github.com/openai/openai-agents-python/blob/main/docs/tools.md | main | openai-agents | [
-0.018420860171318054,
-0.017667347565293312,
-0.06605679541826248,
0.06181517243385315,
0.030910369008779526,
-0.036478687077760696,
-0.0632096379995346,
0.00726522970944643,
0.02431453950703144,
-0.06733255088329315,
0.02342422865331173,
-0.07338941097259521,
0.047870926558971405,
-0.009... | 0.08452 |
async function that receives a [`ToolContext`][agents.tool\_context.ToolContext] and the arguments as a JSON string, and returns tool output (for example, text, structured tool output objects, or a list of outputs). ```python from typing import Any from pydantic import BaseModel from agents import RunContextWrapper, Fu... | https://github.com/openai/openai-agents-python/blob/main/docs/tools.md | main | openai-agents | [
-0.06190110743045807,
0.012593531049787998,
-0.13997820019721985,
0.03434809669852257,
-0.051118068397045135,
-0.05607878044247627,
-0.019953211769461632,
0.06430764496326447,
-0.018636245280504227,
-0.09279420971870422,
0.04726225510239601,
-0.06015127897262573,
0.037101566791534424,
0.02... | 0.022716 |
anything), it runs a `default\_tool\_error\_function` which tells the LLM an error occurred. - If you pass your own error function, it runs that instead, and sends the response to the LLM. - If you explicitly pass `None`, then any tool call errors will be re-raised for you to handle. This could be a `ModelBehaviorError... | https://github.com/openai/openai-agents-python/blob/main/docs/tools.md | main | openai-agents | [
-0.021837493404746056,
-0.027547277510166168,
-0.04113814979791641,
0.029008952900767326,
0.004319388885051012,
-0.044623780995607376,
-0.02850385196506977,
0.10035260021686554,
-0.006028308533132076,
-0.05304277688264847,
0.056397322565317154,
-0.04871971905231476,
0.09219769388437271,
-0... | 0.08814 |
pending items appear in `result.interruptions`; then use `result.to\_state()` and resume after calling `state.approve(...)` or `state.reject(...)`. See the [Human-in-the-loop guide](human\_in\_the\_loop.md) for the full pause/resume pattern. ### Custom output extraction In certain cases, you might want to modify the ou... | https://github.com/openai/openai-agents-python/blob/main/docs/tools.md | main | openai-agents | [
-0.0644262284040451,
0.0538521483540535,
-0.05724450945854187,
0.05252186581492424,
0.029093097895383835,
-0.03588943928480148,
0.03602247312664986,
-0.025858353823423386,
-0.0056878202594816685,
-0.053727712482213974,
-0.010271959938108921,
0.0114408228546381,
-0.024277647957205772,
-0.01... | 0.049254 |
to the user's question in French", is\_enabled=french\_enabled, ), ], ) async def main(): context = RunContextWrapper(LanguageContext(language\_preference="french\_spanish")) result = await Runner.run(orchestrator, "How are you?", context=context.context) print(result.final\_output) asyncio.run(main()) ``` The `is\_ena... | https://github.com/openai/openai-agents-python/blob/main/docs/tools.md | main | openai-agents | [
-0.004347606096416712,
-0.009446538053452969,
-0.092657670378685,
0.06276775896549225,
0.01031587552279234,
0.02502172440290451,
0.020632563158869743,
-0.007173948455601931,
-0.052830569446086884,
-0.09941935539245605,
0.04950939863920212,
-0.06802883744239807,
-0.02428373321890831,
-0.000... | 0.098275 |
- See `examples/tools/codex.py` and `examples/tools/codex\_same\_thread.py` for complete runnable samples. | https://github.com/openai/openai-agents-python/blob/main/docs/tools.md | main | openai-agents | [
-0.062398359179496765,
-0.056239839643239975,
-0.07056967169046402,
0.001289948122575879,
0.01497095637023449,
-0.11645606905221939,
-0.06701746582984924,
0.0491175577044487,
-0.03299221768975258,
-0.051920514553785324,
0.02061447873711586,
0.03699292987585068,
0.0022279699333012104,
-0.06... | 0.025662 |
# Agent orchestration Orchestration refers to the flow of agents in your app. Which agents run, in what order, and how do they decide what happens next? There are two main ways to orchestrate agents: 1. Allowing the LLM to make decisions: this uses the intelligence of an LLM to plan, reason, and decide on what steps to... | https://github.com/openai/openai-agents-python/blob/main/docs/multi_agent.md | main | openai-agents | [
0.006568347103893757,
-0.030541060492396355,
-0.02091183327138424,
-0.0025874979328364134,
-0.03164418786764145,
-0.07014060765504837,
-0.07484868913888931,
0.037851881235837936,
0.03581143543124199,
-0.0018362662522122264,
-0.07639925926923752,
0.023995157331228256,
0.07563800364732742,
-... | 0.135256 |
orchestrating via LLM is powerful, orchestrating via code makes tasks more deterministic and predictable, in terms of speed, cost and performance. Common patterns here are: - Using [structured outputs](https://platform.openai.com/docs/guides/structured-outputs) to generate well formed data that you can inspect with you... | https://github.com/openai/openai-agents-python/blob/main/docs/multi_agent.md | main | openai-agents | [
0.026990169659256935,
-0.03583312779664993,
-0.06918244808912277,
0.02254779264330864,
-0.006153469439595938,
