Port over stuck detector

[ryanhoangt]

Fix #281

• There's one scenario for stuck detection, which happens when the condenser fails to trim events when context exceeding error occurs. This can be added in a separate PR after Trigger condenser when LLM throws context window error #282 is implemented.

[github-actions]

Coverage Report

File	Stmts	Miss	Cover	Missing
openhands
__init__.py	1	0	100%
openhands/agent_server
__init__.py	16	16	0%	1, 3–5, 9–12, 20–22, 25, 27–29, 31
__main__.py	15	15	0%	1, 3, 6–8, 11, 14, 22, 24–27, 29, 38–39
api.py	20	20	0%	1–2, 4, 6, 9, 12, 15, 18, 24–28, 31–32, 36–37, 40–42
config.py	42	42	0%	1–4, 6, 10–11, 14, 17, 21, 27–28, 31, 34, 38, 41, 47, 54, 61, 67, 73, 77, 79–80, 82, 85–87, 89–90, 93–96, 98, 101, 104, 107, 109–110, 113–114
conversation_router.py	50	50	0%	3–4, 6, 8, 11, 18, 21–22, 28–29, 48–50, 55–56, 63–64, 67–68, 70–73, 76–77, 82–84, 90–91, 95–96, 99, 102–106, 109, 112–116, 119–124
conversation_service.py	195	195	0%	1–6, 8, 10–12, 19–22, 25, 28–29, 35–39, 41–48, 50, 57–58, 61–63, 69–70, 72, 75–82, 85–86, 89–93, 96–98, 100–103, 105, 107, 112–113, 115–117, 120–121, 123, 125, 127, 132–136, 140, 144–148, 151–152, 159–160, 173–177, 180, 182–183, 185–191, 193–199, 201–210, 212–215, 217–225, 230–231, 234–236, 238–242, 244, 251–253, 261–263, 265–266, 269–274, 276, 278, 280–281, 283, 285–286, 288, 290–291, 293–294, 297–299, 302, 308–311, 318–319, 323–327, 329, 334, 337, 340, 342–343, 345, 349–351
event_router.py	86	86	0%	5–8, 10, 18, 20, 23, 31–32, 35–37, 42–43, 65–70, 73–74, 84–88, 91–92, 94–100, 103–104, 109–113, 119–120, 122–127, 130, 133, 137–141, 147–148, 152–156, 159–168, 170, 173–175, 177–182
event_service.py	126	126	0%	1–4, 6, 13–15, 21–22, 25–26, 32–36, 38–40, 42–45, 47–50, 53, 56, 58, 65–66, 69–71, 73, 78–79, 82–85, 88–89, 92–96, 99–101, 103–106, 108–109, 111–113, 115, 120–121, 123–125, 127, 132–133, 135, 137, 139–143, 145–152, 154–155, 157–158, 160, 162–166, 182–183, 185, 187–190, 192–194, 196, 198–201, 203–207, 209–212, 214–216, 218–220
middleware.py	26	26	0%	1, 3–6, 9, 14–15, 23–26, 29–30, 33–34, 37, 46–48, 50, 53–57
models.py	50	50	0%	1–4, 6, 8, 12, 19–20, 27, 30, 33–36, 39, 42–43, 46, 52–54, 59–61, 64, 72, 77, 80, 87, 90–93, 96, 99, 102–104, 107–109, 112, 115–116, 119–120, 123–125
pub_sub.py	35	35	0%	1–4, 6–7, 10, 13–15, 18, 22–23, 30, 32, 39–42, 44, 51–54, 56, 59, 61, 68–72, 74–75, 78
utils.py	28	28	0%	1–4, 6, 9, 11, 14–18, 21, 25–30, 33–36, 39–41, 44, 47
openhands/sdk
__init__.py	16	2	87%	27–28
logger.py	73	21	71%	33, 57, 64–67, 69–71, 124, 129–131, 134–135, 141–143, 150, 155–156
openhands/sdk/agent
__init__.py	4	0	100%
agent.py	195	47	75%	64, 71, 78, 82, 99, 117, 124–125, 130–131, 207–208, 210–212, 214–216, 258, 272, 295, 334, 339–341, 344–345, 348, 378–380, 384–386, 396–397, 401–404, 411–412, 416, 420–421, 452, 459
base.py	94	7	92%	113, 127, 135–137, 153, 173
spec.py	15	0	100%
openhands/sdk/context
__init__.py	4	0	100%
agent_context.py	57	2	96%	146, 152
manager.py	3	3	0%	1, 4–5
view.py	97	1	98%	90
