Commit 9830b4e
committed
feat(hooks): widen contract for post_tool_use / before_llm_call + extract loop_detector & max_iterations builtins
Documents that decision="block" / continue=false / exit code 2 from a
post_tool_use or before_llm_call hook signals run termination — not just
the previously-undocumented behaviour of pre_tool_use deny. The executor's
aggregate logic already produced Allowed=false uniformly for these
verdicts; the runtime now actually acts on them.
Adds `hooks.DecisionBlockValue` so in-process builtins don't carry the
"block" string literal.
* `executePostToolHook` and `executeBeforeLLMCallHooks` now return
(stop, message). `processToolCalls` aggregates and propagates the
post-tool stop up to the run loop, synthesising error responses
for any unprocessed tool calls in the batch (same shape as user
cancellation) so the API doesn't reject orphan function calls.
* New `emitHookDrivenShutdown` factors the standard
Error / notification(level=error) / on_error fanout used by both
the post_tool_use and before_llm_call shutdown paths.
* `executeWithApproval` now returns a `toolApprovalOutcome` struct
instead of a single canceled bool to thread the new stop signal
through every approval path (yolo, allow, force-ask, default,
user prompt).
Replaces the inline `tool_loop_detector` previously in pkg/runtime.
Stateful per-session via `builtins.State`, registered through
`builtins.Register` and cleared on session_end. Auto-injected from
`agent.MaxConsecutiveToolCalls()` (defaulting to 5 to preserve the
historical always-on contract).
**Behaviour change:** signatures are now per-call, not per-batch. Single-
tool repetition (the dominant stuck-agent pattern) and parallel-identical
batches still trip; alternating multi-tool batches like `[A,B] [A,B] [A,B]`
no longer trip — that case should now be caught by max_iterations or
manual threshold tuning. Polling tools (view_background_agent,
view_background_job) remain invisible to the counter.
**Additive**, not a replacement: the existing `agent.MaxIterations` flow
(MaxIterationsReachedEvent + interactive resume dialog in TUI/CLI/ACP)
is unchanged because those UIs depend on the special event. The new
builtin gives users a hard-stop alternative they can configure purely
in YAML, with no resume protocol and no special event:
hooks:
before_llm_call:
- {type: builtin, command: max_iterations, args: ["50"]}
* `pkg/hooks/contract_widening_test.go` pins decision=block and
continue=false producing Allowed=false on post_tool_use and
before_llm_call.
* `pkg/hooks/builtins/loop_detector_test.go` covers threshold
tripping, JSON-key-order normalisation, exempt-tool invisibility,
per-session isolation, lenient arg parsing, and concurrent safety.
* `pkg/hooks/builtins/max_iterations_test.go` covers limit
tripping, per-session isolation, lenient arg parsing, and
concurrent safety.
* `hooks_wiring_test.go` updated to reflect that loop_detector is
auto-injected on every agent.
* `pkg/runtime/tool_loop_detector.go` and its test (logic moved to
`pkg/hooks/builtins/loop_detector.go` with per-call semantics).
Lint clean, all tests pass with -race.