-0.08921777456998825,
-0.0859389379620552,
-0.008831681683659554,
0.02450854517519474,
0.01621614582836628,
-0.0734301507472992,
0.00041437125764787197,
0.06501223146915436,
-0.026... | 0.132515 |
# TUI The TUI is the modern front-end for Hermes — a terminal UI backed by the same Python runtime as the [Classic CLI](cli.md). Same agent, same sessions, same slash commands; a cleaner, more responsive surface for interacting with them. It's the recommended way to run Hermes interactively. ## Launch ```bash # Launch ... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/tui.md | main | hermes-agent | [
-0.08388501405715942,
0.014081951230764389,
-0.04846493527293205,
-0.024586357176303864,
-0.02496008388698101,
-0.09647352993488312,
-0.008755289018154144,
-0.03690991550683975,
-0.016067886725068092,
-0.011959072202444077,
-0.029869897291064262,
-0.0179936271160841,
0.0451488271355629,
-0... | 0.15082 |
resets to the defaults. ## Requirements - \*\*Node.js\*\* ≥ 20 — the TUI runs as a subprocess launched from the Python CLI. `hermes doctor` verifies this. - \*\*TTY\*\* — like the classic CLI, piping stdin or running in non-interactive environments falls back to single-query mode. On first launch Hermes installs the TU... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/tui.md | main | hermes-agent | [
-0.07260040193796158,
-0.005771027412265539,
-0.001400891225785017,
-0.009148313663899899,
0.0022368126083165407,
-0.11508476734161377,
-0.05495574325323105,
-0.02740631066262722,
-0.024782491847872734,
0.000881743966601789,
-0.005930974613875151,
-0.006180522032082081,
0.012583650648593903,... | 0.150285 |
- `/sessions` or `/switch`. - `/sessions new` to create a fresh live session immediately. - Click the `N live sessions` count in the status line. Inside the switcher: - `↑` / `↓` move the selection; mouse clicks select rows too. - `Enter` switches to the selected live session. - `Ctrl+D` closes the selected live sessio... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/tui.md | main | hermes-agent | [
-0.09296046197414398,
-0.049400001764297485,
-0.052928242832422256,
0.06011825054883957,
-0.03845132142305374,
-0.030762769281864166,
0.0042974455282092094,
-0.0046891882084310055,
0.017630085349082947,
0.055734310299158096,
0.020402489230036736,
-0.04109259322285652,
-0.05052084103226662,
... | 0.142695 |
(live), frozen to `⏲ 32s / 3m 45s` after the turn completes. First number is time since last user message; second is total session duration. Resets on every new prompt. - \*\*`🗜️ N`\*\* — number of times the running session has been auto-compressed. Appears once the first compression fires. - \*\*`▶ N`\*\* — number of... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/tui.md | main | hermes-agent | [
-0.06325636804103851,
0.04777650907635689,
0.0097960801795125,
-0.0285963024944067,
0.027480484917759895,
-0.05898480489850044,
0.03251338005065918,
-0.05991208180785179,
0.005188674200326204,
0.017728641629219055,
0.041898030787706375,
-0.06181248649954796,
0.03259746730327606,
-0.0067697... | 0.156088 |
can point the TUI at that gateway instead — the TUI then becomes a thin client and shares state with every other surface (messaging platforms, web dashboard, other TUI sessions) that's attached to the same gateway. Set the websocket URL via env before launching: ```bash export HERMES\_TUI\_GATEWAY\_URL="ws://localhost:... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/tui.md | main | hermes-agent | [
-0.08854801207780838,
-0.043178994208574295,
-0.05927027389407158,
-0.007978870533406734,
-0.0031429387163370848,
-0.09719768911600113,
0.02437332086265087,
-0.029821349307894707,
0.03219461441040039,
-0.003949995618313551,
-0.039548829197883606,
-0.019834909588098526,
0.017658870667219162,
... | 0.170081 |
# CLI Interface Hermes Agent's CLI is a full terminal user interface (TUI) — not a web UI. It features multiline editing, slash-command autocomplete, conversation history, interrupt-and-redirect, and streaming tool output. Built for people who live in the terminal. :::tip First-time setup One command — `hermes setup --... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/cli.md | main | hermes-agent | [
-0.0559568926692009,
-0.0029766273219138384,
-0.042760927230119705,
-0.04024515300989151,
-0.0044939881190657616,
-0.09721339493989944,
0.04611402004957199,
-0.03713761642575264,
0.022018907591700554,
-0.004916589241474867,
0.010356869548559189,
-0.03401036933064461,
0.029827313497662544,
... | 0.199412 |
"Previous Conversation" panel appears between the banner and the input prompt, showing a compact recap of the conversation history. See [Sessions — Conversation Recap on Resume](sessions.md#conversation-recap-on-resume) for details and configuration. ## Keybindings | Key | Action | |-----|--------| | `Enter` | Send mes... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/cli.md | main | hermes-agent | [
-0.05991406366229057,
-0.052227385342121124,
0.005491230636835098,
-0.02665833756327629,
-0.01218874566257,
0.0632159635424614,
0.02204243279993534,
-0.009293148294091225,
0.04479561001062393,
