int-upload #1

Merged

Jriyuu merged 26 commits from int-upload into main 2026-11-02 13:08:37 +11:00

Conversation 1 Commits 26 Files changed 19 +1859 -2

Jriyuu commented 2026-11-02 09:16:12 +11:00

Owner

Copy link

No description provided.

Jriyuu added 7 commits 2026-11-02 09:16:12 +11:00

Upload files to "tests" 765d7cb17a

Upload files to "src/bin" 4157287646

Upload files to "src/workbench" db06947844

Upload files to "src" a710ae27ae

Upload files to "docker" 9be8f8c0b1

Upload files to "/" 7815ec47a1

Upload files to ".forgejo/workflows" 1add2d259f

forgejo-actions commented 2026-11-02 09:30:52 +11:00

Copy link

🤖 AI review for 62bcfc2a

Advisory only - this never blocks the merge. You decide.

Review output

[ocr] 19 file(s) changed, reviewing 15 in /workspace/jriyuu-labs/craft-workbench
[ocr] Skipping CHANGELOG.md — filtered by path/extension rules
[ocr] Skipping README.md — filtered by path/extension rules
[ocr] Skipping src/job_instructions.md — filtered by path/extension rules
[ocr] Skipping src/requirements.txt — filtered by path/extension rules
[ocr] Filtered 4 file(s) by include/exclude rules
[ocr] Plan completed for group "tests/mock_litellm.py"
[ocr]   ▶ code_search file_patterns=[['*']] search_text=mock_litellm
[ocr]   ✔ code_search (12ms)
[ocr]   ▶ code_search file_patterns=[['src/']] search_text=litellm|base_url
[ocr]   ✔ code_search (9ms)
[ocr]   ▶ code_search search_text=mock_litellm
[ocr]   ✔ code_search (11ms)
[ocr]   ▶ code_search search_text=litellm
[ocr]   ✔ code_search (11ms)
[ocr]   ▶ file_read end_line=230 file_path=README.md start_line=185
[ocr]   ✔ file_read (3ms)
[ocr]   ▶ file_find query_name=tests/
[ocr]   ✔ file_find (4ms)
[ocr]   ▶ file_read end_line=60 file_path=src/entrypoint.sh start_line=1
[ocr]   ✔ file_read (5ms)
[ocr]   ▶ file_read end_line=60 file_path=src/labs.example.yaml start_line=1
[ocr]   ✔ file_read (5ms)
[ocr]   ▶ file_read file_path=src/workbench/server.py start_line=170 end_line=215
[ocr]   ✔ file_read (5ms)
[ocr]   ▶ code_comment
[ocr]   ✔ code_comment (0s)
[ocr]   ▶ code_search search_text=mock_litellm
[ocr]   ✔ code_search (56ms)
[ocr]   ▶ code_search search_text=litellm
[ocr]   ✔ code_search (9ms)
[ocr]   ▶ code_search search_text=base_url
[ocr]   ✔ code_search (9ms)
[ocr]   ▶ file_read end_line=230 file_path=README.md start_line=190
[ocr]   ✔ file_read (5ms)
[ocr]   ▶ file_read end_line=105 file_path=src/workbench/guard.py start_line=1
[ocr]   ✔ file_read (5ms)
[ocr]   ▶ code_search search_text=UNSAFE_ALLOW_LOCAL
[ocr]   ✔ code_search (10ms)
[ocr]   ▶ code_search file_patterns=[src/workbench/server.py] search_text=Guard
[ocr]   ✔ code_search (10ms)
[ocr]   ▶ code_search search_text=api_token file_patterns=[src/workbench/server.py]
[ocr]   ✔ code_search (11ms)
[ocr]   ▶ file_read file_path=src/workbench/server.py start_line=540 end_line=600
[ocr]   ✔ file_read (5ms)
[ocr]   ▶ file_read start_line=1 end_line=43 file_path=src/labs.example.yaml
[ocr]   ✔ file_read (5ms)
[ocr] Plan completed for group ".forgejo/workflows/build.yml,docker/Dockerfile,docker/compose.yaml,docker/craft-workbench.xml,src/bin/ca,src/bin/chromium,src/entrypoint.sh"
[ocr]   ▶ file_find query_name=requirements