openhands/sdk/context/condenser
__init__.py	5	0	100%
base.py	21	0	100%
llm_summarizing_condenser.py	39	3	92%	44–46
no_op_condenser.py	6	0	100%
pipeline_condenser.py	13	6	53%	45–50
openhands/sdk/context/microagents
__init__.py	4	0	100%
exceptions.py	5	0	100%
microagent.py	143	25	82%	130, 133–136, 218–221, 229, 251–252, 257–258, 260, 264, 271–273, 281–283, 337, 339–340
types.py	21	0	100%
openhands/sdk/context/prompts
__init__.py	2	0	100%
prompt.py	30	5	83%	12, 15, 24, 44–45
openhands/sdk/conversation
__init__.py	8	0	100%
conversation.py	129	17	86%	95, 110, 138, 146–148, 152–153, 202, 204–205, 207, 225, 307–308, 316–317
event_store.py	101	8	92%	50–51, 60, 67, 72–73, 129, 142
persistence_const.py	5	0	100%
secrets_manager.py	41	1	97%	107
serialization_diff.py	0	0	100%
state.py	101	5	95%	141, 164, 200–202
stuck_detector.py	101	20	80%	88, 92–93, 155, 164–165, 168, 193, 196, 202, 209–211, 224, 233, 257–258, 264–265, 271
types.py	6	0	100%
visualizer.py	94	6	93%	90, 147, 169, 186, 218, 220
openhands/sdk/event
__init__.py	7	0	100%
base.py	74	8	89%	55, 75, 87–88, 94, 97–98, 100
condenser.py	28	7	75%	37, 39, 41–45
llm_convertible.py	181	16	91%	54, 64–65, 70–71, 251, 285–286, 291, 299, 340–341, 346, 379–380, 385
metric_events.py	13	0	100%
types.py	7	0	100%
user_action.py	12	1	91%	21
utils.py	12	0	100%
openhands/sdk/io
__init__.py	4	0	100%
base.py	14	4	71%	7, 11, 15, 19
local.py	56	16	71%	43–44, 58, 66–78
memory.py	43	4	90%	16, 20, 53–54
openhands/sdk/llm
__init__.py	8	0	100%
exceptions.py	36	0	100%
llm.py	399	97	75%	229, 234, 253–254, 286, 349, 355–356, 451, 464–465, 470–471, 473–474, 477–479, 484–486, 490–492, 513–516, 523, 541–542, 570, 576–577, 623, 672, 689–690, 699, 710, 731, 733–738, 740–757, 760–764, 766–767, 773–782, 786–797, 810, 824, 829
llm_registry.py	38	0	100%
message.py	109	4	96%	96, 99, 222–223
metadata.py	15	0	100%
openhands/sdk/llm/mixins
fn_call_converter.py	343	94	72%	74, 343, 345, 349, 367, 369, 375, 381, 383, 422, 424, 426, 428, 433–434, 481–482, 518–520, 522, 524, 545–547, 553, 575, 601–602, 610–613, 615, 617, 639, 648, 656, 701–704, 708–711, 723, 727, 738, 748, 797–798, 800, 829, 833, 859, 867, 870–871, 876, 905–908, 912–913, 918–919, 924, 973–974, 980, 994, 1006, 1008–1009, 1012–1014, 1016–1017, 1023–1025, 1027–1028, 1030, 1032, 1036, 1038, 1043, 1045–1046, 1049
non_native_fc.py	39	3	92%	64, 75, 91
openhands/sdk/llm/utils
metrics.py	111	3	97%	17, 117, 311
model_features.py	40	0	100%
retry_mixin.py	50	11	78%	47, 50, 64, 86, 90, 94–95, 105, 110–111, 116
telemetry.py	136	15	88%	71, 94, 99–100, 112–113, 120, 134, 199, 216, 222, 229, 232, 234, 241
unverified_models.py	69	4	94%	45–46, 51, 73
verified_models.py	5	0	100%
openhands/sdk/mcp
__init__.py	5	0	100%
client.py	26	6	76%	48–49, 62–63, 72–73
definition.py	48	16	66%	55, 75–80, 82–90
tool.py	40	13	67%	36–39, 43, 46, 49–52, 101–102, 107
utils.py	30	4	86%	23–24, 27, 30
openhands/sdk/preset
__init__.py	0	0	100%
default.py	20	12	40%	13, 15, 22, 28–29, 31–33, 35–36, 43, 45
openhands/sdk/security
