> ## Documentation Index > Fetch the complete documentation index at: https://docs.egma.ai/llms.txt > Use this file to discover all available pages before exploring further. # Troubleshooting A failed simulation shows its reason on the run page. Find the message below. ## Egma could not start the bot | Message starts with | What to do | | - | - | | `Pipecat Cloud refused the start request for agent "…" (HTTP 401)` or `(HTTP 403)` | The public API key was not accepted. Use a `pk_` key of the Pipecat Cloud organization that deploys the agent. Add the connection again with the same agent name and modality to replace the stored key. | | `Pipecat Cloud has no agent named "…"` | Check the connection's agent name against `agent_name` in `pcc-deploy.toml`, and check that the key belongs to the same organization. | | `Pipecat Cloud had no free capacity for agent "…"` | Every start request for 120 seconds was answered HTTP 429. Raise the agent's `max_agents` in Pipecat Cloud, or run fewer simulations at once with a lower `--concurrency`. | | `Pipecat Cloud refused the start request for agent "…" (HTTP …)` | Check the agent's deployment status in Pipecat Cloud. | | `Egma could not reach Pipecat Cloud to start agent "…"` | Egma retried for 120 seconds. Try again later. A self-hosted Egma instance needs outbound access to Pipecat Cloud and Daily. | | `Egma could not reach your start URL` | Check that your starter is running and that its URL resolves and answers over HTTPS. For a `trycloudflare.com` address, run `egma agent dev` on the machine that runs your bot, and start your bot's development runner there. | | `your start URL (…) refused the start request (HTTP 401)` or `(HTTP 403)` | Check the connection's auth headers against what your starter expects. | | `your start URL (…) answered HTTP 429` | Your starter refused every start request for 120 seconds. Give it more capacity or run fewer simulations at once. | | `your start URL (…) refused the start request (HTTP …)` | Check your starter's logs for that request. | | `the answer to the start request carried no usable Daily room` | Answer with a 2xx JSON object whose `dailyRoom` is an `https://…daily.co/…` URL. Do not redirect. | ## The bot did not get ready Egma waits up to 120 seconds from the first start request for your bot to join the Daily room, report to Egma, and publish audio (voice) or send RTVI `bot-ready` (chat). | Message starts with | What to do | | - | - | | `your bot did not join within 120 seconds` | The start was accepted, but no bot joined. Check the bot's logs, for example with `pipecat cloud agent logs `. A `NotReported` error there means the bot could not reach Egma: check `EGMA_URL` and `EGMA_API_KEY` where the bot runs (on Pipecat Cloud, in the agent's secret set, then redeploy). A crash at start has the same effect. | | `your bot joined but did not report to Egma` | Add `await simulation(worker, runner_args)` from `egma.pipecat` before the runner starts the worker, and install `egma[pipecat]`. Set `EGMA_URL` and `EGMA_API_KEY` where the bot runs. On Pipecat Cloud, add them to the agent's secret set and redeploy. | | `the test mocks "…", and this is a Pipecat Flows function` | Egma does not mock Pipecat Flows functions. A simulation whose test mocks one fails when the flow first offers that function. Remove it from the test's mock tools. Flows functions that are not mocked run normally and are recorded. | | `your bot reported to Egma and Egma refused the report` | Read the reason in parentheses. For a protocol version, upgrade the `egma` package. | | `your bot has RTVI turned off` | Chat simulations need RTVI. Remove `enable_rtvi=False` from your `PipelineWorker`. Voice simulations do not need RTVI. | | `your bot joined and reported to Egma but published no audio track` | Check that the pipeline ends in `transport.output()` and that `DailyParams` has `audio_out_enabled=True`. | A message that ends with `the simulation's configured …s duration expired during startup` means the simulation's duration limit ran out while Egma was still waiting. Fix the cause named before it. ## Errors in your bot's logs * `NotReported` comes from `simulation()` when the bot cannot complete its report to Egma. The bot does not start. Check `EGMA_URL`, `EGMA_API_KEY`, and network access from the bot to Egma. * `Egma could not answer the mocked tool "…"` is the result a mocked tool returns when it cannot reach Egma during a simulation. The real tool did not run. ## The CLI * `Egma answered with an incomplete Agent. Check that this Egma platform is up to date.` comes from egma-cli 0.7.0 or older in a project that has a Pipecat agent. Update the CLI with `npm install --global egma-cli@latest`. Projects without a Pipecat agent are not affected. ## On this machine * Keep `egma agent dev` and your bot's development runner running while simulations run. * Put `DAILY_API_KEY` in your bot's environment, and leave `DAILY_ROOM_URL` unset. * When `egma agent dev` says the tunnel's address is not in public DNS yet, a simulation that starts in the next few minutes can fail to reach it. ## Read the evidence The simulation's transcript is your bot's own record: every turn, and every tool call with its arguments and result. Calls answered by a test mock are marked **mocked**; other tools run for real and are recorded too. The voice recording keeps what the simulated caller heard. To capture real conversations, add the `monitor` line as described in [Monitor a Pipecat agent](/docs/integrations/pipecat/monitor). This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.