[ocr]   ✔ file_find (5ms)
[ocr]   ▶ code_search file_patterns=[['src/workbench/setup.py']] search_text=print\( use_perl_regexp=true
[ocr]   ✔ code_search (9ms)
[ocr]   ▶ file_read_diff path_array=[src/workbench/__main__.py]
[ocr]   ✔ file_read_diff (0s)
[ocr]   ▶ file_read file_path=src/requirements.txt
[ocr]   ✔ file_read (5ms)
[ocr]   ▶ file_read file_path=src/workbench/setup.py
[ocr]   ✔ file_read (5ms)
[ocr]   ▶ code_search use_perl_regexp=true
[ocr]   ✔ code_search (11ms)
[ocr]   ▶ code_comment
[ocr]   ✔ code_comment (0s)
[ocr]   ▶ file_read file_path=src/workbench/config.py
[ocr]   ✔ file_read (5ms)
[ocr]   ▶ file_read file_path=src/workbench/server.py start_line=150 end_line=280
[ocr]   ✔ file_read (5ms)
[ocr]   ▶ file_find query_name=test
[ocr]   ✔ file_find (9ms)
[ocr]   ▶ file_read end_line=80 file_path=src/workbench/server.py start_line=1
[ocr]   ✔ file_read (4ms)
[ocr]   ▶ file_find query_name=job_instructions
[ocr]   ✔ file_find (4ms)
[ocr]   ▶ code_comment
[ocr]   ✔ code_comment (0s)
[ocr] No tool calls parsed for .forgejo/workflows/build.yml,docker/Dockerfile,docker/compose.yaml,docker/craft-workbench.xml,src/bin/ca,src/bin/chromium,src/entrypoint.sh, retrying...
[ocr] Round 2 failed for group ".forgejo/workflows/build.yml,docker/Dockerfile,docker/compose.yaml,docker/craft-workbench.xml,src/bin/ca,src/bin/chromium,src/entrypoint.sh": LLM completion error: POST "https://ai.jriyuu.io/v1/chat/completions": 400 Bad Request {"message":"litellm.BadRequestError: OpenAIException - Invalid request parameters. Please check your input and try again.\n\nLiteLLM: model group 'ci-reviewer' failed with the error above and no fallback model group was found for it, so the request was not retried on another model. Fallbacks are configured for: qwen3.8-27b, qwen3.8-27b-fast, reverse-engineer-core. Add a fallbacks entry for that model group (Router fallbacks or proxy router_settings.fallbacks) to retry on another model.","type":"invalid_request_error","param":"messages[1].content","code":"400"} (keeping earlier findings)
[ocr] Plan phase failed for group "src/labs.example.yaml,src/workbench/__init__.py,src/workbench/__main__.py,src/workbench/config.py,src/workbench/guard.py,src/workbench/server.py,src/workbench/setup.py": plan request: POST "https://ai.jriyuu.io/v1/chat/completions": 504 Gateway Timeout : <html>
<head><title>504 Gateway Time-out</title></head>
<body>
<center><h1>504 Gateway Time-out</h1></center>
<hr><center>nginx</center>
</body>
</html> (continuing without plan)
[ocr] No tool calls parsed for src/labs.example.yaml,src/workbench/__init__.py,src/workbench/__main__.py,src/workbench/config.py,src/workbench/guard.py,src/workbench/server.py,src/workbench/setup.py, retrying...
[ocr] Subtask error for group "workbench Python application & config": LLM completion error: caller deadline exceeded: context deadline exceeded
[ocr] Summary: 15 file(s) reviewed, 4 comment(s), ~486877 token(s) used (input: ~356269, output: ~130608), cache(read: ~220608, write: ~0), 30m13s elapsed
[ocr] Session: e7057185-0384-48d2-8df5-305d23179907
Review partially complete: 4 finding(s); 7 of 15 selected item(s) failed.

