UNPKG

openclaw

Version:

Multi-channel AI gateway with extensible messaging integrations

github.com/openclaw/openclaw

openclaw/openclaw

75 lines (62 loc) • 5.9 kB
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
--- summary: "Common Skill Workshop errors and the tool-policy doctor check" title: "Skill Workshop troubleshooting" read_when: - A Workshop proposal fails to create, apply, or appear in the list - The agent cannot call skill_workshop --- ## Troubleshooting | Problem | Resolution | | ---------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | `Skill proposal description is too large` | Shorten `description` to 160 bytes or less. | | `Skill proposal content is too large` | Shorten the proposal body or raise `skills.workshop.maxSkillBytes`. | | `Target skill changed after proposal creation` | Revise the proposal against the current target, or create a new proposal. | | `Proposal scan failed` | Inspect scanner findings, then revise or quarantine the proposal. | | `Support file paths must be under one of...` | Move support files under `assets/`, `examples/`, `references/`, `scripts/`, or `templates/`. | | Proposal does not show in list | Check the selected agent and `OPENCLAW_STATE_DIR`. | | Agent cannot call `skill_workshop` | Check the active tool policy and run mode. `coding` includes the tool; restrictive `tools.allow` policies must list it explicitly, and sandboxed runs must use a normal host-side agent session or the CLI. | ### Legacy ownership warnings during an update Doctor discovers historical workspace setup files from retained Workshop proposals and collection backups, even when the agent now uses a different workspace. It imports valid setup history before relocating Workshop records, preserving the original timestamps and an archived copy without changing workspace configuration. If an unconfigured historical workspace has missing or unreadable setup state, Doctor preserves its files and reports a warning. Obsolete proposals become stale; proposals with unfinished apply recovery and ambiguous collection backups remain available for manual review. Configured workspace failures and failures after an import has begun still block repair. Doctor leaves legacy proposal metadata and collection backups in place when their workspace has no configured owner or maps to more than one agent. The warning names the retained path and candidate agents. These ownership warnings do not stop the other migrations or later Doctor repairs. Review the retained proposal or backup manifest alongside the configured agent workspaces. Correct a workspace mapping only when it identifies the actual owner; do not assign an arbitrary agent or delete the artifacts to clear the warning. Rerun Doctor after resolving ownership. Outside the historical-workspace deferrals above, invalid metadata, failed writes, and unfinished recovery remain migration failures. A retarget count describes proposals changed during that pass. Any remaining external targets can belong to different proposals whose migration is blocked. Review their identifiers, paths, and migration warnings before retrying the same repair. Package rollback does not reverse the state-schema migration or relocated skill files. An older package can therefore be incompatible with the retained state. See [Schema bumps and older updaters](/reference/database-schemas#schema-bumps-and-older-updaters) and [Downgrade recovery](/reference/database-schemas#downgrade-recovery). ### Tool-policy diagnostic In `propose` and `auto` modes, `openclaw doctor` runs the `core/doctor/skill-workshop-tool-policy` check for each configured agent. Doctor checks the sandbox construction gate before tool policy: sandboxed runs without host-granted library-authoring authority do not construct `skill_workshop`, even when `alsoAllow` lists it. The warning names the effective sandbox setting and recommends a non-sandboxed session, a human turn with library-authoring authority, or the `openclaw skills workshop` CLI. For `non-main` mode, this restriction applies to non-main sessions. Library authority is granted by the host for an authenticated human turn; it is not an allowlist entry or a permission autonomous reviews can grant themselves. See [Personal library authoring](/tools/skill-workshop/personal-library). When the tool can be constructed but policy hides it, the warning names the first excluding config layer and the exact `allow` or `alsoAllow` change to make. A run whose explicit allowlist leaves no callable tools also names the sandbox gate when it excludes the requested Workshop tool. Older runbooks may still use `openclaw plugins inspect skill-workshop`; that command now explains that Skill Workshop is built in and prints the same policy hint when applicable.