codeer-cli 0.1.9__tar.gz → 0.1.10__tar.gz
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- {codeer_cli-0.1.9 → codeer_cli-0.1.10}/API_REFERENCE.md +18 -1
- {codeer_cli-0.1.9 → codeer_cli-0.1.10}/PKG-INFO +30 -1
- {codeer_cli-0.1.9 → codeer_cli-0.1.10}/README.md +29 -0
- {codeer_cli-0.1.9 → codeer_cli-0.1.10}/pyproject.toml +1 -1
- {codeer_cli-0.1.9 → codeer_cli-0.1.10}/src/codeer_cli/__init__.py +2 -1
- {codeer_cli-0.1.9 → codeer_cli-0.1.10}/src/codeer_cli/_validate.py +42 -1
- {codeer_cli-0.1.9 → codeer_cli-0.1.10}/src/codeer_cli/agents.py +9 -1
- {codeer_cli-0.1.9 → codeer_cli-0.1.10}/src/codeer_cli/cli.py +8 -0
- {codeer_cli-0.1.9 → codeer_cli-0.1.10}/src/codeer_cli/commands/agent.py +26 -3
- codeer_cli-0.1.10/src/codeer_cli/commands/model.py +43 -0
- codeer_cli-0.1.10/src/codeer_cli/models.py +16 -0
- codeer_cli-0.1.10/tests/test_agent_handoff.py +153 -0
- codeer_cli-0.1.10/tests/test_models.py +68 -0
- {codeer_cli-0.1.9 → codeer_cli-0.1.10}/uv.lock +1 -1
- {codeer_cli-0.1.9 → codeer_cli-0.1.10}/.gitignore +0 -0
- {codeer_cli-0.1.9 → codeer_cli-0.1.10}/src/codeer_cli/chats.py +0 -0
- {codeer_cli-0.1.9 → codeer_cli-0.1.10}/src/codeer_cli/client.py +0 -0
- {codeer_cli-0.1.9 → codeer_cli-0.1.10}/src/codeer_cli/commands/__init__.py +0 -0
- {codeer_cli-0.1.9 → codeer_cli-0.1.10}/src/codeer_cli/commands/_util.py +0 -0
- {codeer_cli-0.1.9 → codeer_cli-0.1.10}/src/codeer_cli/commands/check.py +0 -0
- {codeer_cli-0.1.9 → codeer_cli-0.1.10}/src/codeer_cli/commands/eval_cmd.py +0 -0
- {codeer_cli-0.1.9 → codeer_cli-0.1.10}/src/codeer_cli/commands/history.py +0 -0
- {codeer_cli-0.1.9 → codeer_cli-0.1.10}/src/codeer_cli/commands/kb.py +0 -0
- {codeer_cli-0.1.9 → codeer_cli-0.1.10}/src/codeer_cli/commands/profile.py +0 -0
- {codeer_cli-0.1.9 → codeer_cli-0.1.10}/src/codeer_cli/constants.py +0 -0
- {codeer_cli-0.1.9 → codeer_cli-0.1.10}/src/codeer_cli/eval_.py +0 -0
- {codeer_cli-0.1.9 → codeer_cli-0.1.10}/src/codeer_cli/histories.py +0 -0
- {codeer_cli-0.1.9 → codeer_cli-0.1.10}/src/codeer_cli/kb.py +0 -0
- {codeer_cli-0.1.9 → codeer_cli-0.1.10}/src/codeer_cli/parse.py +0 -0
- {codeer_cli-0.1.9 → codeer_cli-0.1.10}/tests/test_client_transport.py +0 -0
- {codeer_cli-0.1.9 → codeer_cli-0.1.10}/tests/test_eval_labels.py +0 -0
- {codeer_cli-0.1.9 → codeer_cli-0.1.10}/tests/test_eval_pairs.py +0 -0
- {codeer_cli-0.1.9 → codeer_cli-0.1.10}/tests/test_history_send.py +0 -0
- {codeer_cli-0.1.9 → codeer_cli-0.1.10}/tests/test_kb_nodes.py +0 -0
- {codeer_cli-0.1.9 → codeer_cli-0.1.10}/tests/test_kb_ranges.py +0 -0
- {codeer_cli-0.1.9 → codeer_cli-0.1.10}/tests/test_util.py +0 -0
|
@@ -49,6 +49,23 @@ like `knowledge_node_ids`, `domain`, `agent_id`, `http_request` config.
|
|
|
49
49
|
|
|
50
50
|
Limits: 10 tools per agent, ≤5 `call_agent`, ≤1 `memory`.
|
|
51
51
|
|
|
52
|
+
Agent create/update payloads may also include:
|
|
53
|
+
|
|
54
|
+
```json
|
|
55
|
+
{
|
|
56
|
+
"human_handoff": {
|
|
57
|
+
"enabled": true,
|
|
58
|
+
"idle_timeout_minutes": null,
|
|
59
|
+
"handoff_instructions": "Hand off when the user asks for a person."
|
|
60
|
+
}
|
|
61
|
+
}
|
|
62
|
+
```
|
|
63
|
+
|
|
64
|
+
`idle_timeout_minutes` must be greater than zero when provided. The runtime
|
|
65
|
+
only exposes the handoff tool to an external conversation with a non-empty
|
|
66
|
+
`external_user_id`, or to an evaluation run. Internal editor Live Test does not
|
|
67
|
+
activate human mode.