─── tests/mock_litellm.py:138-139 ───
[maintainability · low] The CLI args are consumed positionally with no validation: running `python3
tests/mock_litellm.py` with no args crashes with a raw `IndexError` traceback from
`int(sys.argv[1])` instead of the usage message the docstring already provides, and `LOG_PATH` is
bound from `sys.argv` at module import time, coupling the module to the process argv even when it's
only imported. Cheap fix: check argc once in the `__main__` block and print the usage.

  if __name__ == "__main__":
+     if not 2 <= len(sys.argv) <= 3:
+         sys.exit("usage: python3 tests/mock_litellm.py <port> <request-log-file>")
      ThreadingHTTPServer(("127.0.0.1", int(sys.argv[1])), Handler).serve_forever()


─── tests/mock_litellm.py:44-45 ───
[bug · low] A request body that parses as valid JSON but isn't an object (e.g. `null`, `[]`, or a
bare string) slips past this `except`: `body.get('model')` then raises `AttributeError`, which isn't
in the caught tuple, so the handler thread dies in socketserver's handle_error and the client gets a
dropped connection instead of the 400 this block is meant to produce for malformed input.
Normalizing the shape inside the try (e.g. `if not isinstance(body, dict): raise ValueError("request
body must be a JSON object")`) routes it through the existing 400 path.

              body = json.loads(self.rfile.read(int(self.headers.get("content-length", 0))))
+             if not isinstance(body, dict):
+                 raise ValueError("request body must be a JSON object")
              auth = self.headers.get("authorization") or self.headers.get("x-api-key")


─── src/entrypoint.sh:32-35 ───
[bug · low] The condition doesn't match the comment: docker stop is handled by the TERM/INT trap
above (`wait -n` is interruptible, so the trap runs and exits 0 before this line is reached). Signal
numbers 129-255 observed at `wait -n` actually mean a lab server itself was killed by a signal —
e.g. OOM-kill (137, plausible given one 4G Ghidra heap per running job under a 20g limit) or
segfault (139) — and that case is silently swallowed here: no "a lab server exited" line, no
diagnostic before the automatic restart. Consider logging it (and exiting 1 so restart policies like
on-failure still restart):

  if [ "$rc" -gt 128 ]; then
-   # A stop signal interrupted the wait (docker stop): the trap already shut down cleanly.
-   exit 0
+   # A lab server died from a signal (e.g. OOM-killed 137, segfault 139); the TERM/INT
+   # trap already handles docker stop, so anything seen here is a crash worth a log line.
+   echo "craft-workbench: a lab server died from a signal (rc=$rc); stopping"
+   kill "${pids[@]}" 2>/dev/null || true
+   wait
+   exit 1
  fi


─── src/entrypoint.sh:21-25 ───
[test · low] The bootstrap enforces an unmachined contract between three sides: setup must print
exactly "<lab> <port>" to stdout (stderr must carry the logs), serve must run as `python -m
workbench serve <lab>` with the lab name as the sole positional arg, and the image HEALTHCHECK must
match the child cmdline via pgrep -f "[w]orkbench serve". It all holds today, but a format or argv
drift fails silently — zero job APIs, container stays up, only symptom is pgrep going unhealthy.
Nothing in the repo (only tests/mock_litellm.py was added) exercises this layer; a smoke test
running setup/serve with WORKBENCH_* overrides pointed at temp dirs would lock the stdout format,
the module invocation form and the healthcheck pattern together.