-0.014860708266496658,
-0.0145949088037014,
-0.01757165417075157,
-0.03005543351173401,
0.0038000... | 0.060298 |
Commands are case-insensitive — `/HELP` works the same as `/help`. Installed skills also become slash commands automatically. ::: ## Quick Commands You can define custom commands that run shell commands instantly without invoking the LLM. These work in both the CLI and messaging platforms (Telegram, Discord, etc.). ```... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/cli.md | main | hermes-agent | [
-0.012391203083097935,
-0.032065726816654205,
-0.051200658082962036,
0.009881474077701569,
-0.027623198926448822,
-0.028868919238448143,
0.005930792074650526,
0.05554892122745514,
-0.012778831645846367,
0.030219068750739098,
0.017644334584474564,
-0.03390110656619072,
-0.01243062037974596,
... | 0.097384 |
commands are killed immediately (SIGTERM, then SIGKILL after 1s) - Multiple messages typed during interrupt are combined into one prompt ### Busy Input Mode The `display.busy\_input\_mode` config key controls what happens when you press Enter while the agent is working: | Mode | Behavior | |------|----------| | `"inter... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/cli.md | main | hermes-agent | [
0.047440242022275925,
-0.015378478914499283,
0.0007540888036601245,
0.016038494184613228,
-0.012759056873619556,
-0.0723232626914978,
0.03474624454975128,
-0.03228779509663582,
0.0616387240588665,
0.0457463264465332,
0.08931543678045273,
0.004735386930406094,
0.04703318700194359,
-0.061042... | 0.134284 |
a specific session by ID hermes --resume "refactoring auth" # Resume by title hermes -r 20260225\_143052\_a1b2c3 # Short form ``` Resuming restores the full conversation history from SQLite. The agent sees all previous messages, tool calls, and responses — just as if you never left. Use `/title My Session Name` inside ... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/cli.md | main | hermes-agent | [
-0.08134587854146957,
0.011094083078205585,
-0.014211313799023628,
-0.028114980086684227,
-0.0017455642810091376,
-0.0412532314658165,
0.05362901836633682,
-0.055719707161188126,
0.06139139458537102,
0.005950598977506161,
-0.037986207753419876,
0.025623958557844162,
0.00496377469971776,
-0... | 0.117498 |
in quiet mode which: - Suppresses verbose logging from tools - Enables kawaii-style animated feedback - Keeps output clean and user-friendly For debug output: ```bash hermes chat --verbose ``` | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/cli.md | main | hermes-agent | [
0.03741865232586861,
0.028450515121221542,
-0.005470150150358677,
0.06696050614118576,
0.06341216713190079,
-0.11007187515497208,
0.03456398472189903,
-0.07622072845697403,
0.03050093539059162,
0.023696783930063248,
0.01632133685052395,
-0.033076584339141846,
-0.011377576738595963,
-0.0308... | 0.109658 |
# Profiles: Running Multiple Agents Run multiple independent Hermes agents on the same machine — each with its own config, API keys, memory, sessions, skills, and gateway state. ## What are profiles? A profile is a separate Hermes home directory. Each profile gets its own directory containing its own `config.yaml`, `.e... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/profiles.md | main | hermes-agent | [
-0.03289991244673729,
-0.0691026896238327,
-0.09369955956935883,
-0.013603473082184792,
-0.017195943742990494,
-0.08104649186134338,
0.010546614415943623,
0.03033183142542839,
-0.005623598583042622,
0.0003215093747712672,
0.026914725080132484,
-0.039937324821949005,
0.10235180705785751,
-0... | 0.086696 |
# now targets coder hermes tools # configures coder's tools hermes profile use default # switch back ``` Sets a default so plain `hermes` commands target that profile. Like `kubectl config use-context`. ### Knowing where you are The CLI always shows which profile is active: - \*\*Prompt\*\*: `coder ❯` instead of `❯` - ... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/profiles.md | main | hermes-agent | [
0.009305180050432682,
-0.02664935775101185,
-0.02894747443497181,
-0.04162435233592987,
0.027607351541519165,
-0.04037059098482132,
0.009298022836446762,
-0.007937882095575333,
0.024850264191627502,
-0.010407721623778343,
0.015580488368868828,
-0.08125430345535278,
0.046241458505392075,
-0... | 0.098357 |
`hermes update` pulls code once (shared) and syncs new bundled skills to \*\*all\*\* profiles automatically: ```bash hermes update # → Code updated (12 commits) # → Skills synced: default (up to date), coder (+2 new), assistant (+2 new) ``` User-modified skills are never overwritten. ## Managing profiles ```bash hermes... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/profiles.md | main | hermes-agent | [
-0.07215303182601929,
-0.008898939937353134,
-0.03998566418886185,
-0.031546369194984436,
0.01648535020649433,
-0.05690716579556465,
0.02008027769625187,
-0.02272668108344078,
-0.057745397090911865,
0.029427679255604744,
0.03674456477165222,
0.009785152971744537,
0.07828757166862488,
-0.03... | 0.009187 |
