# General **URL:** https://community.crewai.com/c/general/4.md?page=1 [Latest](https://community.crewai.com/latest.md) · [Categories](https://community.crewai.com/categories.md) · [Tags](https://community.crewai.com/tags.md) **Page:** 2 --- ## [Pattern: giving each agent a hard, server-enforced spend ceiling — instead of trusting the prompt to behavePattern: giving each agent a hard, server-enforced spend ceiling instead of trusting the prompt to behave](https://community.crewai.com/t/pattern-giving-each-agent-a-hard-server-enforced-spend-ceiling-instead-of-trusting-the-prompt-to-behavepattern-giving-each-agent-a-hard-server-enforced-spend-ceiling-instead-of-trusting-the-prompt-to-behave/7727)
**Author:** [@Debo\_Jolaosho](https://community.crewai.com/u/Debo_Jolaosho)\ **Replies:** 0\ **Last updated:** [August 16, 2026, 7:39pm UTC](https://community.crewai.com/t/pattern-giving-each-agent-a-hard-server-enforced-spend-ceiling-instead-of-trusting-the-prompt-to-behavepattern-giving-each-agent-a-hard-server-enforced-spend-ceiling-instead-of-trusting-the-prompt-to-behave/7727 "2026-08-16T19:39:01Z")
Ran into this building agent workflows that touch real payment methods / paid APIs, curious if others have hit the same wall. The failure mode isn’t malicious agents — it’s boring stuff: a retry loop that doesn’t back o… --- ## [I measured where a tool-using agent's input tokens actually go — fixed overhead dwarfed the user turn](https://community.crewai.com/t/i-measured-where-a-tool-using-agents-input-tokens-actually-go-fixed-overhead-dwarfed-the-user-turn/7677)
**Author:** [@reelwiselab](https://community.crewai.com/u/reelwiselab)\ **Replies:** 1\ **Last updated:** [August 16, 2026, 12:17pm UTC](https://community.crewai.com/t/i-measured-where-a-tool-using-agents-input-tokens-actually-go-fixed-overhead-dwarfed-the-user-turn/7677 "2026-08-16T12:17:40Z")
A cost lesson from putting LLM agents into production that I wish I’d measured on day one: once you give a model tools/functions and retrieved context, your token bill stops tracking “how much the user said” and starts t… --- ## [Field note: treating your agent's token spend like a CI cost gate](https://community.crewai.com/t/field-note-treating-your-agents-token-spend-like-a-ci-cost-gate/7725)
**Author:** [@wartzarbee2](https://community.crewai.com/u/wartzarbee2)\ **Replies:** 0\ **Last updated:** [August 10, 2026, 3:30pm UTC](https://community.crewai.com/t/field-note-treating-your-agents-token-spend-like-a-ci-cost-gate/7725 "2026-08-10T15:30:53Z")
Field note: treat your agent’s token spend like a CI cost gate Running multi-agent pipelines, the failure mode that keeps biting me isn’t a crash — it’s a silent cost blowup. A run finishes, produces something plausible… --- ## [Agents timing out and failing to extract structured data from custom web portals](https://community.crewai.com/t/agents-timing-out-and-failing-to-extract-structured-data-from-custom-web-portals/7724)
**Author:** [@felixelle768](https://community.crewai.com/u/felixelle768)\ **Replies:** 0\ **Last updated:** [August 8, 2026, 7:21pm UTC](https://community.crewai.com/t/agents-timing-out-and-failing-to-extract-structured-data-from-custom-web-portals/7724 "2026-08-08T19:21:03Z")
Hi everyone, I am working on an automated multi-agent workflow where one of my researcher agents is supposed to pull billing breakdowns and reference data directly from online portals. However, I am hitting a wall when … --- ## [Add Custom MCP server with OAuth: default form fields is an invalid config](https://community.crewai.com/t/add-custom-mcp-server-with-oauth-default-form-fields-is-an-invalid-config/7720)