- while read -r lab _port; do
-   [ -n "$lab" ] || continue
-   "$PY" -m workbench serve "$lab" &
-   pids+=("$!")
- done <<< "$labs"
+ # e.g. a smoke test that asserts, for a temp labs.yaml reached via WORKBENCH_* env:
+ #   setup output lines match '^[a-z][a-z0-9-]* [0-9]+$'   (this loop's parse assumptions)
+ #   a spawned `python -m workbench serve <name>` cmdline matches pgrep -f "[w]orkbench serve"
+ #   `python -m workbench serve` with no arg / unknown arg exits non-zero

[ocr] WARNING [review_round_failed] .forgejo/workflows/build.yml,docker/Dockerfile,docker/compose.yaml,docker/craft-workbench.xml,src/bin/ca,src/bin/chromium,src/entrypoint.sh: round 2: LLM completion error: POST "https://ai.jriyuu.io/v1/chat/completions": 400 Bad Request {"message":"litellm.BadRequestError: OpenAIException - Invalid request parameters. Please check your input and try again.\n\nLiteLLM: model group 'ci-reviewer' failed with the error above and no fallback model group was found for it, so the request was not retried on another model. Fallbacks are configured for: qwen3.8-27b, qwen3.8-27b-fast, reverse-engineer-core. Add a fallbacks entry for that model group (Router fallbacks or proxy router_settings.fallbacks) to retry on another model.","type":"invalid_request_error","param":"messages[1].content","code":"400"}

LLM retry report summary: 5 of 30 requests affected -- 3 requests failed, 2 requests recovered after retry

Review planning (2 requests):
- src/labs.example.yaml,src/workbench/__init__.py,src/workbench/__main__.py,src/workbench/config.py,src/workbench/guard.py,src/workbench/server.py,src/workbench/setup.py: timed out (HTTP 504) -> timed out (HTTP 504) -> timed out (HTTP 504) -> timed out (HTTP 504) -> timed out (HTTP 504) -> timed out (HTTP 504) -> failed
- .forgejo/workflows/build.yml,docker/Dockerfile,docker/compose.yaml,docker/craft-workbench.xml,src/bin/ca,src/bin/chromium,src/entrypoint.sh: timed out (HTTP 504) -> succeeded

Core review (3 requests):
- .forgejo/workflows/build.yml,docker/Dockerfile,docker/compose.yaml,docker/craft-workbench.xml,src/bin/ca,src/bin/chromium,src/entrypoint.sh: rejected by provider (HTTP 400) -> failed
- src/labs.example.yaml,src/workbench/__init__.py,src/workbench/__main__.py,src/workbench/config.py,src/workbench/guard.py,src/workbench/server.py,src/workbench/setup.py: timed out -> failed
- .forgejo/workflows/build.yml,docker/Dockerfile,docker/compose.yaml,docker/craft-workbench.xml,src/bin/ca,src/bin/chromium,src/entrypoint.sh: timed out (HTTP 504) -> succeeded

Per-attempt detail: --format json (retry_report).