|
|
68
|
+
|
|
52
69
|
## Stage 2 — Knowledge bases
|
|
53
70
|
|
|
54
71
|
Base path: `/organizations/{org_id}/workspaces/{ws_id}/knowledge_bases`
|
|
@@ -280,7 +297,7 @@ Non-destructive: older versions stay in `GET /agents/{id}/histories`.
|
|
|
280
297
|
| --- | --- |
|
|
281
298
|
| `GET /accounts/me` | Sanity-check session, read workspace_organization_map |
|
|
282
299
|
| `GET /organizations` | List orgs visible to the user |
|
|
283
|
-
| `GET /llm/models` | List available LLM model IDs to use as `llm_model` |
|
|
300
|
+
| `GET /llm/models` | List available LLM model IDs to use as `llm_model` (`codeer model list`) |
|
|
284
301
|
| `GET /retrieval/...` | Shared retrieval helpers (file upload for attachments, markdown conversion) |
|
|
285
302
|
|
|
286
303
|
---
|
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
Metadata-Version: 2.4
|
|
2
2
|
Name: codeer-cli
|
|
3
|
-
Version: 0.1.
|
|
3
|
+
Version: 0.1.10
|
|
4
4
|
Summary: Command line tools for managing Codeer agents over the Codeer API.
|
|
5
5
|
Project-URL: Homepage, https://www.codeer.ai
|
|
6
6
|
Author: Codeer.AI
|
|
@@ -117,6 +117,35 @@ Validate setup before API work:
|
|
|
117
117
|
codeer check
|
|
118
118
|
```
|
|
119
119
|
|
|
120
|
+
List the active cloud models without opening the Codeer web app:
|
|
121
|
+
|
|
122
|
+
```bash
|
|
123
|
+
codeer model list --type text
|
|
124
|
+
```
|
|
125
|
+
|
|
126
|
+
## Agent human handoff
|
|
127
|
+
|
|
128
|
+
`codeer agent apply` accepts the same `human_handoff` object as the Agent API.
|
|
129
|
+
The dry-run validates it and shows whether handoff is enabled before any server
|
|
130
|
+
write:
|
|
131
|
+
|
|
132
|
+
```json
|
|
133
|
+
{
|
|
134
|
+
"name": "Support Agent",
|
|
135
|
+
"system_prompt": "Help the user safely.",
|
|
136
|
+
"human_handoff": {
|
|
137
|
+
"enabled": true,
|
|
138
|
+
"idle_timeout_minutes": null,
|
|
139
|
+
"handoff_instructions": "Hand off when the user asks for a person."
|
|
140
|
+
}
|
|
141
|
+
}
|
|
142
|
+
```
|
|
143
|
+
|
|
144
|
+
`idle_timeout_minutes` must be a positive integer or `null`. Human handoff only
|
|
145
|
+
becomes available in live published-agent conversations with a non-empty
|
|
146
|
+
`external_user_id`; editor Live Test conversations are internal and cannot
|
|
147
|
+
activate human mode.
|
|
148
|
+
|
|
120
149
|
## Upgrade and uninstall
|
|
121
150
|
|
|
122
151
|
Upgrade the CLI:
|
|
@@ -99,6 +99,35 @@ Validate setup before API work:
|
|
|
99
99
|
codeer check
|
|
100
100
|
```
|
|
101
101
|
|
|
102
|
+
List the active cloud models without opening the Codeer web app:
|
|
103
|
+
|
|
104
|
+
```bash
|
|
105
|
+
codeer model list --type text
|
|
106
|
+
```
|
|
107
|
+
|
|
108
|
+
## Agent human handoff
|
|
109
|
+
|
|
110
|
+
`codeer agent apply` accepts the same `human_handoff` object as the Agent API.
|
|
111
|
+
The dry-run validates it and shows whether handoff is enabled before any server
|
|
112
|
+
write:
|
|
113
|
+
|
|
114
|
+
```json
|
|
115
|
+
{
|
|
116
|
+
"name": "Support Agent",
|
|
117
|
+
"system_prompt": "Help the user safely.",
|
|
118
|
+
"human_handoff": {
|
|
119
|
+
"enabled": true,
|
|
120
|
+
"idle_timeout_minutes": null,
|
|
121
|
+
"handoff_instructions": "Hand off when the user asks for a person."
|
|
122
|
+
}
|
|
123
|
+
}
|
|
124
|
+
```
|
|
125
|
+
|
|
126
|
+
`idle_timeout_minutes` must be a positive integer or `null`. Human handoff only
|
|
127
|
+
becomes available in live published-agent conversations with a non-empty
|
|
128
|
+
`external_user_id`; editor Live Test conversations are internal and cannot
|
|
129
|
+
activate human mode.
|
|
130
|
+
|
|
102
131
|
## Upgrade and uninstall
|
|
103
132
|
|
|
104
133
|
Upgrade the CLI:
|
|
@@ -18,7 +18,7 @@ production and can be overridden from process env. The CLI does not read
|
|
|
18
18
|
workspace-local dotenv files or credential files.