__init__.py	2	0	100%
analyzer.py	36	8	77%	41, 74, 76–77, 79–80, 82, 85
llm_analyzer.py	9	0	100%
risk.py	12	2	83%	21, 33
openhands/sdk/tool
__init__.py	5	0	100%
schema.py	115	11	90%	23–25, 27, 36, 216–219, 239, 254
spec.py	15	0	100%
tool.py	92	10	89%	66, 107, 167, 170–176
openhands/sdk/tool/builtins
__init__.py	4	0	100%
finish.py	26	1	96%	33
think.py	32	13	59%	24, 27–28, 31, 33–37, 39, 51, 57, 74
openhands/sdk/utils
__init__.py	3	0	100%
async_executor.py	52	7	86%	39, 55–56, 84, 88, 102–103
async_utils.py	12	0	100%
discriminated_union.py	168	22	86%	119–127, 141–142, 236, 329, 356, 399–401, 418, 451, 464, 471, 474
json.py	28	28	0%	1–3, 5, 7–8, 11, 14–21, 25, 28, 30–31, 34, 37–38, 40, 43, 45–48
protocol.py	3	0	100%
pydantic_diff.py	57	15	73%	36, 44, 50–58, 60–62, 65
truncate.py	10	0	100%
visualize.py	17	4	76%	14–16, 22
openhands/tools
__init__.py	17	3	82%	29–30, 79
openhands/tools/browser_use
__init__.py	3	3	0%	3, 37, 40
definition.py	109	109	0%	3, 5, 7–10, 14, 17, 20–22, 26–29, 31, 35, 37–38, 40, 46, 49–50, 55, 68, 83, 86–88, 101, 104, 107, 113, 125, 140, 143–145, 158, 161, 164, 167, 179, 194, 197–199, 212, 215, 221, 230, 245, 248–250, 263, 266, 270, 276, 281, 296, 299–301, 314, 317, 323, 332, 347, 350–352, 365, 368, 371, 377, 392, 395–397, 410, 413, 416, 422, 437, 440–442, 455, 458, 464, 472, 487, 490–492, 505, 508, 513, 521, 536, 539–541, 551, 558–561
impl.py	113	113	0%	3–4, 6–8, 12, 15, 18, 25, 28, 33–34, 36, 38, 42, 44, 58–59, 61–70, 73–82, 84–85, 87–91, 93, 95, 97–98, 101, 103–104, 106, 108–109, 112, 114–115, 117, 119–120, 122, 124–125, 127, 129, 131–132, 134–137, 140–141, 144, 146, 148, 151, 153–154, 156, 158–159, 161, 163–164, 167, 169–170, 174, 176–180, 182, 184–189, 191, 193, 195–197, 200, 202, 204–207
server.py	45	45	0%	1, 4, 10–11, 13–14, 16–17, 20–21, 24–25, 28, 30–32, 34–35, 38–39, 41, 44, 47–48, 51, 54–55, 57–61, 64–66, 68, 73–78, 80, 87, 89
openhands/tools/execute_bash
__init__.py	4	0	100%
constants.py	9	0	100%
definition.py	95	61	35%	38, 41, 44–45, 47, 50–52, 54–56, 58, 88, 92–101, 106, 109–111, 114, 116–118, 120, 124–125, 128–130, 132–133, 136–139, 143–145, 150, 154–156, 159–161, 165–166, 168, 245, 247–248, 251, 261
impl.py	40	22	45%	52–55, 57–58, 60–62, 64–67, 69, 72, 82–83, 86–89, 91
metadata.py	50	23	54%	67–73, 77–78, 83, 85, 87–96, 100–101
openhands/tools/execute_bash/terminal
__init__.py	6	0	100%
factory.py	49	28	42%	24–25, 30, 32, 35, 37–38, 44–46, 74–77, 79–83, 87–89, 91, 97, 107, 111–113
interface.py	69	18	73%	43, 52, 62, 71, 76, 85, 94, 99, 104, 112, 145, 157, 162, 171, 180, 185, 191, 193
subprocess_terminal.py	236	203	13%	33–34, 50–53, 56, 59–61, 67–68, 71–74, 76, 79, 81–83, 97–100, 102, 105–106, 109, 112–113, 117–118, 124–126, 128, 131–132, 134, 138–139, 141–142, 144–150, 152–158, 161–166, 168–169, 171–172, 177, 179–187, 191–193, 195–196, 198–199, 202–204, 206–209, 211–212, 214–215, 217–222, 227–229, 231, 234, 237–238, 241–242, 248–250, 252–259, 261–264, 268–276, 289–290, 292, 309–310, 313–314, 316, 318, 320–322, 325–326, 328–330, 332–333, 335–336, 345–346, 350, 352–357, 361–362, 364–367, 369–375, 377–378, 380–384, 388–389, 391–397, 401–402, 405–406, 408–409, 411–413