_{Open Code Review via ci-actions · model ci-reviewer · job log}

<!-- ci-actions:open-code-review --> ### 🤖 AI review for `62bcfc2a` Advisory only - this never blocks the merge. You decide. <details open><summary>Review output</summary> ```text [ocr] 19 file(s) changed, reviewing 15 in /workspace/jriyuu-labs/craft-workbench [ocr] Skipping CHANGELOG.md — filtered by path/extension rules [ocr] Skipping README.md — filtered by path/extension rules [ocr] Skipping src/job_instructions.md — filtered by path/extension rules [ocr] Skipping src/requirements.txt — filtered by path/extension rules [ocr] Filtered 4 file(s) by include/exclude rules [ocr] Plan completed for group "tests/mock_litellm.py" [ocr] ▶ code_search file_patterns=[['*']] search_text=mock_litellm [ocr] ✔ code_search (12ms) [ocr] ▶ code_search file_patterns=[['src/']] search_text=litellm|base_url [ocr] ✔ code_search (9ms) [ocr] ▶ code_search search_text=mock_litellm [ocr] ✔ code_search (11ms) [ocr] ▶ code_search search_text=litellm [ocr] ✔ code_search (11ms) [ocr] ▶ file_read end_line=230 file_path=README.md start_line=185 [ocr] ✔ file_read (3ms) [ocr] ▶ file_find query_name=tests/ [ocr] ✔ file_find (4ms) [ocr] ▶ file_read end_line=60 file_path=src/entrypoint.sh start_line=1 [ocr] ✔ file_read (5ms) [ocr] ▶ file_read end_line=60 file_path=src/labs.example.yaml start_line=1 [ocr] ✔ file_read (5ms) [ocr] ▶ file_read file_path=src/workbench/server.py start_line=170 end_line=215 [ocr] ✔ file_read (5ms) [ocr] ▶ code_comment [ocr] ✔ code_comment (0s) [ocr] ▶ code_search search_text=mock_litellm [ocr] ✔ code_search (56ms) [ocr] ▶ code_search search_text=litellm [ocr] ✔ code_search (9ms) [ocr] ▶ code_search search_text=base_url [ocr] ✔ code_search (9ms) [ocr] ▶ file_read end_line=230 file_path=README.md start_line=190 [ocr] ✔ file_read (5ms) [ocr] ▶ file_read end_line=105 file_path=src/workbench/guard.py start_line=1 [ocr] ✔ file_read (5ms) [ocr] ▶ code_search search_text=UNSAFE_ALLOW_LOCAL [ocr] ✔ code_search (10ms) [ocr] ▶ code_search file_patterns=[src/workbench/server.py] search_text=Guard [ocr] ✔ code_search (10ms) [ocr] ▶ code_search search_text=api_token file_patterns=[src/workbench/server.py] [ocr] ✔ code_search (11ms) [ocr] ▶ file_read file_path=src/workbench/server.py start_line=540 end_line=600 [ocr] ✔ file_read (5ms) [ocr] ▶ file_read start_line=1 end_line=43 file_path=src/labs.example.yaml [ocr] ✔ file_read (5ms) [ocr] Plan completed for group ".forgejo/workflows/build.yml,docker/Dockerfile,docker/compose.yaml,docker/craft-workbench.xml,src/bin/ca,src/bin/chromium,src/entrypoint.sh" [ocr] ▶ file_find query_name=requirements [ocr] ✔ file_find (5ms) [ocr] ▶ code_search file_patterns=[['src/workbench/setup.py']] search_text=print\( use_perl_regexp=true [ocr] ✔ code_search (9ms) [ocr] ▶ file_read_diff path_array=[src/workbench/__main__.py] [ocr] ✔ file_read_diff (0s) [ocr] ▶ file_read file_path=src/requirements.txt [ocr] ✔ file_read (5ms) [ocr] ▶ file_read file_path=src/workbench/setup.py [ocr] ✔ file_read (5ms) [ocr] ▶ code_search use_perl_regexp=true [ocr] ✔ code_search (11ms) [ocr] ▶ code_comment [ocr] ✔ code_comment (0s) [ocr] ▶ file_read file_path=src/workbench/config.py [ocr] ✔ file_read (5ms) [ocr] ▶ file_read file_path=src/workbench/server.py start_line=150 end_line=280 [ocr] ✔ file_read (5ms) [ocr] ▶ file_find query_name=test [ocr] ✔ file_find (9ms) [ocr] ▶ file_read end_line=80 file_path=src/workbench/server.py start_line=1 [ocr] ✔ file_read (4ms) [ocr] ▶ file_find query_name=job_instructions [ocr] ✔ file_find (4ms) [ocr] ▶ code_comment [ocr] ✔ code_comment (0s) [ocr] No tool calls parsed for .forgejo/workflows/build.yml,docker/Dockerfile,docker/compose.yaml,docker/craft-workbench.xml,src/bin/ca,src/bin/chromium,src/entrypoint.sh, retrying... [ocr] Round 2 failed for group ".forgejo/workflows/build.yml,docker/Dockerfile,docker/compose.yaml,docker/craft-workbench.xml,src/bin/ca,src/bin/chromium,src/entrypoint.sh": LLM completion error: POST "https://ai.jriyuu.io/v1/chat/completions": 400 Bad Request {"message":"litellm.BadRequestError: OpenAIException - Invalid request parameters. Please check your input and try again.