# Windows (Native) Guide Hermes runs natively on Windows 10 and Windows 11 — no WSL, no Cygwin, no Docker. This page is the deep dive: what works natively, what's WSL-only, what the installer actually does, and the Windows-specific knobs you might need to touch. If you just want to install, the one-liner on the [landin... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/windows-native.md | main | hermes-agent | [
0.013492962345480919,
0.05647802725434303,
-0.047038208693265915,
-0.019827984273433685,
0.021774252876639366,
-0.026232432574033737,
-0.01148177683353424,
0.03433215618133545,
-0.0014475563075393438,
0.009162207134068012,
-0.017879467457532883,
-0.06985189765691757,
0.021120600402355194,
... | -0.046893 |
has a `shutil.which(...)`-style check; if a binary is missing and the run is interactive, `dep\_ensure` offers to install it (deferring to `scripts\install.ps1 -ensure ` for the actual install logic). Non-interactive runs (gateway, cron, headless desktop launches) skip the prompt and surface a clear `this feature needs... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/windows-native.md | main | hermes-agent | [
-0.05886911600828171,
0.011168821714818478,
0.06566791236400604,
-0.021637164056301117,
0.11881700158119202,
-0.11103104054927826,
-0.031570274382829666,
0.055738791823387146,
-0.04387546330690384,
0.04469793289899826,
0.010822628624737263,
-0.020177854225039482,
0.038613252341747284,
0.04... | -0.025039 |
commands through \*\*Git Bash\*\*, same strategy Claude Code uses. This sidesteps the POSIX-vs-Windows gap without rewriting every tool. Resolution order for `bash.exe`: 1. `HERMES\_GIT\_BASH\_PATH` environment variable if set. 2. `%LOCALAPPDATA%\hermes\git\usr\bin\bash.exe` (installer-managed PortableGit). 3. `%LOCALA... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/windows-native.md | main | hermes-agent | [
0.03347053751349449,
-0.01014704816043377,
-0.048785220831632614,
0.028043165802955627,
-0.0008463357225991786,
0.004837564658373594,
0.039321206510066986,
0.006179280113428831,
0.05611609295010567,
0.04274069145321846,
-0.02433546632528305,
0.02516673319041729,
0.04447438195347786,
0.0331... | -0.020449 |
default on Windows 11). ## Running the gateway at Windows login `hermes gateway install` on Windows uses \*\*Scheduled Tasks\*\* with a Startup-folder fallback — no admin required. ### Install ```powershell hermes gateway install ``` What happens under the hood: 1. `schtasks /Create /SC ONLOGON /RL LIMITED /TN HermesGa... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/windows-native.md | main | hermes-agent | [
-0.03074236772954464,
-0.010762276127934456,
-0.06063134968280792,
0.03523735702037811,
0.0006584576913155615,
-0.016432883217930794,
-0.03287745267152786,
-0.03654787689447403,
0.04616512358188629,
0.0402754507958889,
-0.06118715554475784,
0.04455067589879036,
0.03458121418952942,
0.03976... | 0.077839 |
up — open a new PowerShell window (or Windows Terminal tab) after installation. Close-and-reopen, don't `$env:PATH += …` by hand unless you know what you're doing. Verify: ```powershell Get-Command hermes # should print C:\Users\\AppData\Local\hermes\bin\hermes.cmd hermes --version ``` ### Environment variables Hermes ... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/windows-native.md | main | hermes-agent | [
0.04811323061585426,
-0.013602277263998985,
-0.051620956510305405,
-0.016139773651957512,
-0.034341391175985336,
-0.02533136121928692,
0.04096245765686035,
0.029831621795892715,
-0.003944839350879192,
0.09586801379919052,
-0.032829198986291885,
-0.06367338448762894,
0.010271611623466015,
0... | 0.012407 |
switch to the `.cmd` variant (e.g., `npx.cmd`, not `npx`). \*\*`[scriptblock]::Create(...)` fails with `The assignment expression is not valid`.\*\* Your download of `install.ps1` picked up a UTF-8 BOM. The `irm | iex` form strips BOMs automatically; `[scriptblock]::Create((irm ...))` does not. Re-run with the simple `... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/windows-native.md | main | hermes-agent | [
-0.011033521965146065,
-0.06372855603694916,
-0.10393472015857697,
0.03718425706028938,
0.02861810475587845,
-0.01564253494143486,
0.004685642197728157,
-0.018211115151643753,
0.02250727266073227,
0.0734758973121643,
-0.007175588980317116,
0.004516961053013802,
0.029969321563839912,
-0.038... | 0.075523 |
# Desktop App The Hermes desktop app is a native app built around the \*\*same\*\* agent you get from the CLI and the gateway — same config, same API keys, same sessions, same skills, same memory. It is not a separate product or a lightweight clone; it uses the same Hermes Agent core and settings, and drives it through... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/desktop.md | main | hermes-agent | [
-0.050678085535764694,
0.0016480570193380117,
-0.0474228672683239,
-0.09461165964603424,
-0.023644190281629562,
-0.11169321089982986,
-0.0031209271401166916,
0.022266320884227753,
0.02560938522219658,
0.009739656932651997,
0.020578887313604355,
-0.03589801862835884,
0.07327625155448914,
-0... | 0.15523 |