Assisted-By: docker-agent1 parent 2e3457f commit 9830b4e
16 files changed
Lines changed: 1121 additions & 451 deletions
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
3 | 3 | | |
4 | 4 | | |
5 | 5 | | |
6 | | - | |
7 | | - | |
8 | | - | |
9 | | - | |
10 | | - | |
11 | | - | |
12 | | - | |
13 | | - | |
| 6 | + | |
| 7 | + | |
| 8 | + | |
| 9 | + | |
| 10 | + | |
| 11 | + | |
| 12 | + | |
| 13 | + | |
| 14 | + | |
| 15 | + | |
| 16 | + | |
14 | 17 | | |
15 | | - | |
16 | | - | |
17 | | - | |
18 | | - | |
| 18 | + | |
| 19 | + | |
| 20 | + | |
| 21 | + | |
| 22 | + | |
19 | 23 | | |
20 | | - | |
21 | | - | |
22 | | - | |
23 | | - | |
| 24 | + | |
| 25 | + | |
| 26 | + | |
| 27 | + | |
| 28 | + | |
| 29 | + | |
24 | 30 | | |
25 | 31 | | |
26 | 32 | | |
27 | 33 | | |
| 34 | + | |
28 | 35 | | |
29 | 36 | | |
30 | 37 | | |
31 | 38 | | |
32 | | - | |
33 | | - | |
34 | | - | |
| 39 | + | |
| 40 | + | |
| 41 | + | |
| 42 | + | |
| 43 | + | |
| 44 | + | |
| 45 | + | |
| 46 | + | |
| 47 | + | |
| 48 | + | |
| 49 | + | |
| 50 | + | |
| 51 | + | |
| 52 | + | |
| 53 | + | |
| 54 | + | |
| 55 | + | |
| 56 | + | |
| 57 | + | |
| 58 | + | |
| 59 | + | |
| 60 | + | |
| 61 | + | |
| 62 | + | |
| 63 | + | |
| 64 | + | |
35 | 65 | | |
36 | 66 | | |
37 | 67 | | |
| |||
40 | 70 | | |
41 | 71 | | |
42 | 72 | | |
43 | | - | |
| 73 | + | |
| 74 | + | |
| 75 | + | |
| 76 | + | |
| 77 | + | |
| 78 | + | |
44 | 79 | | |
45 | 80 | | |
46 | 81 | | |
47 | | - | |
| 82 | + | |
| 83 | + | |
| 84 | + | |
| 85 | + | |
| 86 | + | |
| 87 | + | |
| 88 | + | |
48 | 89 | | |
49 | | - | |
50 | | - | |
51 | | - | |
| 90 | + | |
| 91 | + | |
| 92 | + | |
| 93 | + | |
| 94 | + | |
52 | 95 | | |
53 | 96 | | |
54 | 97 | | |
55 | | - | |
56 | | - | |
57 | | - | |
58 | | - | |
| 98 | + | |
| 99 | + | |
59 | 100 | | |
60 | 101 | | |
61 | 102 | | |
| |||
69 | 110 | | |
70 | 111 | | |
71 | 112 | | |
| 113 | + | |
| 114 | + | |
| 115 | + | |
| 116 | + | |
| 117 | + | |
| 118 | + | |
| 119 | + | |
72 | 120 | | |
73 | 121 | | |
74 | 122 | | |
| |||
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
21 | 21 | | |
22 | 22 | | |
23 | 23 | | |
24 | | - | |
| 24 | + | |
| 25 | + | |
25 | 26 | | |
26 | 27 | | |
27 | 28 | | |
| |||
32 | 33 | | |
33 | 34 | | |
34 | 35 | | |
| 36 | + | |
| 37 | + | |
35 | 38 | | |
36 | 39 | | |
37 | 40 | | |
| |||
172 | 175 | | |
173 | 176 | | |
174 | 177 | | |
175 | | - | |
| 178 | + | |
| 179 | + | |
176 | 180 | | |
177 | 181 | | |
178 | 182 | | |
| |||
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
| 1 | + | |
| 2 | + | |
| 3 | + | |
| 4 | + | |
| 5 | + | |
| 6 | + | |
| 7 | + | |