|
|
19
19
|
"""
|
|
20
20
|
|
|
21
|
-
from ._validate import ToolValidationError
|
|
21
|
+
from ._validate import HumanHandoffValidationError, ToolValidationError
|
|
22
22
|
from .client import AuthError, CodeerClient, CodeerError, TransportError
|
|
23
23
|
from .parse import (
|
|
24
24
|
AgentSummary,
|
|
@@ -44,6 +44,7 @@ from .parse import (
|
|
|
44
44
|
|
|
45
45
|
__all__ = [
|
|
46
46
|
"CodeerClient", "CodeerError", "TransportError", "AuthError", "ToolValidationError",
|
|
47
|
+
"HumanHandoffValidationError",
|
|
47
48
|
# parsers
|
|
48
49
|
"AgentSummary", "ConversationTurn", "EvalResultSummary", "HistorySummary",
|
|
49
50
|
"KBNode", "ToolCall", "EvalToolCall",
|
|
@@ -1,4 +1,4 @@
|
|
|
1
|
-
"""Client-side validation for
|
|
1
|
+
"""Client-side validation for agent payloads.
|
|
2
2
|
|
|
3
3
|
These checks exist because the backend's form-schema validator is lenient
|
|
4
4
|
(``extra="allow"``) and silently accepts unknown ``type`` strings, which then
|
|
@@ -24,6 +24,47 @@ class ToolValidationError(ValueError):
|
|
|
24
24
|
"""Raised when a unified_tools payload is definitely wrong."""
|
|
25
25
|
|
|
26
26
|
|
|
27
|
+
class HumanHandoffValidationError(ValueError):
|
|
28
|
+
"""Raised when a human_handoff payload is definitely wrong."""
|
|
29
|
+
|
|
30
|
+
|
|
31
|
+
def validate_human_handoff(config: Any) -> dict[str, Any] | None:
|
|
32
|
+
"""Validate and normalize an optional human-handoff configuration."""
|
|
33
|
+
if config is None:
|
|
34
|
+
return None
|
|
35
|
+
if not isinstance(config, dict):
|
|
36
|
+
raise HumanHandoffValidationError("human_handoff must be an object.")
|
|
37
|
+
|
|
38
|
+
allowed_keys = {"enabled", "idle_timeout_minutes", "handoff_instructions"}
|
|
39
|
+
unknown_keys = sorted(set(config) - allowed_keys)
|
|
40
|
+
if unknown_keys:
|
|
41
|
+
raise HumanHandoffValidationError(
|
|
42
|
+
f"human_handoff contains unsupported field(s): {', '.join(unknown_keys)}."
|
|
43
|
+
)
|
|
44
|
+
|
|
45
|
+
enabled = config.get("enabled", False)
|
|
46
|
+
if not isinstance(enabled, bool):
|
|
47
|
+
raise HumanHandoffValidationError("human_handoff.enabled must be true or false.")
|
|
48
|
+
|
|
49
|
+
timeout = config.get("idle_timeout_minutes")
|
|
50
|
+
if timeout is not None and (type(timeout) is not int or timeout <= 0):
|
|
51
|
+
raise HumanHandoffValidationError(
|
|
52
|
+
"human_handoff.idle_timeout_minutes must be a positive integer or null."
|
|
53
|
+
)
|
|
54
|
+
|
|
55
|
+
instructions = config.get("handoff_instructions")
|
|
56
|
+
if instructions is not None and not isinstance(instructions, str):
|
|
57
|
+
raise HumanHandoffValidationError(
|
|
58
|
+
"human_handoff.handoff_instructions must be a string or null."
|
|
59
|
+
)
|
|
60
|
+
|
|
61
|
+
return {
|
|
62
|
+
"enabled": enabled,
|
|
63
|
+
"idle_timeout_minutes": timeout,
|
|
64
|
+
"handoff_instructions": instructions,
|
|
65
|
+
}
|
|
66
|
+
|
|
67
|
+
|
|
27
68
|
def validate_unified_tools(tools: Iterable[dict[str, Any]] | None) -> list[dict[str, Any]]:
|
|
28
69
|
"""Raise ToolValidationError on invalid payloads; otherwise return the list.