**Author:** [@sw0rdf1sh](https://community.crewai.com/u/sw0rdf1sh)\ **Replies:** 0\ **Last updated:** [August 5, 2026, 10:15am UTC](https://community.crewai.com/t/add-custom-mcp-server-with-oauth-default-form-fields-is-an-invalid-config/7720 "2026-08-05T10:15:31Z")
When adding a new Custom MCP Server and OAuth is selected as the authentication method, the default Token Auth Method is “Standard (POST body)”, even though Client Secret is not a mandatory form field. Without a Client … --- ## [How are you isolating credentials + controlling egress when running CrewAI agents with tool access?](https://community.crewai.com/t/how-are-you-isolating-credentials-controlling-egress-when-running-crewai-agents-with-tool-access/7705)
**Author:** [@wartzarbee2](https://community.crewai.com/u/wartzarbee2)\ **Replies:** 0\ **Last updated:** [July 31, 2026, 9:21pm UTC](https://community.crewai.com/t/how-are-you-isolating-credentials-controlling-egress-when-running-crewai-agents-with-tool-access/7705 "2026-07-31T21:21:26Z")
When I run a CrewAI crew that has tool access — shell, HTTP, a code interpreter — I keep running into the same worry: the crew logic and the trust boundary around it are two different problems, and CrewAI (rightly) only … --- ## [Failed to start execution (500)](https://community.crewai.com/t/failed-to-start-execution-500/7698)
**Author:** [@skerdi](https://community.crewai.com/u/skerdi)\ **Replies:** 1\ **Last updated:** [July 31, 2026, 9:14pm UTC](https://community.crewai.com/t/failed-to-start-execution-500/7698 "2026-07-31T21:14:01Z")
Studio automation fails on Run with ‘Failed to start execution (500)’. The identical crew runs fine locally with gpt-4o-mini. LLM connection (OpenAI) is valid, $5 credit, no agent warnings. Traces stay on ‘waiting for e… --- ## [HDP - cryptographic human authorization for CrewAI agents](https://community.crewai.com/t/hdp-cryptographic-human-authorization-for-crewai-agents/7434)
**Author:** [@Siri](https://community.crewai.com/u/Siri)\ **Replies:** 1\ **Last updated:** [July 31, 2026, 7:21am UTC](https://community.crewai.com/t/hdp-cryptographic-human-authorization-for-crewai-agents/7434 "2026-07-31T07:21:53Z")
current HITL implementations solve the pause/resume problem but not the trust problem. but there’s no way for the agent to verify the approval actually came from an authorized human vs an injected signal. HDP (human de… --- ## [At-least-once tool calls: retries can double-fire your side-effecting tools in a crew](https://community.crewai.com/t/at-least-once-tool-calls-retries-can-double-fire-your-side-effecting-tools-in-a-crew/7697)
**Author:** [@reelwiselab](https://community.crewai.com/u/reelwiselab)\ **Replies:** 0\ **Last updated:** [July 31, 2026, 7:01am UTC](https://community.crewai.com/t/at-least-once-tool-calls-retries-can-double-fire-your-side-effecting-tools-in-a-crew/7697 "2026-07-31T07:01:46Z")
A quieter multi-agent failure mode than token cost: when an agent retries a tool call, non-idempotent tools double-fire. This one cost us real duplicate side effects before we caught it, so writing it down. The setup ev… --- ## [How use web content credibility of web pages in crewai?](https://community.crewai.com/t/how-use-web-content-credibility-of-web-pages-in-crewai/6923)
**Author:** [@Andy60rm](https://community.crewai.com/u/Andy60rm)\ **Replies:** 1\ **Last updated:** [July 29, 2026, 8:09am UTC](https://community.crewai.com/t/how-use-web-content-credibility-of-web-pages-in-crewai/6923 "2026-07-29T08:09:30Z")
how use web content credibility of web pages in crewai? with custom tools and external api? any suggestions? thx --- ## [A cost gotcha in multi-agent crews: tool schemas are re-billed every turn](https://community.crewai.com/t/a-cost-gotcha-in-multi-agent-crews-tool-schemas-are-re-billed-every-turn/7690)
**Author:** [@reelwiselab](https://community.crewai.com/u/reelwiselab)\ **Replies:** 0\ **Last updated:** [July 28, 2026, 10:54pm UTC](https://community.crewai.com/t/a-cost-gotcha-in-multi-agent-crews-tool-schemas-are-re-billed-every-turn/7690 "2026-07-28T22:54:47Z")
When you scale a crew from 2–3 agents to a dozen, the cost curve often bends up faster than the number of agents — and a big, easy-to-miss driver is tool/function schemas being re-billed as input tokens on every turn. T… --- ## [Starting Projects](https://community.crewai.com/t/starting-projects/7046)