\*\*Agents\*\* and \*\*Command Center\*\* — orchestration surfaces for multi-agent work. ## Updating The app checks for updates in the background and offers a one-click update when one is ready. The [manual update process](https://hermes-agent.nousresearch.com/docs/getting-started/updating) also works with the GUI. ## ... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/desktop.md | main | hermes-agent | [
-0.04071549326181412,
0.034250956028699875,
-0.011248200200498104,
-0.05842849239706993,
-0.032005004584789276,
-0.09110284596681595,
-0.09916844964027405,
-0.05212227627635002,
0.014999542385339737,
0.022502856329083443,
0.010412937961518764,
-0.07270099222660065,
0.0629417672753334,
-0.0... | 0.11197 |
~/.hermes/.env chmod 600 ~/.hermes/.env echo "$TOKEN" # copy this value into the desktop app # 2. Run the dashboard bound to a reachable address. # --tui enables the embedded chat (the /api/ws + /api/pty WebSockets the # desktop drives) — without it the app connects but chat stays dead. # --insecure is required for any... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/desktop.md | main | hermes-agent | [
-0.07152116298675537,
0.03745236247777939,
-0.09483802318572998,
-0.05841489136219025,
0.02166328765451908,
-0.11840303987264633,
0.030915897339582443,
0.07765109837055206,
0.030529990792274475,
0.016855355352163315,
-0.001392449950799346,
-0.016428645700216293,
0.09484768658876419,
0.0289... | 0.101515 |
repo root once, then run the dev server from `apps/desktop`: ```bash npm install # from repo root — links apps/desktop, web, apps/shared cd apps/desktop npm run dev # Vite renderer + Electron, which boots the Python backend ``` Point the app at a specific checkout, or sandbox it from your real config: ```bash HERMES\_D... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/desktop.md | main | hermes-agent | [
-0.04288799688220024,
-0.012174027040600777,
-0.02606862410902977,
-0.037145961076021194,
0.04698500782251358,
-0.09981553256511688,
-0.024590663611888885,
0.049298856407403946,
-0.02365688607096672,
0.05762378126382828,
0.019220327958464622,
-0.0686371698975563,
0.06732913106679916,
0.004... | 0.047665 |
# Profile Distributions: Share a Whole Agent A \*\*profile distribution\*\* packages a complete Hermes agent — personality, skills, cron jobs, MCP connections, config — as a git repository. Anyone with access to the repo can install the whole agent with one command, update it in place, and keep their own memories, sess... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/profile-distributions.md | main | hermes-agent | [
-0.04197417199611664,
-0.059358149766922,
-0.017788317054510117,
-0.03763251751661301,
0.013114982284605503,
-0.07097044587135315,
0.01515132188796997,
-0.0009656785405240953,
-0.010154381394386292,
0.022647138684988022,
0.06146720051765442,
0.0013129855506122112,
0.07551886886358261,
-0.0... | 0.082341 |
share API keys alongside the agent.\*\* `auth.json` and `.env` are deliberately excluded from distributions. Each installer brings their own credentials. - \*\*You want to share memories / sessions / conversation history.\*\* Those are user data, not distribution content. Never shipped. ## The lifecycle: author to inst... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/profile-distributions.md | main | hermes-agent | [
-0.022412553429603577,
-0.045494530349969864,
-0.07600194215774536,
-0.012115160934627056,
-0.04008606821298599,
-0.03328825905919075,
-0.036563899368047714,
0.03482755273580551,
-0.007364410441368818,
0.03181665763258934,
0.06752164661884308,
-0.011647681705653667,
0.07329966127872467,
0.... | 0.054646 |
skills stay - cron/digest.json ``` When omitted, the defaults above apply — which is what most distributions want. --- ## For installers: using a distribution ### Install ```bash hermes profile install github.com/you/research-bot --alias ``` What happens: 1. Clones the repo into a temporary directory. 2. Reads `distrib... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/profile-distributions.md | main | hermes-agent | [
-0.0207405686378479,
-0.024446384981274605,
-0.04007943719625473,
-0.056766003370285034,
0.01700395904481411,
-0.08445833623409271,
0.009521779604256153,
0.04705152288079262,
-0.04796278104186058,
0.03369101881980896,
0.04276500642299652,
-0.015578115358948708,
0.07818293571472168,
-0.0072... | 0.017608 |
you to confirm: ``` Profile: research-bot Path: ~/.hermes/profiles/research-bot Model: claude-opus-4 (anthropic) Skills: 12 Distribution: research-bot@1.0.0 Installed from: https://github.com/you/research-bot This will permanently delete: • All config, API keys, memories, sessions, skills, cron jobs • Command alias (~/... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/profile-distributions.md | main | hermes-agent | [
-0.043983008712530136,
-0.03855237364768982,
-0.06861254572868347,
-0.027207985520362854,
0.04280567169189453,
-0.03324126452207565,
-0.031268298625946045,
0.02798428386449814,
-0.012504437938332558,
0.03640015423297882,
0.07221446931362152,
0.02458837255835533,
0.06835315376520157,
0.0088... | 0.098817 |