|
|
29
70
|
|
|
@@ -7,7 +7,7 @@ from __future__ import annotations
|
|
|
7
7
|
|
|
8
8
|
from typing import Any, List, Optional
|
|
9
9
|
|
|
10
|
-
from ._validate import validate_unified_tools
|
|
10
|
+
from ._validate import validate_human_handoff, validate_unified_tools
|
|
11
11
|
from .client import CodeerClient
|
|
12
12
|
|
|
13
13
|
|
|
@@ -24,8 +24,10 @@ def create(
|
|
|
24
24
|
suggested_questions: Optional[List[str]] = None,
|
|
25
25
|
primary_object_ids: Optional[List[int]] = None,
|
|
26
26
|
attachment_ids: Optional[List[str]] = None,
|
|
27
|
+
human_handoff: Optional[dict[str, Any]] = None,
|
|
27
28
|
) -> dict:
|
|
28
29
|
validated_tools = validate_unified_tools(unified_tools)
|
|
30
|
+
validated_handoff = validate_human_handoff(human_handoff)
|
|
29
31
|
body: dict[str, Any] = {
|
|
30
32
|
"name": name,
|
|
31
33
|
"system_prompt": system_prompt,
|
|
@@ -39,6 +41,8 @@ def create(
|
|
|
39
41
|
body["description"] = description
|
|
40
42
|
if llm_model is not None:
|
|
41
43
|
body["llm_model"] = llm_model
|
|
44
|
+
if validated_handoff is not None:
|
|
45
|
+
body["human_handoff"] = validated_handoff
|
|
42
46
|
return client.post("/external/agents", json=body)
|
|
43
47
|
|
|
44
48
|
|
|
@@ -56,9 +60,11 @@ def update(
|
|
|
56
60
|
suggested_questions: Optional[List[str]] = None,
|
|
57
61
|
primary_object_ids: Optional[List[int]] = None,
|
|
58
62
|
attachment_ids: Optional[List[str]] = None,
|
|
63
|
+
human_handoff: Optional[dict[str, Any]] = None,
|
|
59
64
|
) -> dict:
|
|
60
65
|
"""PUT creates a new AgentHistory snapshot (draft)."""
|
|
61
66
|
validated_tools = validate_unified_tools(unified_tools)
|
|
67
|
+
validated_handoff = validate_human_handoff(human_handoff)
|
|
62
68
|
body: dict[str, Any] = {
|
|
63
69
|
"name": name,
|
|
64
70
|
"system_prompt": system_prompt,
|
|
@@ -73,6 +79,8 @@ def update(
|
|
|
73
79
|
body["description"] = description
|
|
74
80
|
if llm_model is not None:
|
|
75
81
|
body["llm_model"] = llm_model
|
|
82
|
+
if validated_handoff is not None:
|
|
83
|
+
body["human_handoff"] = validated_handoff
|
|
76
84
|
return client.patch(f"/external/agents/{agent_id}", json=body)
|
|
77
85
|
|
|
78
86
|
|
|
@@ -2,6 +2,7 @@
|
|
|
2
2
|
|
|
3
3
|
codeer check
|
|
4
4
|
codeer agent list|get|apply|diff|versions
|
|
5
|
+
codeer model list
|
|
5
6
|
codeer kb list|files|upload|node-rename|node-delete|faq-list|faq-get|faq-create|faq-update|faq-delete
|
|
6
7
|
codeer eval list|label-list|label-create|label-update|label-delete|case-update|case-delete|evaluators|evaluator-create|evaluator-update|run|export|reconcile|cases-apply|rubrics|rubrics-apply
|
|
7
8
|
codeer history list|get|conversations|negative-feedback|create|send
|
|
@@ -25,6 +26,7 @@ def main(argv: list[str] | None = None) -> int:
|
|
|
25
26
|
epilog="""\
|
|
26
27
|
Safe workflow for coding agents:
|
|
27
28
|
codeer check --json
|
|
29
|
+
codeer model list --type text
|
|
28
30
|
codeer agent list
|
|
29
31
|
codeer agent get <agent-id> --full
|
|
30
32
|
codeer kb list
|
|
@@ -55,6 +57,12 @@ Use --out <path> for large raw artifacts; stdout defaults to compact summaries.
|
|
|
55
57
|
|
|
56
58
|
check.register(sub)
|
|
57
59
|
|
|
60
|
+
try:
|
|
61
|
+
from .commands import model as model_cmd
|
|
62
|
+
model_cmd.register(sub)
|
|
63
|
+
except ImportError:
|
|
64
|
+
pass
|
|
65
|
+
|
|
58
66
|
# Phase 2-4: agent, kb, eval commands will register here
|
|
59
67
|
try:
|
|
60
68
|
from .commands import agent as agent_cmd
|
|
@@ -6,6 +6,7 @@ from pathlib import Path
|
|
|
6
6
|
from typing import Optional
|
|
7
7
|
|
|
8
8
|
from .. import agents as agents_mod
|
|
9
|
+
from .._validate import validate_human_handoff, validate_unified_tools
|
|
9
10
|
from ..client import CodeerClient
|
|
10
11
|
from ._util import log, print_json, strip_noisy_fields, truncate, write_json
|
|
11
12
|
|
|
@@ -102,6 +103,12 @@ def _tool_summary(tools: list[dict] | None) -> list[dict]:
|
|
|
102
103
|
|
|
103
104
|
def _agent_summary(agent: dict, *, full: bool = False) -> dict:
|
|
104
105
|
tools = agent.get("unified_tools") or agent.get("tools") or []
|
|
106
|
+
human_handoff = agent.get("human_handoff")
|
|
107
|
+
if not isinstance(human_handoff, dict):
|
|
108
|
+
meta = agent.get("meta")
|
|
109
|
+
human_handoff = meta.get("human_handoff") if isinstance(meta, dict) else {}
|
|
110
|
+
if not isinstance(human_handoff, dict):
|
|
111
|
+
human_handoff = {}
|
|
105
112
|
row = {
|
|
106
113
|
"id": agent.get("id"),
|
|
107
114
|
"name": agent.get("name"),
|
|
@@ -118,6 +125,7 @@ def _agent_summary(agent: dict, *, full: bool = False) -> dict:
|
|
|
118
125
|
"agent_type": agent.get("agent_type"),
|
|
119
126
|
"updated_at": agent.get("updated_at"),
|
|
120
127
|
"tool_count": len(tools),
|
|
128
|
+
"human_handoff_enabled": bool(human_handoff.get("enabled")),
|
|
121
129
|
"system_prompt_chars": len(agent.get("system_prompt") or ""),
|
|
122
130
|
}
|
|
123
131
|
if full:
|
|
@@ -125,6 +133,7 @@ def _agent_summary(agent: dict, *, full: bool = False) -> dict:
|
|
|
125
133
|
row["use_search"] = agent.get("use_search")
|
|
126
134
|
row["suggested_questions"] = agent.get("suggested_questions") or []
|
|
127
135
|
row["tools"] = _tool_summary(tools)
|
|
136
|
+
row["human_handoff"] = human_handoff
|
|
128
137
|
row["system_prompt_preview"] = truncate(agent.get("system_prompt") or "", 1200)
|
|
129
138
|
return row
|
|
130
139
|
|
|
@@ -153,6 +162,13 @@ def run_apply(args, client) -> int:
|
|
|
153
162
|
log(f"error: payload missing required field(s): {', '.join(missing)}")
|
|
154
163
|
return 2
|
|
155
164
|
|
|
165
|
+
try:
|
|
166
|
+
validated_tools = validate_unified_tools(body.get("unified_tools") or [])
|
|
167
|
+
validated_handoff = validate_human_handoff(body.get("human_handoff"))
|
|
168
|
+
except ValueError as exc:
|
|
169
|
+
log(f"error: invalid agent payload: {exc}")
|
|
170
|
+
return 2
|
|
171
|
+
|
|
156
172
|
if args.dry_run:
|
|
157
173
|
operation = "update" if args.agent_id else "create"
|
|
158
174
|
result = {
|
|
@@ -162,9 +178,14 @@ def run_apply(args, client) -> int:
|
|
|
162
178
|
"payload": str(Path(args.payload)),
|
|
163
179
|
"name": body.get("name"),
|
|
164
180
|
"system_prompt_chars": len(body.get("system_prompt") or ""),
|
|
165
|
-
"tool_count": len(
|
|
181
|
+
"tool_count": len(validated_tools),
|
|
166
182
|
"use_search": body.get("use_search", False),
|
|
167
183
|
"llm_model": body.get("llm_model"),
|
|
184
|
+
"human_handoff": {
|
|
185
|
+
"enabled": bool((validated_handoff or {}).get("enabled")),
|
|
186
|
+
"idle_timeout_minutes": (validated_handoff or {}).get("idle_timeout_minutes"),
|
|
187
|
+
"instructions_chars": len((validated_handoff or {}).get("handoff_instructions") or ""),
|
|
188
|
+
},
|
|
168
189
|
"version_note": args.note if args.agent_id else None,
|
|
169
190
|
"would_write_server_state": True,
|
|
170
191
|
"next_step": "Review this summary, then rerun without --dry-run after approval.",
|
|
@@ -178,7 +199,7 @@ def run_apply(args, client) -> int:
|
|
|
178
199
|
client, args.agent_id,
|
|
179
200
|
name=body["name"],
|
|
180
201
|
system_prompt=body["system_prompt"],
|
|
181
|
-
unified_tools=
|
|
202
|
+
unified_tools=validated_tools,
|
|
182
203
|
use_search=body.get("use_search", False),
|
|
183
204
|
version_note=args.note,
|
|
184
205
|
description=body.get("description"),
|
|
@@ -186,6 +207,7 @@ def run_apply(args, client) -> int:
|
|
|
186
207
|
suggested_questions=body.get("suggested_questions") or [],
|
|
187
208
|
primary_object_ids=body.get("primary_object_ids") or [],
|
|
188
209
|
attachment_ids=body.get("attachment_ids") or [],
|
|
210
|
+
human_handoff=validated_handoff,
|
|
189
211
|
)
|
|
190
212
|
agent_id = args.agent_id
|
|
191
213
|
log(f"PUT /agents/{agent_id} ok")
|
|
@@ -197,13 +219,14 @@ def run_apply(args, client) -> int:
|
|
|
197
219
|
