starred/openclaude
1
0
Fork 0
mirror of https://github.com/Gitlawb/openclaude.git synced 2026-08-24 10:14:19 -05:00
Code Issues Packages Projects Releases Wiki Activity

diagnostics(query): trace interruption causality (#2111)

Browse Source 
* diagnostics(issue-1830): trace interruption causality

* test(issue-1830): lock interruption ownership matrix

* fix(codex): preserve stream deadline contract

* fix(diagnostics): harden interruption trace lifecycle

Refs #1830

* fix(diagnostics): harden interruption trace settlement

Refs #1830

* fix(diagnostics): preserve interruption causality

* fix(diagnostics): address interruption trace review

* fix(diagnostics): preserve tracing observer contracts

* fix(diagnostics): preserve interruption trace contracts

* test(permissions): cover interactive hook interrupts
This commit is contained in:
Bogdan
2026-08-16 15:54:29 +08:00
committed by GitHub
parent ea655163d3
commit c30578819e
91 changed files with 7545 additions and 389 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/advanced-setup.md
+2
View File
@@ -481,6 +481,8 @@ host. Without this variable the behavior is unchanged.
| `OPENCLAUDE_MAX_TURNS` | No | Per-prompt **local** interactive REPL turn cap for the in-process query loop. Defaults to `50`. Set a larger positive integer for long autonomous local interactive sessions (for example models that take many small tool steps). CLI `--max-turns 0` explicitly disables this cap and prints a cautionary warning. Precedence for a valid override: CLI `--max-turns` → this env var → legacy `CLAUDE_CODE_MAX_TURNS` (only when this var is unset/empty) → `/config` → Max turns (interactive) → `50`. If this env var is set but invalid (zero, negative, non-integer), the default `50` is used and lower layers are not consulted — same pattern as `OPENCLAUDE_MAX_RETRIES`. Does not apply to remote-backed interactive sessions (`connect` / `ssh` / `--remote`). |
| `OPENCLAUDE_RETRY_DELAY_MS` | No | Base retry delay in milliseconds for APIs that do not send `Retry-After`; exponential backoff starts from this value, capped at 60000 (default: 500) |
| `OPENCLAUDE_QUERY_HARD_MAX_MS` | No | Foreground query hard maximum in milliseconds. Defaults to 1800000 (30 minutes). Use a larger positive integer for long autonomous sessions; invalid, zero, negative, fractional, or timer-overflow values are ignored with a warning. |
| `OPENCLAUDE_INTERRUPT_TRACE` | No | Set to `1` or `true` to retain a bounded, privacy-safe interruption lifecycle trace in memory. Disabled by default. The trace contains only allowlisted lifecycle metadata—never prompts, responses, tool arguments, credentials, or raw error messages. |
| `OPENCLAUDE_INTERRUPT_TRACE_FILE` | No | Optional absolute JSONL output path used only when `OPENCLAUDE_INTERRUPT_TRACE` is enabled. On Linux, missing parent directories are created privately and every parent is opened through `/proc/self/fd` without following symbolic links before the final regular file is appended. If the file already exists, its mode is reset to `0600` on every append, so do not configure a shared file. Other platforms retain the bounded trace in memory but do not write this file because Node does not expose an equivalent safe descriptor-relative traversal API there. Writes are best-effort and never change request behavior. Use a separate path per OpenClaude process and keep the resulting diagnostic file private. |
| `OPENCLAUDE_DISABLE_CO_AUTHORED_BY` | No | Suppress the default `Co-Authored-By` trailer in generated git commits |
| `OPENCLAUDE_LOG_TOKEN_USAGE` | No | When truthy (e.g. `verbose`), emits one JSON line on stderr per API request with input/output/cache tokens and the resolved provider. **User-facing debug output** — complements the REPL display controlled by `/config showCacheStats`. Distinct from `CLAUDE_CODE_ENABLE_TOKEN_USAGE_ATTACHMENT`, which is **model-facing** (injects context usage info into the prompt itself). Both can run together. |