lead. You want a temporary agent that diagnoses a production incident — a canned SOUL with the right tools and MCP connections — and runs on three on-call engineers' laptops for the next week. ```bash # You # Build the profile, commit, push a private repo git push -u origin main # Each on-call hermes profile install gi... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/profile-distributions.md | main | hermes-agent | [
-0.02762189880013466,
-0.034971099346876144,
0.03254228085279465,
-0.056508176028728485,
0.05304238200187683,
-0.0758185014128685,
-0.028697656467556953,
0.03880719840526581,
0.00043319101678207517,
0.04091132432222366,
0.06530220061540604,
0.027599675580859184,
0.025078127160668373,
-0.03... | 0.067276 |
prints `hermes -p cron list` and you enable them explicitly. SOUL.md and skills ARE active as soon as you start chatting with the profile, so read them before your first run if you're installing from someone you don't know. Rough analogy: installing a distribution is like installing a browser extension or a VS Code ext... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/profile-distributions.md | main | hermes-agent | [
-0.022692406550049782,
-0.0005577711272053421,
-0.0003699272347148508,
-0.08250048756599426,
0.029040442779660225,
-0.09089666604995728,
0.007154433522373438,
-0.009982982650399208,
-0.0029338044114410877,
0.03383658081293106,
0.07732846587896347,
0.022007063031196594,
-0.00531623512506485,
... | 0.015076 |
# Windows (WSL2) Guide Hermes Agent now supports \*\*both\*\* native Windows and WSL2. This page covers the WSL2 path; for the native PowerShell install see the dedicated \*\*[Windows (Native) Guide](./windows-native.md)\*\*. \*\*When to pick WSL2 over native:\*\* - You want to use the dashboard's embedded terminal (`/... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/windows-wsl-quickstart.md | main | hermes-agent | [
0.015192163176834583,
0.013914041221141815,
-0.022894179448485374,
-0.011585352011024952,
-0.007300457451492548,
-0.08170786499977112,
0.0026158473920077085,
0.04379275068640709,
0.04599880799651146,
0.00501071335747838,
-0.028145458549261093,
-0.04152080789208412,
0.008759932592511177,
0.... | -0.003871 |
### Distro choice Ubuntu (LTS) is what we test against. Debian works. Arch and NixOS work for people who want them, but the one-line installer assumes a Debian-derived `apt` system — see the [Nix setup guide](/getting-started/nix-setup) for that path. ### Enable systemd (recommended) The hermes gateway (and anything el... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/windows-wsl-quickstart.md | main | hermes-agent | [
0.03142218664288521,
0.03298163786530495,
-0.03134285286068916,
-0.04971398040652275,
0.009263742715120316,
-0.052000436931848526,
-0.03809884563088417,
0.12710827589035034,
0.023956038057804108,
-0.00046152749564498663,
-0.016255149617791176,
0.022785091772675514,
-0.009616184048354626,
0... | 0.026779 |
```bash cp ~/reports/output.pdf /mnt/c/Users/you/Desktop/ ``` \*\*Open a WSL file in a Windows app\*\* (GUI editor, browser, etc.): use `explorer.exe` or `wslview`: ```bash sudo apt install wslu # once — gives you wslview, wslpath, wslopen, etc. wslview ~/reports/output.pdf # opens with the Windows default handler expl... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/windows-wsl-quickstart.md | main | hermes-agent | [
0.04118616133928299,
-0.02446444146335125,
-0.1540149450302124,
0.055839959532022476,
0.01225272472947836,
0.023631831631064415,
-0.0318794921040535,
0.10274132341146469,
0.05254487320780754,
-0.0020100194960832596,
-0.0060539464466273785,
0.00009763130219653249,
0.009241357445716858,
0.05... | -0.042971 |
rather than raw port forwarding. #### Subcase 2a: from the Windows host itself On \*\*Windows 11 22H2+ with mirrored mode enabled\*\*, there is nothing to do. A process in WSL that binds to `0.0.0.0:8080` (or even `127.0.0.1:8080`) is reachable from a Windows browser at `http://localhost:8080`. WSL publishes the bind b... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/windows-wsl-quickstart.md | main | hermes-agent | [
0.030290227383375168,
0.05743839964270592,
-0.04173795133829117,
-0.009816566482186317,
-0.03654845431447029,
-0.0636269673705101,
-0.11263182014226913,
-0.008436313830316067,
0.02718251571059227,
-0.009093981236219406,
0.0058821639977395535,
0.015876825898885727,
-0.022558927536010742,
0.... | -0.066376 |
Inside WSL with systemd (recommended) If you enabled systemd per the setup section above, `hermes gateway` and the API server work the way they do on any Linux machine. Use the gateway setup wizard: ```bash hermes gateway setup ``` It will offer to install a systemd user unit so the gateway comes up automatically when ... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/windows-wsl-quickstart.md | main | hermes-agent | [
-0.012234562076628208,
0.019585859030485153,
-0.09584847837686539,
0.032740507274866104,
-0.016216784715652466,
0.011832685209810734,
-0.05969617888331413,
0.009752195328474045,
0.0520680770277977,
0.005066209007054567,
-0.04354485496878624,
0.05578871816396713,
0.011112798936665058,
0.009... | -0.011065 |