\n\nLiteLLM: model group 'ci-reviewer' failed with the error above and no fallback model group was found for it, so the request was not retried on another model. Fallbacks are configured for: qwen3.8-27b, qwen3.8-27b-fast, reverse-engineer-core. Add a fallbacks entry for that model group (Router fallbacks or proxy router_settings.fallbacks) to retry on another model.","type":"invalid_request_error","param":"messages[1].content","code":"400"} (keeping earlier findings) [ocr] Plan phase failed for group "src/labs.example.yaml,src/workbench/__init__.py,src/workbench/__main__.py,src/workbench/config.py,src/workbench/guard.py,src/workbench/server.py,src/workbench/setup.py": plan request: POST "https://ai.jriyuu.io/v1/chat/completions": 504 Gateway Timeout : <html> <head><title>504 Gateway Time-out</title></head> <body> <center><h1>504 Gateway Time-out</h1></center> <hr><center>nginx</center> </body> </html> (continuing without plan) [ocr] No tool calls parsed for src/labs.example.yaml,src/workbench/__init__.py,src/workbench/__main__.py,src/workbench/config.py,src/workbench/guard.py,src/workbench/server.py,src/workbench/setup.py, retrying... [ocr] Subtask error for group "workbench Python application & config": LLM completion error: caller deadline exceeded: context deadline exceeded [ocr] Summary: 15 file(s) reviewed, 4 comment(s), ~486877 token(s) used (input: ~356269, output: ~130608), cache(read: ~220608, write: ~0), 30m13s elapsed [ocr] Session: e7057185-0384-48d2-8df5-305d23179907 Review partially complete: 4 finding(s); 7 of 15 selected item(s) failed. ─── tests/mock_litellm.py:138-139 ─── [maintainability · low] The CLI args are consumed positionally with no validation: running `python3 tests/mock_litellm.py` with no args crashes with a raw `IndexError` traceback from `int(sys.argv[1])` instead of the usage message the docstring already provides, and `LOG_PATH` is bound from `sys.argv` at module import time, coupling the module to the process argv even when it's only imported. Cheap fix: check argc once in the `__main__` block and print the usage. if __name__ == "__main__": + if not 2 <= len(sys.argv) <= 3: + sys.exit("usage: python3 tests/mock_litellm.py <port> <request-log-file>") ThreadingHTTPServer(("127.0.0.1", int(sys.argv[1])), Handler).serve_forever() ─── tests/mock_litellm.py:44-45 ─── [bug · low] A request body that parses as valid JSON but isn't an object (e.g. `null`, `[]`, or a bare string) slips past this `except`: `body.get('model')` then raises `AttributeError`, which isn't in the caught tuple, so the handler thread dies in socketserver's handle_error and the client gets a dropped connection instead of the 400 this block is meant to produce for malformed input. Normalizing the shape inside the try (e.g. `if not isinstance(body, dict): raise ValueError("request body must be a JSON object")`) routes it through the existing 400 path. body = json.loads(self.rfile.read(int(self.headers.get("content-length", 0)))) + if not isinstance(body, dict): + raise ValueError("request body must be a JSON object") auth = self.headers.get("authorization") or self.headers.get("x-api-key") ─── src/entrypoint.sh:32-35 ─── [bug · low] The condition doesn't match the comment: docker stop is handled by the TERM/INT trap above (`wait -n` is interruptible, so the trap runs and exits 0 before this line is reached). Signal numbers 129-255 observed at `wait -n` actually mean a lab server itself was killed by a signal — e.g. OOM-kill (137, plausible given one 4G Ghidra heap per running job under a 20g limit) or segfault (139) — and that case is silently swallowed here: no "a lab server exited" line, no diagnostic before the automatic restart. Consider logging it (and exiting 1 so restart policies like on-failure