| 8 | + | |
| 9 | + | |
| 10 | + | |
| 11 | + | |
| 12 | + | |
| 13 | + | |
| 14 | + | |
| 15 | + | |
| 16 | + | |
| 17 | + | |
| 18 | + | |
| 19 | + | |
| 20 | + | |
| 21 | + | |
| 22 | + | |
| 23 | + | |
| 24 | + | |
| 25 | + | |
| 26 | + | |
| 27 | + | |
| 28 | + | |
| 29 | + | |
| 30 | + | |
| 31 | + | |
| 32 | + | |
| 33 | + | |
| 34 | + | |
| 35 | + | |
| 36 | + | |
| 37 | + | |
| 38 | + | |
| 39 | + | |
| 40 | + | |
| 41 | + | |
| 42 | + | |
| 43 | + | |
| 44 | + | |
| 45 | + | |
| 46 | + | |
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
| 1 | + | |
| 2 | + | |
| 3 | + | |
| 4 | + | |
| 5 | + | |
| 6 | + | |
| 7 | + | |
| 8 | + | |
| 9 | + | |
| 10 | + | |
| 11 | + | |
| 12 | + | |
| 13 | + | |
| 14 | + | |
| 15 | + | |
| 16 | + | |
| 17 | + | |
| 18 | + | |
| 19 | + | |
| 20 | + | |
| 21 | + | |
| 22 | + | |
| 23 | + | |
| 24 | + | |
| 25 | + | |
| 26 | + | |
| 27 | + | |
| 28 | + | |
| 29 | + | |
| 30 | + | |
| 31 | + | |
| 32 | + | |
| 33 | + | |
| 34 | + | |
| 35 | + | |
| 36 | + | |
| 37 | + | |
| 38 | + | |
| 39 | + | |
| 40 | + | |
| 41 | + | |
| 42 | + | |
| 43 | + | |
| 44 | + | |
| 45 | + | |
| 46 | + | |
| 47 | + | |
| 48 | + | |
| 49 | + | |
| 50 | + | |
| 51 | + | |
| 52 | + | |
| 53 | + | |
| 54 | + | |
| 55 | + | |
| 56 | + | |
| 57 | + | |
| 58 | + | |
| 59 | + | |
| 60 | + | |
| 61 | + | |
| 62 | + | |
| 63 | + | |
| 64 | + | |
| 65 | + | |
| 66 | + | |
| 67 | + | |
| 68 | + | |
| 69 | + | |
| 70 | + | |
| 71 | + | |
| 72 | + | |
| 73 | + | |
| 74 | + | |
| 75 | + | |
| 76 | + | |
| 77 | + | |
| 78 | + | |
| 79 | + | |
| 80 | + | |
| 81 | + | |
| 82 | + | |
| 83 | + | |
| 84 | + | |
| 85 | + | |
| 86 | + | |
| 87 | + | |
| 88 | + | |
| 89 | + | |
| 90 | + | |
| 91 | + | |
| 92 | + | |
| 93 | + | |
| 94 | + | |
| 95 | + | |
| 96 | + | |
| 97 | + | |
| 98 | + | |
| 99 | + | |
| 100 | + | |
| 101 | + | |
| 102 | + | |
| 103 | + | |
| 104 | + | |
| 105 | + | |
| 106 | + | |
| 107 | + | |
| 108 | + | |
| 109 | + | |
| 110 | + | |
| 111 | + | |
| 112 | + | |
| 113 | + | |
| 114 | + | |
| 115 | + | |
| 116 | + | |
| 117 | + | |
| 118 | + | |
| 119 | + | |
| 120 | + | |
| 121 | + | |
| 122 | + | |
| 123 | + | |
| 124 | + | |
| 125 | + | |
| 126 | + | |
| 127 | + | |
| 128 | + | |
| 129 | + | |
| 130 | + | |
| 131 | + | |
| 132 | + | |
| 133 | + | |
| 134 | + | |
| 135 | + | |
| 136 | + | |
| 137 | + | |
| 138 | + | |
| 139 | + | |
| 140 | + | |
| 141 | + | |
| 142 | + | |
| 143 | + | |
| 144 | + | |
| 145 | + | |
| 146 | + | |
| 147 | + | |
| 148 | + | |
| 149 | + | |
| 150 | + | |
| 151 | + | |
| 152 | + | |
| 153 | + | |
0 commit comments