files.\*\* Defender scans files via the 9P bridge when accessed from Windows, which magnifies the slowness of `/mnt/c`-style cross-boundary access. If you only touch WSL files from inside WSL, this doesn't matter. If you use Windows tools against `\\wsl$\...` frequently, consider excluding the WSL distro path from real... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/windows-wsl-quickstart.md | main | hermes-agent | [
0.033420491963624954,
0.07282718271017075,
-0.05182419344782829,
0.030644452199339867,
0.08882854878902435,
0.002554115140810609,
0.02667011134326458,
0.017432747408747673,
0.07720941305160522,
-0.009625902399420738,
0.07208533585071564,
0.0652150958776474,
-0.012110761366784573,
0.0076361... | -0.005992 |
# Running Many Gateways at Once Operate multiple [profiles](./profiles.md) — each with its own bot tokens, sessions, and memory — as managed services on a single machine. This page covers the operational concerns: starting them all together, viewing logs across profiles, preventing the host from sleeping, and recoverin... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/multi-profile-gateways.md | main | hermes-agent | [
-0.02884535677731037,
-0.08664494752883911,
-0.05979708582162857,
-0.04380517825484276,
-0.03778846934437752,
-0.05976537615060806,
-0.004903004504740238,
0.046201445162296295,
0.00874539278447628,
0.028950156643986702,
-0.013499902561306953,
0.01785428822040558,
0.09470071643590927,
-0.01... | 0.118519 |
| | macOS | `~/Library/LaunchAgents/ai.hermes.gateway-.plist` | | Linux | `~/.config/systemd/user/hermes-gateway-.service` | The default profile keeps the historical names: `ai.hermes.gateway.plist` / `hermes-gateway.service`. ## Viewing logs Each profile writes to its own log files: ```bash # Default profile tail -f ~... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/multi-profile-gateways.md | main | hermes-agent | [
0.017634719610214233,
-0.026937756687402725,
-0.04337938502430916,
-0.0692581832408905,
0.034032683819532394,
-0.055775053799152374,
0.054315052926540375,
0.05200595781207085,
0.042021676898002625,
-0.023546693846583366,
0.005316656548529863,
-0.009027219377458096,
0.05916406214237213,
0.0... | 0.064052 |
code `hermes update` pulls the latest code once and syncs new bundled skills into every profile: ```bash hermes update hermes-gateways restart ``` User-modified skills are never overwritten. ## Troubleshooting ### "Could not find service in domain for user gui: 501" You ran `hermes gateway start` after a previous `herm... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/multi-profile-gateways.md | main | hermes-agent | [
-0.05629545450210571,
-0.0474153570830822,
-0.039546918123960495,
-0.10172727704048157,
0.0014089293545112014,
-0.054861512035131454,
-0.005313629284501076,
-0.04069248214364052,
0.009209328331053257,
0.031300995498895645,
-0.008386882953345776,
0.07120522111654282,
0.047636713832616806,
0... | 0.017938 |
# Configuring Models Hermes uses two kinds of model slots: - \*\*Main model\*\* — what the agent thinks with. Every user message, every tool-call loop, every streamed response goes through this model. - \*\*Auxiliary models\*\* — smaller side-jobs the agent offloads. Context compression, vision (image analysis), web-pa... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/configuring-models.md | main | hermes-agent | [
0.01316309068351984,
-0.02325727976858616,
-0.041962772607803345,
-0.023462962359189987,
-0.0035580494441092014,
-0.08250860869884491,
-0.044327545911073685,
-0.02347201481461525,
0.02099441923201084,
0.009515147656202316,
0.01568148285150528,
0.0013471832498908043,
0.03928406909108162,
-0... | 0.151605 |
Opus. Default config sets this to `google/gemini-3-flash-preview` on OpenRouter. | | \*\*Vision\*\* | When your main model lacks vision support. Point it at `google/gemini-2.5-flash` or `gpt-4o-mini`. | | \*\*Compression\*\* | When you're burning reasoning tokens on Opus/M2.7 just to summarize context. A fast chat mode... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/configuring-models.md | main | hermes-agent | [
-0.09276075661182404,
-0.0737963393330574,
0.04361671581864357,
-0.005018291994929314,
0.10399778932332993,
-0.00028437984292395413,
-0.009522439911961555,
0.0034060196485370398,
0.05096469447016716,
-0.0020981961861252785,
-0.031047852709889412,
0.03193466737866402,
-0.11082828789949417,
... | 0.075804 |
has a working credential. Check \*\*Keys\*\* in the sidebar — you should see one of: an API key, a successful OAuth, or a custom endpoint URL. If the provider you want isn't there, run `hermes setup` to wire it up, or go to \*\*Keys\*\* and add the env var. ### Main model didn't change in my running chat Expected. The ... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/configuring-models.md | main | hermes-agent | [
-0.041755300015211105,
-0.04289805144071579,
0.015866290777921677,
-0.022780615836381912,
-0.030071280896663666,
-0.08556725829839706,
-0.020423024892807007,
-0.05074658989906311,
0.07098200172185898,
0.062270160764455795,
-0.06121279299259186,
-0.015687651932239532,
0.014395182020962238,
... | 0.038333 |
