Commit 88c21b2
authored
ci: pin claude-code-action to v1.0.107 (#696)
## Summary
- `anthropics/claude-code-action@v1` (rolling tag) auto-pulled
**v1.0.108** at 2026-04-28 00:32 UTC and the new release regresses
OAuth-token auth — the agent SDK throws `Could not resolve
[authentication]` from `validateHeaders` before any work runs.
Refreshing `CLAUDE_CODE_OAUTH_TOKEN` does **not** help.
- Pin `.github/workflows/claude.yml` to the v1.0.107 SHA
(`7eab1296cc65117d50ac2a2fa5f00a30ec84d3d5`) until upstream ships a fix.
Inline comment notes when to lift the pin.
## Evidence
| Time (UTC) | Event |
| --- | --- |
| 2026-04-28 00:32:48 | `claude-code-action` v1.0.108 released; `v1` tag
updated |
| 2026-04-28 00:28 | last successful `@claude` run (still on v1.0.107) |
| 2026-04-28 00:41 → | every run fails with the same SDK auth error |
Failing run after rotating the OAuth token:
https://github.com/bootjp/elastickv/actions/runs/25039054927
Stack tail (same on all failing runs):
```
SDK execution error: 54 | new Anthropic({ apiKey, dangerouslyAllowBrowser: true });
... validateHeaders({values, nulls}){ ... throw Error('Could not resolve [authentication]')
##[error]Action failed with error: SDK execution error: Error: Claude Code process exited with code 1
```
The `git hash-object 'adapter/sqs_*.go' ...` warnings earlier in the
same logs are unrelated noise — the action runs `git hash-object` on
PR-only files before checking out the PR branch; they don't fail the
run.
## Test plan
- [ ] After merge, trigger `@claude` on any PR and confirm `Run Claude
Code` succeeds (currently 100% failure rate on `@v1`).
- [ ] Once upstream ships a fix (probably v1.0.109+), revert the pin
back to `@v1` in a follow-up.
## Self-review
Single-line `uses:` change in a CI workflow; the 5-lens review collapses
to: no data-loss / concurrency / consistency surface, no perf impact, no
test coverage applicable. The change is reversible by editing one line.1 file changed
Lines changed: 6 additions & 1 deletion
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
32 | 32 | | |
33 | 33 | | |
34 | 34 | | |
35 | | - | |
| 35 | + | |
| 36 | + | |
| 37 | + | |
| 38 | + | |
| 39 | + | |
| 40 | + | |
36 | 41 | | |
37 | 42 | | |
38 | 43 | | |
| |||
0 commit comments