Spreadsheets plugin does not register required artifact-tool capabilities
Nobody has claimed this yet.
- Dominant language
- Rust
- Stars
- 125k
- Forks
- 19.4k
- PR merge metrics
- PR metrics pending
Description
What version of Codex CLI is running?
codex-cli 0.152.1
What subscription do you have?
Pro
Which model were you using?
gpt-5.6-sol
What platform is your computer?
Microsoft Windows NT 10.0.26200.0 x64
What terminal emulator and version are you using (if applicable)?
Windows Terminal 1.24.11911.0 with PowerShell; no multiplexer
Codex doctor report
- overallStatus: warning
- Codex CLI: 0.152.1
- Desktop component: 26.831.2377.0
- No installation, authentication, configuration, runtime, or network failures
- Unrelated warnings: non-Dev-Drive workspace and stale thread-history rows
To generate a path-redacted report directly into your clipboard, run this in PowerShell:
$raw = (codex doctor --json) -join "`n"
$profile = ($env:USERPROFILE | ConvertTo-Json -Compress).Trim('"')
$workspace = ('C:\dev\XXXX' | ConvertTo-Json -Compress).Trim('"')
$raw.Replace($profile, '<USERPROFILE>').Replace($workspace, '<WORKSPACE>').Replace('XXXX', '<PROJECT>') | Set-Clipboard
What issue are you seeing?
The spreadsheets@openai-primary-runtime plugin is installed and enabled, and
its Spreadsheets skill loads successfully. However,
load_workspace_dependencies is absent from the live callable-tool registry.
A direct registry check shows that the function is missing/undefined; this is
not an invocation error. Consequently, the skill cannot obtain the authoritative
Node executable and node_modules paths required to initialize
@oai/artifact-tool.
The failure reproduces in both:
- A brand-new ChatGPT Desktop Codex chat after updating and fully restarting
- A separate fresh standalone Codex CLI session
Both correctly stop with:
TOOL-BLOCKER
Environment:
- ChatGPT desktop app:
26.831.21537 - Codex CLI:
0.152.1 - Windows:
Microsoft Windows NT 10.0.26200.0 x64 - Model:
gpt-5.6-sol - Plugin:
spreadsheets@openai-primary-runtime - Plugin version:
26.826.12353 - Plugin state: installed and enabled
codex doctor: no installation, authentication, configuration, runtime, or
network failures
The installed plugin/runtime state is not being treated as proof that the
capability is available. The failure is based on inspection of the live
callable-tool registry in each fresh session.
Clarification from the original report:
I originally described mark_artifact_operation_started as a second required
live-registry capability. Public evidence in #38735 shows it documented as the
helper script container_tools/mark_artifact_operation_started.mjs, invoked
using the approved runtime, rather than necessarily being a host-injected
registry function.
I am therefore withdrawing the claim that
mark_artifact_operation_started must itself appear in the live tool registry.
The independently confirmed blocker is the missing
load_workspace_dependencies capability. Because that loader is unavailable,
the supported artifact runtime cannot be initialized and the marker step cannot
be reached.
No dependency paths were guessed, no local package installation or cache repair
was attempted, and no alternative spreadsheet library was used.
Related reports include #34270, #39021, and #42183. This report adds a current
Windows reproduction across both ChatGPT Desktop Codex and standalone Codex CLI
0.152.1
What steps can reproduce the bug?
Uploaded thread: 01a06105-2f7b-7451-b971-66bcf38119e4
What is the expected behavior?
When spreadsheets@openai-primary-runtime is installed and enabled and its skill
is exposed to a fresh eligible session, the host should register a callable
load_workspace_dependencies capability.
The capability should return the authoritative runtime executable and dependency
paths required to initialize @oai/artifact-tool. The skill should then be able
to execute its artifact-operation marker through the documented helper-script
mechanism.
If the standalone CLI or another surface intentionally cannot provide this
loader, the Spreadsheets skill should not be exposed there as usable, and the
unsupported limitation should be stated explicitly rather than failing only
after the skill is selected.
Additional information
I amended the title and report to distinguish the confirmed missing host
capability (load_workspace_dependencies) from the artifact-operation marker
helper script.
I am no longer asserting that mark_artifact_operation_started must appear as a
separate live-registry function. The confirmed loader-registration failure
remains reproducible in fresh Windows ChatGPT Desktop Codex and standalone Codex
CLI 0.152.1 sessions.
- ChatGPT desktop app version:
26.831.21537 - Codex CLI version:
0.152.1 - Plugin:
spreadsheets@openai-primary-runtime - Plugin version:
26.826.12353 - Plugin state: installed and enabled
- Reproduces in both ChatGPT desktop Codex and Codex CLI
codex doctorreports 0 failures- Authentication, configuration, installation, network, and runtime checks pass
- Date/time zone: 2026-09-02, Europe/Rome
/feedbacksession ID: [insert it here if available]
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by tracing plugin capability registration into the live callable-tool registry, focusing on the missing load_workspace_dependencies entry and the spreadsheets@openai-primary-runtime skill. Reproduce in a fresh ChatGPT Desktop Codex session and standalone Codex CLI session, then verify that the loader provides the authoritative Node executable and node_modules paths. Done means the capability is registered and usable, or the skill is not exposed where the loader is unavailable.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- node.js, rust
- Domain
- cli, devtools
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Active
- Clarity
- Mostly clear
- Newbie friendliness
- 45/100