`~/.hermes/config.yaml` and restart whatever reads it. See the [Configuration reference](./configuration.md) for the full schema. ### REST API The dashboard uses three endpoints. Useful for scripting: ```bash # List authenticated providers + curated model lists curl -H "X-Hermes-Session-Token: $TOKEN" http://localhost:... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/configuring-models.md | main | hermes-agent | [
0.017602387815713882,
0.015431121923029423,
-0.07377054542303085,
-0.039898812770843506,
-0.008232341147959232,
-0.026885081082582474,
-0.06910139322280884,
-0.0011105694575235248,
0.047913458198308945,
0.05279652774333954,
-0.010546861216425896,
-0.03260893002152443,
0.019294792786240578,
... | 0.051364 |
# Configuration All settings are stored in the `~/.hermes/` directory for easy access. :::tip Easiest path to a working `config.yaml` Run `hermes setup --portal` — one OAuth gets you a model provider and all four Tool Gateway tools without hand-editing YAML. Portal subscribers also get 10% off token-billed providers. S... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/configuration.md | main | hermes-agent | [
0.008079181425273418,
-0.003914354834705591,
-0.12611493468284607,
0.0037403700407594442,
-0.027979694306850433,
-0.05520248785614967,
-0.04004768282175064,
0.020517250522971153,
0.012170745991170406,
0.0088066840544343,
0.041710153222084045,
0.04264695197343826,
0.040107980370521545,
-0.0... | 0.104086 |
(direct or via the Nous-managed gateway), a Daytona workspace, or a Singularity/Apptainer container. ```yaml terminal: backend: local # local | docker | ssh | modal | daytona | singularity cwd: "." # Gateway/cron working directory (CLI always uses launch dir) timeout: 180 # Per-command timeout in seconds env\_passthrou... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/configuration.md | main | hermes-agent | [
-0.027736084535717964,
0.03246701881289482,
-0.018279632553458214,
-0.0321488119661808,
0.08480113744735718,
-0.06964309513568878,
-0.08031484484672546,
-0.010685584507882595,
0.01882118172943592,
0.019926175475120544,
-0.0009687910787761211,
0.026546359062194824,
0.02770140767097473,
-0.0... | 0.111248 |
Sweep abandoned Exited containers at startup # Cross-backend lifecycle settings (apply to docker as well) timeout: 180 # Per-command timeout in seconds lifetime\_seconds: 300 # Idle-reaper window; also feeds 2× orphan-reaper threshold ``` \*\*`docker\_env`\*\* vs \*\*`docker\_forward\_env`\*\*: the former injects liter... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/configuration.md | main | hermes-agent | [
0.03272486478090286,
0.08319181203842163,
-0.011841772124171257,
-0.019597576931118965,
0.009466967545449734,
-0.08523118495941162,
-0.04067128896713257,
-0.0021057864651083946,
0.02456413395702839,
0.023669416084885597,
0.02237165905535221,
-0.0697905421257019,
-0.004514655563980341,
-0.0... | 0.040247 |
bg processes do not. - \*\*Switching profiles\*\* isolates containers from each other — a container labeled `hermes-profile=work` is invisible to a Hermes process running under `hermes-profile=research`. The orphan reaper is profile-scoped too, so cross-profile containers don't get reaped accidentally, but they also wo... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/configuration.md | main | hermes-agent | [
-0.010791616514325142,
-0.04490120708942413,
0.0054770871065557,
-0.04035230353474617,
0.031015176326036453,
-0.09501900523900986,
-0.007094235625118017,
-0.011945657432079315,
-0.022102514281868935,
-0.008243925869464874,
-0.0019480935297906399,
-0.035833705216646194,
0.05478741601109505,
... | 0.068354 |
\*\*Persistence:\*\* When enabled, the sandbox filesystem is snapshotted on cleanup and restored on next session. Snapshots are tracked in `~/.hermes/modal\_snapshots.json`. This preserves filesystem state, not live processes, PID space, or background jobs. \*\*Credential files:\*\* Automatically mounted from `~/.herme... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/configuration.md | main | hermes-agent | [
-0.09387315809726715,
0.002186859492212534,
-0.047044701874256134,
0.010239341296255589,
0.02329998090863228,
-0.030740056186914444,
-0.04694638401269913,
-0.05924609676003456,
0.012249989435076714,
-0.0071121081709861755,
-0.014395568519830704,
0.05053829774260521,
0.026467682793736458,
0... | 0.161113 |
cloud sandboxes that get destroyed after the session ends, without having to tell the agent to explicitly `scp` or `modal volume put` every artifact. ### Docker Volume Mounts When using the Docker backend, `docker\_volumes` lets you share host directories with the container. Each entry uses standard Docker `-v` syntax:... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/configuration.md | main | hermes-agent | [
0.038033291697502136,
0.02874794788658619,
0.008334423415362835,
0.01329998578876257,
0.07117890566587448,
-0.023930320516228676,
-0.013994034379720688,
-0.03991961479187012,
0.04634552076458931,
0.05002667382359505,
-0.02878303825855255,
0.015975292772054672,
0.05803581327199936,
-0.00714... | 0.043074 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.