workspace_id=body["workspace_id"],
|
|
198
220
|
name=body["name"],
|
|
199
221
|
system_prompt=body["system_prompt"],
|
|
200
|
-
unified_tools=
|
|
222
|
+
unified_tools=validated_tools,
|
|
201
223
|
use_search=body.get("use_search", False),
|
|
202
224
|
description=body.get("description"),
|
|
203
225
|
llm_model=body.get("llm_model"),
|
|
204
226
|
suggested_questions=body.get("suggested_questions") or [],
|
|
205
227
|
primary_object_ids=body.get("primary_object_ids") or [],
|
|
206
228
|
attachment_ids=body.get("attachment_ids") or [],
|
|
229
|
+
human_handoff=validated_handoff,
|
|
207
230
|
)
|
|
208
231
|
agent_id = agent["id"]
|
|
209
232
|
log(f"POST /agents ok, id={agent_id}")
|
|
@@ -0,0 +1,43 @@
|
|
|
1
|
+
from __future__ import annotations
|
|
2
|
+
|
|
3
|
+
from .. import models as models_mod
|
|
4
|
+
from ._util import print_json, strip_noisy_fields, write_json
|
|
5
|
+
|
|
6
|
+
|
|
7
|
+
def register(subparsers):
|
|
8
|
+
model = subparsers.add_parser("model", help="Available LLM models")
|
|
9
|
+
sub = model.add_subparsers(dest="action", required=True)
|
|
10
|
+
|
|
11
|
+
p = sub.add_parser("list", help="List active cloud LLM models available to agents")
|
|
12
|
+
p.add_argument("--type", choices=("text", "image"), default=None,
|
|
13
|
+
help="Filter by model type")
|
|
14
|
+
p.add_argument("--full", action="store_true",
|
|
15
|
+
help="Include modalities, pricing, and creation metadata")
|
|
16
|
+
p.add_argument("--out", default=None,
|
|
17
|
+
help="Write the complete model response to this file")
|
|
18
|
+
p.set_defaults(func=run_list)
|
|
19
|
+
|
|
20
|
+
|
|
21
|
+
def _model_summary(model: dict, *, full: bool = False) -> dict:
|
|
22
|
+
row = {
|
|
23
|
+
"display_name": model.get("display_name"),
|
|
24
|
+
"model_id": model.get("model_id"),
|
|
25
|
+
"provider": model.get("provider"),
|
|
26
|
+
"model_type": model.get("model_type"),
|
|
27
|
+
}
|
|
28
|
+
if full:
|
|
29
|
+
row.update({
|
|
30
|
+
"input_modalities": model.get("input_modalities") or [],
|
|
31
|
+
"input_credits_per_million_tokens": model.get("input_credits_per_million_tokens"),
|
|
32
|
+
"output_credits_per_million_tokens": model.get("output_credits_per_million_tokens"),
|
|
33
|
+
"created_at": model.get("created_at"),
|
|
34
|
+
})
|
|
35
|
+
return row
|
|
36
|
+
|
|
37
|
+
|
|
38
|
+
def run_list(args, client) -> int:
|
|
39
|
+
result = models_mod.list_available(client, model_type=args.type)
|
|
40
|
+
full_result = strip_noisy_fields(result)
|
|
41
|
+
write_json(args.out, full_result)
|
|
42
|
+
print_json([_model_summary(model, full=args.full) for model in result])
|
|
43
|
+
return 0
|
|
@@ -0,0 +1,16 @@
|
|
|
1
|
+
"""Available LLM model discovery."""
|
|
2
|
+
|
|
3
|
+
from __future__ import annotations
|
|
4
|
+
|
|
5
|
+
from typing import Optional
|
|
6
|
+
|
|
7
|
+
from .client import CodeerClient
|
|
8
|
+
|
|
9
|
+
|
|
10
|
+
def list_available(
|
|
11
|
+
client: CodeerClient,
|
|
12
|
+
*,
|
|
13
|
+
model_type: Optional[str] = None,
|
|
14
|
+
) -> list[dict]:
|
|
15
|
+
params = {"model_type": model_type} if model_type else None
|
|
16
|
+
return client.get("/llm/models", params=params)
|
|
@@ -0,0 +1,153 @@
|
|
|
1
|
+
from __future__ import annotations
|
|
2
|
+
|
|
3
|
+
import contextlib
|
|
4
|
+
import io
|
|
5
|
+
import json
|
|
6
|
+
import tempfile
|
|
7
|
+
import unittest
|
|
8
|
+
from pathlib import Path
|
|
9
|
+
from types import SimpleNamespace
|
|
10
|
+
|
|
11
|
+
from codeer_cli import agents
|
|
12
|
+
from codeer_cli._validate import HumanHandoffValidationError, validate_human_handoff
|
|
13
|
+
from codeer_cli.commands import agent as agent_cmd
|
|
14
|
+
|
|
15
|
+
|
|
16
|
+
class FakeClient:
|
|
17
|
+
def __init__(self) -> None:
|
|
18
|
+
self.calls: list[tuple[str, str, dict]] = []
|
|
19
|
+
|
|
20
|
+
def post(self, path: str, **kwargs):
|
|
21
|
+
self.calls.append(("POST", path, kwargs))
|
|
22