terminal_session.py	178	135	24%	43, 92, 96–98, 107–108, 119–121, 123–126, 135–136, 140, 144, 147–148, 150, 155, 161, 163–164, 169, 177, 182–185, 198–200, 205, 208–209, 213, 219, 233–235, 240, 243–244, 248, 255, 264, 273–274, 276, 279–283, 285, 288, 290–292, 296–297, 300–302, 306, 311–312, 316–317, 323–325, 336–337, 340–342, 344–346, 349, 359, 363, 366, 369–370, 376–377, 383, 388–389, 392–396, 402, 404–406, 412–416, 419, 422, 425–426, 428–431, 438, 442, 447–448, 455–457, 461, 465, 470–471, 475–476, 479–482, 488–489, 492
tmux_terminal.py	80	29	63%	36, 45, 91, 95, 97, 107–108, 110, 118–119, 121, 128, 133, 145–152, 160–161, 163–164, 166, 168–170
openhands/tools/execute_bash/utils
command.py	81	72	11%	15–19, 27, 32, 34–35, 37–38, 41–46, 48, 51–53, 55, 58–67, 74–75, 77–79, 81, 83, 90–91, 93, 95–97, 99, 101–102, 105–106, 109, 117, 120–121, 123–124, 127–129, 132–138, 141–145, 150
openhands/tools/str_replace_editor
__init__.py	3	0	100%
definition.py	62	12	80%	84, 96, 116, 119, 122, 129, 131, 133, 135, 207, 210, 213
editor.py	228	170	25%	74, 76–77, 80, 98, 101, 111–115, 121–129, 131, 149–150, 170–171, 175, 179–180, 189, 193–196, 204–205, 209–211, 217, 220, 225, 228, 231–232, 235, 238–239, 243, 247, 262–264, 273, 276, 282, 287, 289–293, 295, 297, 301–302, 306–307, 316–317, 319–322, 324, 331–332, 338–340, 348–352, 356, 358–359, 366, 369, 374–375, 377, 401–402, 424–425, 427–428, 434, 437, 441–447, 450–451, 454–459, 462, 465–466, 470, 473–474, 477, 479–480, 486, 490, 509, 514–517, 519, 527, 534, 541, 552–555, 557, 559, 586–589, 598–599, 628–630, 632–641, 646–649, 662–663, 668, 673, 679, 685
exceptions.py	22	13	40%	5–6, 9, 16–18, 25–27, 38–41
impl.py	26	11	57%	31–32, 51–52, 54–56, 65–68
openhands/tools/str_replace_editor/utils
__init__.py	0	0	100%
config.py	2	0	100%
constants.py	5	0	100%
diff.py	64	15	76%	24, 40–42, 45–47, 50, 90–93, 107–108, 115
encoding.py	54	27	50%	42–43, 46–48, 51, 54, 60, 64–65, 67, 69, 78, 80–81, 84, 87–90, 93, 96–97, 114, 128, 130–131
file_cache.py	95	48	49%	43–46, 49–50, 54, 59, 61, 64, 69, 72–73, 87, 95–98, 108–112, 115–120, 126–130, 133–135, 138–140, 143–148, 151, 154
history.py	66	36	45%	58–60, 66–68, 70–71, 74–76, 78–79, 82, 85–86, 88, 92–94, 98–99, 102–104, 107, 111–113, 115–120, 122
shell.py	23	16	30%	30, 32–34, 38, 40, 51–55, 62–63, 70–72
openhands/tools/task_tracker
__init__.py	2	2	0%	1, 10
definition.py	132	132	0%	1–4, 6–7, 9–11, 20, 23–26, 33, 36, 40, 45–46, 48, 51–53, 55–56, 59–60, 62, 65, 68, 71–72, 76–78, 80–81, 83, 85, 87–88, 91, 94–96, 98–99, 102–108, 110–112, 115, 117–120, 122, 125, 128–129, 131–132, 134–135, 137, 140, 143, 150–151, 154–155, 157, 159, 161, 163–165, 171, 173–174, 179–180, 184, 191, 193–194, 196–198, 202–203, 205–208, 210, 212, 214–215, 217–219, 221–225, 229, 231, 233–234, 236–237, 239, 241–245, 249, 382, 396, 399–400, 407, 410
openhands/tools/utils
__init__.py	0	0	100%
TOTAL	6807	2673	60%