still restart): if [ "$rc" -gt 128 ]; then - # A stop signal interrupted the wait (docker stop): the trap already shut down cleanly. - exit 0 + # A lab server died from a signal (e.g. OOM-killed 137, segfault 139); the TERM/INT + # trap already handles docker stop, so anything seen here is a crash worth a log line. + echo "craft-workbench: a lab server died from a signal (rc=$rc); stopping" + kill "${pids[@]}" 2>/dev/null || true + wait + exit 1 fi ─── src/entrypoint.sh:21-25 ─── [test · low] The bootstrap enforces an unmachined contract between three sides: setup must print exactly "<lab> <port>" to stdout (stderr must carry the logs), serve must run as `python -m workbench serve <lab>` with the lab name as the sole positional arg, and the image HEALTHCHECK must match the child cmdline via pgrep -f "[w]orkbench serve". It all holds today, but a format or argv drift fails silently — zero job APIs, container stays up, only symptom is pgrep going unhealthy. Nothing in the repo (only tests/mock_litellm.py was added) exercises this layer; a smoke test running setup/serve with WORKBENCH_* overrides pointed at temp dirs would lock the stdout format, the module invocation form and the healthcheck pattern together. - while read -r lab _port; do - [ -n "$lab" ] || continue - "$PY" -m workbench serve "$lab" & - pids+=("$!") - done <<< "$labs" + # e.g. a smoke test that asserts, for a temp labs.yaml reached via WORKBENCH_* env: + # setup output lines match '^[a-z][a-z0-9-]* [0-9]+$' (this loop's parse assumptions) + # a spawned `python -m workbench serve <name>` cmdline matches pgrep -f "[w]orkbench serve" + # `python -m workbench serve` with no arg / unknown arg exits non-zero [ocr] WARNING [review_round_failed] .forgejo/workflows/build.yml,docker/Dockerfile,docker/compose.yaml,docker/craft-workbench.xml,src/bin/ca,src/bin/chromium,src/entrypoint.sh: round 2: LLM completion error: POST "https://ai.jriyuu.io/v1/chat/completions": 400 Bad Request {"message":"litellm.BadRequestError: OpenAIException - Invalid request parameters. Please check your input and try again.\n\nLiteLLM: model group 'ci-reviewer' failed with the error above and no fallback model group was found for it, so the request was not retried on another model. Fallbacks are configured for: qwen3.8-27b, qwen3.8-27b-fast, reverse-engineer-core. Add a fallbacks entry for that model group (Router fallbacks or proxy router_settings.fallbacks) to retry on another model.","type":"invalid_request_error","param":"messages[1].content","code":"400"} LLM retry report summary: 5 of 30 requests affected -- 3 requests failed, 2 requests recovered after retry Review planning (2 requests): - src/labs.example.yaml,src/workbench/__init__.py,src/workbench/__main__.py,src/workbench/config.py,src/workbench/guard.py,src/workbench/server.py,src/workbench/setup.py: timed out (HTTP 504) -> timed out (HTTP 504) -> timed out (HTTP 504) -> timed out (HTTP 504) -> timed out (HTTP 504) -> timed out (HTTP 504) -> failed - .forgejo/workflows/build.yml,docker/Dockerfile,docker/compose.yaml,docker/craft-workbench.xml,src/bin/ca,src/bin/chromium,src/entrypoint.sh: timed out (HTTP 504) -> succeeded Core review (3 requests): - .forgejo/workflows/build.yml,docker/Dockerfile,docker/compose.yaml,docker/craft-workbench.xml,src/bin/ca,src/bin/chromium,src/entrypoint.sh: rejected by provider (HTTP 400) -> failed - src/labs.example.yaml,src/workbench/__init__.py,src/workbench/__main__.py,src/workbench/config.py,src/workbench/guard.py,src/workbench/server.py,src/workbench/setup.py: timed out -> failed - .forgejo/workflows/build.yml,docker/Dockerfile,docker/compose.yaml,docker/craft-workbench.xml,src/bin/ca,src/bin/chromium,src/entrypoint.sh: timed out (HTTP 504) -> succeeded Per-attempt detail: --format json (retry_report). ``` </details> _{Open Code Review via ci-actions · model `ci-reviewer` · [job log](https://nidavellir.jriyuu.io/jriyuu-labs/craft-workbench/actions/runs/40)}