**Author:** [@Maaz\_Akhtar](https://community.crewai.com/u/Maaz_Akhtar)\ **Replies:** 2\ **Last updated:** [July 25, 2026, 5:45pm UTC](https://community.crewai.com/t/starting-projects/7046 "2026-07-25T17:45:37Z")
What projects would you all recommend to a crewAI learner who has reviewed the theory and now wants to build his own project from scratch using CrewAI docs and crewGPT? --- ## [Seeking insights from developers building with Crew Agents](https://community.crewai.com/t/seeking-insights-from-developers-building-with-crew-agents/7507)
**Author:** [@oabolade](https://community.crewai.com/u/oabolade)\ **Replies:** 1\ **Last updated:** [July 23, 2026, 9:03am UTC](https://community.crewai.com/t/seeking-insights-from-developers-building-with-crew-agents/7507 "2026-07-23T09:03:35Z")
Hi there, My name is Olasile, I am a product manager at Amazon working on cloud and data infrastructure I am seeking insights from people building agentic workflows with CrewAI. Specifically I’m looking to chat with AI… --- ## [Trying to understand CrewAI- is this really about agents, or just managing LLM calls?](https://community.crewai.com/t/trying-to-understand-crewai-is-this-really-about-agents-or-just-managing-llm-calls/7244)
**Author:** [@Kami](https://community.crewai.com/u/Kami)\ **Replies:** 6\ **Last updated:** [July 23, 2026, 8:59am UTC](https://community.crewai.com/t/trying-to-understand-crewai-is-this-really-about-agents-or-just-managing-llm-calls/7244 "2026-07-23T08:59:36Z")
I am a faculty member at a US University. I got interested in crewai because we have an upcoming project where we evaluate our courses and compare them against industry trends and so on for relevancy discussions. I tho… --- ## [Statefold gives CrewAI agents durable, replayable and inspectable state using an append-only event log](https://community.crewai.com/t/statefold-gives-crewai-agents-durable-replayable-and-inspectable-state-using-an-append-only-event-log/7657)
**Author:** [@Joshua\_Nishanth](https://community.crewai.com/u/Joshua_Nishanth)\ **Replies:** 0\ **Last updated:** [July 19, 2026, 5:31am UTC](https://community.crewai.com/t/statefold-gives-crewai-agents-durable-replayable-and-inspectable-state-using-an-append-only-event-log/7657 "2026-07-19T05:31:55Z")
Hi everyone, I’m sharing Statefold, an open-source, framework-agnostic state and observability platform for AI agents. GitHub: GitHub - ioteverythin/statefold: Framework-agnostic, event-sourced state platform for AI ag… --- ## [Can someone please explain hierarchical crews and how do they work especially regarding tasks?](https://community.crewai.com/t/can-someone-please-explain-hierarchical-crews-and-how-do-they-work-especially-regarding-tasks/7629)
**Author:** [@Sarthak\_Rawat](https://community.crewai.com/u/Sarthak_Rawat)\ **Replies:** 0\ **Last updated:** [July 14, 2026, 6:35am UTC](https://community.crewai.com/t/can-someone-please-explain-hierarchical-crews-and-how-do-they-work-especially-regarding-tasks/7629 "2026-07-14T06:35:32Z")
Hello I am trying to build my first hierarchical crew but I do not understand how task delegation works here. So we define bunch of tasks and manager decides who does tasks but what about order. Documents only say about … --- ## [Agents that write code, debug themselves, and keep a notebook between runs (the Stratix Cup is worth a look)](https://community.crewai.com/t/agents-that-write-code-debug-themselves-and-keep-a-notebook-between-runs-the-stratix-cup-is-worth-a-look/7602)
**Author:** [@ajdevrel](https://community.crewai.com/u/ajdevrel)\ **Replies:** 0\ **Last updated:** [June 24, 2026, 5:48pm UTC](https://community.crewai.com/t/agents-that-write-code-debug-themselves-and-keep-a-notebook-between-runs-the-stratix-cup-is-worth-a-look/7602 "2026-06-24T17:48:05Z")
Sharing this in case anyone here likes watching agents get stress-tested in public: the Stratix Cup, run by LayerLens (an AI eval company). The premise is a coding benchmark dressed up as a football tournament. 16 front… --- ## [Show & Tell: Orchid – Local-first record and replay for AI agent debug & testing](https://community.crewai.com/t/show-tell-orchid-local-first-record-and-replay-for-ai-agent-debug-testing/7601)