|
+
return {"id": "agent-1"}
|
|
23
|
+
|
|
24
|
+
def patch(self, path: str, **kwargs):
|
|
25
|
+
self.calls.append(("PATCH", path, kwargs))
|
|
26
|
+
return {"id": "agent-1"}
|
|
27
|
+
|
|
28
|
+
|
|
29
|
+
class HumanHandoffValidationTests(unittest.TestCase):
|
|
30
|
+
def test_accepts_enabled_config_without_timeout(self) -> None:
|
|
31
|
+
result = validate_human_handoff({
|
|
32
|
+
"enabled": True,
|
|
33
|
+
"idle_timeout_minutes": None,
|
|
34
|
+
"handoff_instructions": "Hand off when safety approval is required.",
|
|
35
|
+
})
|
|
36
|
+
|
|
37
|
+
self.assertEqual(result, {
|
|
38
|
+
"enabled": True,
|
|
39
|
+
"idle_timeout_minutes": None,
|
|
40
|
+
"handoff_instructions": "Hand off when safety approval is required.",
|
|
41
|
+
})
|
|
42
|
+
|
|
43
|
+
def test_rejects_non_positive_timeout(self) -> None:
|
|
44
|
+
with self.assertRaises(HumanHandoffValidationError):
|
|
45
|
+
validate_human_handoff({"enabled": True, "idle_timeout_minutes": 0})
|
|
46
|
+
|
|
47
|
+
def test_rejects_unknown_fields(self) -> None:
|
|
48
|
+
with self.assertRaises(HumanHandoffValidationError):
|
|
49
|
+
validate_human_handoff({"enabled": True, "unexpected": "value"})
|
|
50
|
+
|
|
51
|
+
|
|
52
|
+
class AgentHumanHandoffPayloadTests(unittest.TestCase):
|
|
53
|
+
def test_create_forwards_human_handoff(self) -> None:
|
|
54
|
+
client = FakeClient()
|
|
55
|
+
handoff = {
|
|
56
|
+
"enabled": True,
|
|
57
|
+
"idle_timeout_minutes": 15,
|
|
58
|
+
"handoff_instructions": "Hand off on request.",
|
|
59
|
+
}
|
|
60
|
+
|
|
61
|
+
agents.create(
|
|
62
|
+
client, # type: ignore[arg-type]
|
|
63
|
+
workspace_id="workspace-1",
|
|
64
|
+
name="Agent",
|
|
65
|
+
system_prompt="Help safely.",
|
|
66
|
+
human_handoff=handoff,
|
|
67
|
+
)
|
|
68
|
+
|
|
69
|
+
self.assertEqual(client.calls[0][0:2], ("POST", "/external/agents"))
|
|
70
|
+
self.assertEqual(client.calls[0][2]["json"]["human_handoff"], handoff)
|
|
71
|
+
|
|
72
|
+
def test_update_forwards_human_handoff(self) -> None:
|
|
73
|
+
client = FakeClient()
|
|
74
|
+
handoff = {
|
|
75
|
+
"enabled": False,
|
|
76
|
+
"idle_timeout_minutes": None,
|
|
77
|
+
"handoff_instructions": None,
|
|
78
|
+
}
|
|
79
|
+
|
|
80
|
+
agents.update(
|
|
81
|
+
client, # type: ignore[arg-type]
|
|
82
|
+
"agent-1",
|
|
83
|
+
name="Agent",
|
|
84
|
+
system_prompt="Help safely.",
|
|
85
|
+
unified_tools=[],
|
|
86
|
+
use_search=False,
|
|
87
|
+
human_handoff=handoff,
|
|
88
|
+
)
|
|
89
|
+
|
|
90
|
+
self.assertEqual(client.calls[0][0:2], ("PATCH", "/external/agents/agent-1"))
|
|
91
|
+
self.assertEqual(client.calls[0][2]["json"]["human_handoff"], handoff)
|
|
92
|
+
|
|
93
|
+
|
|
94
|
+
class AgentHumanHandoffCommandTests(unittest.TestCase):
|
|
95
|
+
def _run_apply(self, payload: dict) -> tuple[int, dict]:
|
|
96
|
+
with tempfile.TemporaryDirectory() as tmpdir:
|
|
97
|
+
payload_path = Path(tmpdir) / "agent.json"
|
|
98
|
+
payload_path.write_text(json.dumps(payload))
|
|
99
|
+
args = SimpleNamespace(
|
|
100
|
+
payload=str(payload_path),
|
|
101
|
+
agent_id=None,
|
|
102
|
+
dry_run=True,
|
|
103
|
+
note=None,
|
|
104
|
+
)
|
|
105
|
+
stdout = io.StringIO()
|
|
106
|
+
with contextlib.redirect_stdout(stdout):
|
|
107
|
+
status = agent_cmd.run_apply(args, object())
|
|
108
|
+
return status, json.loads(stdout.getvalue()) if stdout.getvalue() else {}
|
|
109
|
+
|
|
110
|
+
def test_apply_dry_run_summarizes_human_handoff_without_api_call(self) -> None:
|
|
111
|
+
status, result = self._run_apply({
|
|
112
|
+
"name": "Agent",
|
|
113
|
+
"system_prompt": "Help safely.",
|
|
114
|
+
"human_handoff": {
|
|
115
|
+
"enabled": True,
|
|
116
|
+
"idle_timeout_minutes": 10,
|
|
117
|
+
"handoff_instructions": "Hand off on request.",
|
|
118
|
+
},
|
|
119
|
+
})
|
|
120
|
+
|
|
121
|
+
self.assertEqual(status, 0)
|
|
122
|
+
self.assertEqual(result["human_handoff"], {
|
|
123
|
+
"enabled": True,
|
|
124
|
+