Jriyuu added 1 commit 2026-11-02 09:34:18 +11:00

Upload files to "src/workbench" 92e1aa4728

Jriyuu added 1 commit 2026-11-02 09:34:31 +11:00

Upload files to "src" 9a72500887

Jriyuu added 1 commit 2026-11-02 09:34:48 +11:00

Upload files to "tests" 28a489d081

Jriyuu added 1 commit 2026-11-02 09:50:56 +11:00

Upload files to "docker" 970d92ee9c

Jriyuu added 1 commit 2026-11-02 09:51:17 +11:00

Upload files to "src/workbench" af159c77db

Jriyuu added 1 commit 2026-11-02 09:51:34 +11:00

Upload files to "tests" a99fee0e6c

Jriyuu added 1 commit 2026-11-02 10:56:18 +11:00

Upload files to "src/workbench" 09100f543b

Jriyuu added 1 commit 2026-11-02 10:56:38 +11:00

Upload files to "src/bin" 847f5cdd12

Jriyuu added 1 commit 2026-11-02 10:57:03 +11:00

Upload files to "src" c2a9adafab

Jriyuu added 1 commit 2026-11-02 10:57:22 +11:00

Upload files to "docker" 209cf50b8f

Jriyuu added 1 commit 2026-11-02 10:57:37 +11:00

Upload files to "/" 595bd17df0

Jriyuu added 1 commit 2026-11-02 10:57:52 +11:00

Upload files to ".forgejo/workflows" 84ba04de94

Jriyuu added 1 commit 2026-11-02 11:37:56 +11:00

Upload files to "tests" afed4540ed

Jriyuu added 1 commit 2026-11-02 11:38:23 +11:00

Upload files to "src/workbench" ba438773df

Jriyuu added 1 commit 2026-11-02 11:38:34 +11:00

Upload files to "src" 6518f1f65c

Jriyuu added 1 commit 2026-11-02 11:38:54 +11:00

Upload files to "docker" d7455b8257

Jriyuu added 1 commit 2026-11-02 11:39:15 +11:00

Upload files to "/" 331ad6ee3e

Jriyuu added 1 commit 2026-11-02 11:39:28 +11:00

Upload files to ".forgejo/workflows" 9ba0d29be8

Jriyuu added 1 commit 2026-11-02 12:21:12 +11:00

Upload files to "src/workbench" 62bcfc2a6b

Jriyuu merged commit 01f33e200a into main 2026-11-02 13:08:37 +11:00

Jriyuu referenced this pull request from a commit 2026-11-02 13:08:38 +11:00

Merge pull request 'int-upload' (#1) from int-upload into main

Jriyuu deleted branch int-upload 2026-11-02 13:08:43 +11:00

Sign in to join this conversation.

Reviewers

No reviewers

Labels

Clear labels

No items

No labels

Milestone

Clear milestone

No items

No milestone

Projects

Clear projects

No items

No project

Assignees

Clear assignees

No assignees

2 participants

Notifications

Subscribe

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference

jriyuu-labs/craft-workbench!1

No description provided.