**Author:** [@marioguerra](https://community.crewai.com/u/marioguerra)\ **Replies:** 0\ **Last updated:** [June 24, 2026, 4:09pm UTC](https://community.crewai.com/t/show-tell-orchid-local-first-record-and-replay-for-ai-agent-debug-testing/7601 "2026-06-24T16:09:24Z")
Hey everyone, just released Orchid, (Orchestration interactive debugger) a zero-instrumentation proxy that captures every API & LLM call in your agentic pipeline, then lets you inspect and replay the entire run locally, … --- ## [Cant use MCP server in Flows](https://community.crewai.com/t/cant-use-mcp-server-in-flows/7556)
**Author:** [@Jai\_Pande](https://community.crewai.com/u/Jai_Pande)\ **Replies:** 1\ **Last updated:** [June 23, 2026, 1:54pm UTC](https://community.crewai.com/t/cant-use-mcp-server-in-flows/7556 "2026-06-23T13:54:29Z")
Hi @maxmoura/ @Pratik\_Kumar , Any idea how to use MCP servers in Flows, its working perfectly in Crews but not in flows, only when i deploy it to AMP it gives me this MCPServerHTTP is not serializable issue --- ## [Academic survey: How do people use and debug AI agents for multi-step tasks?](https://community.crewai.com/t/academic-survey-how-do-people-use-and-debug-ai-agents-for-multi-step-tasks/7590)
**Author:** [@zekun\_wu](https://community.crewai.com/u/zekun_wu)\ **Replies:** 0\ **Last updated:** [June 19, 2026, 1:32pm UTC](https://community.crewai.com/t/academic-survey-how-do-people-use-and-debug-ai-agents-for-multi-step-tasks/7590 "2026-06-19T13:32:42Z")
Hi everyone, I’m running a short academic survey about how people use chat-based AI agents for multi-step tasks, and how this compares with reusing or editing workflow-style automations. The survey asks about your expe… --- ## [Budget guardrail for crews](https://community.crewai.com/t/budget-guardrail-for-crews/7583)
**Author:** [@achris7](https://community.crewai.com/u/achris7)\ **Replies:** 0\ **Last updated:** [June 16, 2026, 9:58pm UTC](https://community.crewai.com/t/budget-guardrail-for-crews/7583 "2026-06-16T21:58:40Z")
I added a 5-line budget guardrail to my CrewAI crew repo here GitHub - Floe-Labs/floe-guard: Open-source budget guardrail for AI agents — hard-stop before a runaway loop burns your bill. Local, framework-agnostic, no acc… --- ## [CrewAI Examples Repo](https://community.crewai.com/t/crewai-examples-repo/2791)
**Author:** [@Nate\_Mar](https://community.crewai.com/u/Nate_Mar)\ **Replies:** 3\ **Last updated:** [June 12, 2026, 2:38am UTC](https://community.crewai.com/t/crewai-examples-repo/2791 "2026-06-12T02:38:47Z")
Hi! Massive shoutout to the folks who put together all of these crew ai examples. Super cool. I submitted a PR to fix the surprise example and I saw there were a few other prs in the repo from the last month. No rush obv… --- ## [Null-drift: A bare-metal O(1) Memory Backend for CrewAI (No VectorDB bloat)](https://community.crewai.com/t/null-drift-a-bare-metal-o-1-memory-backend-for-crewai-no-vectordb-bloat/7579)
**Author:** [@ansh](https://community.crewai.com/u/ansh)\ **Replies:** 0\ **Last updated:** [June 11, 2026, 7:44pm UTC](https://community.crewai.com/t/null-drift-a-bare-metal-o-1-memory-backend-for-crewai-no-vectordb-bloat/7579 "2026-06-11T19:44:29Z")
Got tired of standard vectordbs eating up disk space and causing python bloat when running continuous local agents. so i gutted the db approach and built a headless rust daemon (null-drift). ​instead of storing exact te… --- ## [Adding OrcaRouter as a native LLM provider — PR open, looking for review](https://community.crewai.com/t/adding-orcarouter-as-a-native-llm-provider-pr-open-looking-for-review/7540)