"idle_timeout_minutes": 10,
|
|
125
|
+
"instructions_chars": 20,
|
|
126
|
+
})
|
|
127
|
+
|
|
128
|
+
def test_apply_dry_run_rejects_invalid_handoff(self) -> None:
|
|
129
|
+
stderr = io.StringIO()
|
|
130
|
+
with contextlib.redirect_stderr(stderr):
|
|
131
|
+
status, result = self._run_apply({
|
|
132
|
+
"name": "Agent",
|
|
133
|
+
"system_prompt": "Help safely.",
|
|
134
|
+
"human_handoff": {"enabled": "yes"},
|
|
135
|
+
})
|
|
136
|
+
|
|
137
|
+
self.assertEqual(status, 2)
|
|
138
|
+
self.assertEqual(result, {})
|
|
139
|
+
self.assertIn("human_handoff.enabled must be true or false", stderr.getvalue())
|
|
140
|
+
|
|
141
|
+
def test_agent_summary_reads_meta_fallback_safely(self) -> None:
|
|
142
|
+
fallback = agent_cmd._agent_summary({
|
|
143
|
+
"id": "agent-1",
|
|
144
|
+
"meta": {"human_handoff": {"enabled": True}},
|
|
145
|
+
})
|
|
146
|
+
malformed = agent_cmd._agent_summary({"id": "agent-2", "meta": "legacy"})
|
|
147
|
+
|
|
148
|
+
self.assertTrue(fallback["human_handoff_enabled"])
|
|
149
|
+
self.assertFalse(malformed["human_handoff_enabled"])
|
|
150
|
+
|
|
151
|
+
|
|
152
|
+
if __name__ == "__main__":
|
|
153
|
+
unittest.main()
|
|
@@ -0,0 +1,68 @@
|
|
|
1
|
+
from __future__ import annotations
|
|
2
|
+
|
|
3
|
+
import contextlib
|
|
4
|
+
import io
|
|
5
|
+
import json
|
|
6
|
+
import unittest
|
|
7
|
+
from types import SimpleNamespace
|
|
8
|
+
|
|
9
|
+
from codeer_cli import models
|
|
10
|
+
from codeer_cli.commands import model as model_cmd
|
|
11
|
+
|
|
12
|
+
|
|
13
|
+
class FakeClient:
|
|
14
|
+
def __init__(self) -> None:
|
|
15
|
+
self.calls: list[tuple[str, dict]] = []
|
|
16
|
+
|
|
17
|
+
def get(self, path: str, **kwargs):
|
|
18
|
+
self.calls.append((path, kwargs))
|
|
19
|
+
return [{"display_name": "Qwen", "model_id": "provider/qwen"}]
|
|
20
|
+
|
|
21
|
+
|
|
22
|
+
class ModelDiscoveryTests(unittest.TestCase):
|
|
23
|
+
def test_lists_text_models_with_api_filter(self) -> None:
|
|
24
|
+
client = FakeClient()
|
|
25
|
+
|
|
26
|
+
result = models.list_available(client, model_type="text") # type: ignore[arg-type]
|
|
27
|
+
|
|
28
|
+
self.assertEqual(result[0]["model_id"], "provider/qwen")
|
|
29
|
+
self.assertEqual(client.calls, [("/llm/models", {"params": {"model_type": "text"}})])
|
|
30
|
+
|
|
31
|
+
def test_lists_all_models_without_query_params(self) -> None:
|
|
32
|
+
client = FakeClient()
|
|
33
|
+
|
|
34
|
+
models.list_available(client) # type: ignore[arg-type]
|
|
35
|
+
|
|
36
|
+
self.assertEqual(client.calls, [("/llm/models", {"params": None})])
|
|
37
|
+
|
|
38
|
+
def test_command_full_output_includes_model_metadata(self) -> None:
|
|
39
|
+
class FullFakeClient(FakeClient):
|
|
40
|
+
def get(self, path: str, **kwargs):
|
|
41
|
+
self.calls.append((path, kwargs))
|
|
42
|
+
return [{
|
|
43
|
+
"display_name": "Qwen",
|
|
44
|
+
"model_id": "provider/qwen",
|
|
45
|
+
"provider": "Provider",
|
|
46
|
+
"model_type": "text",
|
|
47
|
+
"input_modalities": ["text"],
|
|
48
|
+
"input_credits_per_million_tokens": 1,
|
|
49
|
+
"output_credits_per_million_tokens": 2,
|
|
50
|
+
"created_at": "2026-07-22T00:00:00Z",
|
|
51
|
+
}]
|
|
52
|
+
|
|
53
|
+
client = FullFakeClient()
|
|
54
|
+
args = SimpleNamespace(type="text", full=True, out=None)
|
|
55
|
+
stdout = io.StringIO()
|
|
56
|
+
|
|
57
|
+
with contextlib.redirect_stdout(stdout):
|
|
58
|
+
status = model_cmd.run_list(args, client) # type: ignore[arg-type]
|
|
59
|
+
|
|
60
|
+
result = json.loads(stdout.getvalue())
|
|
61
|
+
self.assertEqual(status, 0)
|
|
62
|
+
self.assertEqual(result[0]["model_id"], "provider/qwen")
|
|
63
|
+
self.assertEqual(result[0]["input_modalities"], ["text"])
|
|
64
|
+
self.assertEqual(client.calls, [("/llm/models", {"params": {"model_type": "text"}})])
|
|
65
|
+
|
|
66
|
+
|
|
67
|
+
if __name__ == "__main__":
|
|
68
|
+
unittest.main()
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|