[xingyaoww]

Mostly LGTM - a few things we need to address

[xingyaoww]

You should add a new AgentStatus to Agent stuck in loop maybe?

[ryanhoangt]

Sounds good, I added a new status in the latest commit.

BTW, WDYT about the idea that, instead of raising an error and stop, we can (1) send a MessageAction to prompt the agent to do differently? Or even better but more complex, we can try to switch to a different model to handle this current turn (given that strong LLMs rarely get stuck I believe) -- If that makes sense I can create an issue for it.

[openhands-ai]

Looks like there are a few issues preventing this PR from being merged!

• GitHub Actions are failing:
  • Run tests

If you'd like me to help, just leave a comment, like

@OpenHands please fix the failing actions on PR #335 at branch `ht/stuck-detector`

Feel free to include any additional details that might help me get this PR into a better state.

_{^{You can manage your notification settings}}

[xingyaoww]

LGTM

[github-actions]

Agent Server image for this PR

Pull (multi-arch manifest):

docker pull ghcr.io/all-hands-ai/agent-server:83a04d5

Run:

docker run -it --rm \
  -p 8000:8000 \
  --name agent-server-83a04d5 \
  ghcr.io/all-hands-ai/agent-server:83a04d5

This tag is a multi-arch manifest (amd64/arm64). Your client pulls the right arch automatically.