**Author:** [@Song\_Jinhao](https://community.crewai.com/u/Song_Jinhao)\ **Replies:** 0\ **Last updated:** [May 26, 2026, 3:52am UTC](https://community.crewai.com/t/adding-orcarouter-as-a-native-llm-provider-pr-open-looking-for-review/7540 "2026-05-26T03:52:51Z")
Hi team :waving\_hand: I opened a PR that adds OrcaRouter as a native OpenAI-compatible provider for CrewAI, and figured I’d start a thread here in case folks want to discuss or share thoughts: … --- ## [Cant' find the module](https://community.crewai.com/t/cant-find-the-module/7531)
**Author:** [@Ruid](https://community.crewai.com/u/Ruid)\ **Replies:** 0\ **Last updated:** [May 21, 2026, 10:40pm UTC](https://community.crewai.com/t/cant-find-the-module/7531 "2026-05-21T22:40:12Z")
I have this error. Why? ModuleNotFoundError: No module named ‘THE NAME OF THE PROJECT’ An error occurred while running the crew: Command ‘\[‘uv’, ‘run’, ‘run\_crew’\]’ returned non-zero exit status 1. --- ## [Client session memory](https://community.crewai.com/t/client-session-memory/708)
**Author:** [@rolmovel](https://community.crewai.com/u/rolmovel)\ **Replies:** 4\ **Last updated:** [May 21, 2026, 7:42pm UTC](https://community.crewai.com/t/client-session-memory/708 "2026-05-21T19:42:15Z")
Hi, I would like to build a kind of personal shopper under rest technology using the crewai framework… I’m thinking about the better approach to agent memory keeping in mind next: Agent should have a session memory for… --- ## [Own CrewAI Ollama Cloud Provider](https://community.crewai.com/t/own-crewai-ollama-cloud-provider/7511)
**Author:** [@Hackbard](https://community.crewai.com/u/Hackbard)\ **Replies:** 0\ **Last updated:** [May 15, 2026, 9:38am UTC](https://community.crewai.com/t/own-crewai-ollama-cloud-provider/7511 "2026-05-15T09:38:24Z")
My Problem CrewAI’s built-in Ollama support has a catch: it routes through the OpenAI-compatible shim (/v1/chat/completions). That’s not the real Ollama API — it’s a translation layer bolted onto Ollama to make it look … --- ## [CrewAI Flow Infinite Loop - Steps Repeating Endlessly](https://community.crewai.com/t/crewai-flow-infinite-loop-steps-repeating-endlessly/6780)
**Author:** [@Hiba-Chaabnia](https://community.crewai.com/u/Hiba-Chaabnia)\ **Replies:** 2\ **Last updated:** [May 5, 2026, 10:45pm UTC](https://community.crewai.com/t/crewai-flow-infinite-loop-steps-repeating-endlessly/6780 "2026-05-05T22:45:30Z")
Hi everyone, I’m experiencing instability with CrewAI Flow where certain steps are getting stuck in an infinite loop and repeating non-stop. Issue Details: Steps keep repeating even after progressing to the next sta… --- ## [Show & Tell: MCP Server for Brazilian Gov Procurement Data (stdio + MCPServerAdapter working example)](https://community.crewai.com/t/show-tell-mcp-server-for-brazilian-gov-procurement-data-stdio-mcpserveradapter-working-example/7502)
**Author:** [@guimaster97](https://community.crewai.com/u/guimaster97)\ **Replies:** 0\ **Last updated:** [May 4, 2026, 10:59pm UTC](https://community.crewai.com/t/show-tell-mcp-server-for-brazilian-gov-procurement-data-stdio-mcpserveradapter-working-example/7502 "2026-05-04T22:59:09Z")
Hey CrewAI community! Sharing a working example of the new MCPServerAdapter integration that might be useful for anyone building sales or prospecting agents. The Problem: Brazilian government procurement portals (wher… --- ## [Agent eval tools?](https://community.crewai.com/t/agent-eval-tools/7453)
**Author:** [@Derfe](https://community.crewai.com/u/Derfe)\ **Replies:** 1\ **Last updated:** [May 1, 2026, 4:20pm UTC](https://community.crewai.com/t/agent-eval-tools/7453 "2026-05-01T16:20:18Z")
Hiya all, I’m curious how people here are evaluating their CrewAI agents once things start getting a bit more complex? I mean stuff like: output quality if tool usage is actually improving regressions after changes c… [Previous page](https://community.crewai.com/c/general/4.md) [Next page](https://community.crewai.com/c/general/4.md?page=2)