fix: resolve Claude CLI not found on Windows - PATH, prompt size, cwd (#1661)#1843
fix: resolve Claude CLI not found on Windows - PATH, prompt size, cwd (#1661)#1843
Conversation
…e cap, and cwd (#1661) Three root causes addressed: 1. PATH overwrite: pythonEnv.PATH was overwriting the augmented PATH (with npm globals) in spawn env. Now merges PATH entries instead, prepending python-specific paths (pywin32_system32) while preserving all augmented entries. 2. System prompt size: On Windows, SDK passes system_prompt as --system-prompt CLI arg. Large CLAUDE.md files exceed CreateProcessW's 32,768 char limit, causing misleading "Claude Code not found" error. Now caps CLAUDE.md content on Windows to stay under the limit. 3. Cross-drive cwd: Agent processes were spawned with autoBuildSource as cwd. On Windows with cross-drive setups, this caused file access issues. Now uses projectPath as cwd since all script paths are absolute. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Summary of ChangesHello @AndyMik90, I'm Gemini Code Assist1! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed! This pull request delivers critical fixes for Windows users encountering "Claude CLI not found" errors. It comprehensively addresses three distinct root causes: incorrect handling of environment variable PATH merging, limitations on command-line argument length for system prompts, and cross-drive current working directory inconsistencies. These changes significantly enhance the robustness and reliability of agent processes on Windows. Highlights
Changelog
Activity
Using Gemini Code AssistThe full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips. Invoking Gemini You can request assistance from Gemini at any point by creating a comment using either
Customization To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a Limitations & Feedback Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here. You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension. Footnotes
|
github-actions
bot
added
area/fullstack
📝 WalkthroughWalkthroughAdds Windows-specific truncation for CLAUDE.md in system prompts, changes subprocess cwd to the project directory, and merges Python PATH entries into spawned Python process environments to avoid Windows cross-drive and PATH override issues. Changes
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~20 minutes Possibly related PRs
Suggested labels
Poem
🚥 Pre-merge checks | ✅ 6✅ Passed checks (6 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing touches
🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
This pull request effectively addresses several critical Windows-specific issues that could cause "Claude code not found" errors. The fixes for PATH variable overwriting, command-line length limits, and cross-drive current working directory problems are all well-implemented and clearly explained. I have one suggestion to improve maintainability by replacing a magic number with a constant in apps/backend/core/client.py. Overall, these are excellent changes that will significantly improve the stability of the application on Windows.
apps/backend/core/client.py
Outdated
| max_claude_md_chars = 24000 - len(base_prompt) | ||
| if len(claude_md_content) > max_claude_md_chars > 0: | ||
| claude_md_content = ( | ||
| claude_md_content[:max_claude_md_chars] | ||
| + "\n\n[... CLAUDE.md truncated due to Windows command-line length limit ...]" | ||
| ) |
There was a problem hiding this comment.
The logic to truncate CLAUDE.md content on Windows is a good fix for the command-line length limit. To improve maintainability and readability, it would be better to define the magic number 24000 and the truncation message as constants.
You could define these at the top of the module:
WINDOWS_MAX_CMD_LINE_CHARS = 24000
TRUNCATION_MESSAGE = "\n\n[... CLAUDE.md truncated due to Windows command-line length limit ...]"Then use them in the function as suggested.
| max_claude_md_chars = 24000 - len(base_prompt) | |
| if len(claude_md_content) > max_claude_md_chars > 0: | |
| claude_md_content = ( | |
| claude_md_content[:max_claude_md_chars] | |
| + "\n\n[... CLAUDE.md truncated due to Windows command-line length limit ...]" | |
| ) | |
| max_claude_md_chars = WINDOWS_MAX_CMD_LINE_CHARS - len(base_prompt) | |
| if len(claude_md_content) > max_claude_md_chars > 0: | |
| claude_md_content = ( | |
| claude_md_content[:max_claude_md_chars] | |
| + TRUNCATION_MESSAGE | |
| ) |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 1
🤖 Fix all issues with AI agents
In `@apps/backend/core/client.py`:
- Around line 824-841: The current Windows truncation logic for
claude_md_content uses a fixed reserve (max_claude_md_chars = 24000 -
len(base_prompt)), which can still overflow when other ClaudeAgentOptions fields
(MCP servers, tools, settings path, etc.) are large; update the truncation to
(1) compute the reserved budget dynamically by serializing the other CLI
arguments (the fields used to build the command line from ClaudeAgentOptions)
and subtracting their length from the 32768 Windows limit, or alternatively use
a much more conservative cap (e.g., reserve 16000 chars for CLAUDE.md) so the
prompt cannot cause CreateProcessW to exceed 32768, and (2) prevent confusing
duplicate logs by making the "CLAUDE.md: included in system prompt" print
conditional (use an else branch or only print the included message when no
truncation occurred). Reference the truncation logic around is_windows(),
max_claude_md_chars, claude_md_content, and the two print statements.
🤖 Auto Claude PR ReviewMerge Verdict: 🔴 BLOCKED
Blocked: 4 CI check(s) failing. Fix CI before merge. Risk Assessment
🚨 Blocking Issues (Must Fix)
Generated by Auto Claude PR Review This review identified blockers that must be resolved before merge. Generated by Auto Claude. |
- Extract magic number 24000 into WINDOWS_MAX_SYSTEM_PROMPT_CHARS constant (set to 20000 for more conservative ~12KB CLI headroom) - Extract truncation suffix into WINDOWS_TRUNCATION_MESSAGE constant - Fix double-print when truncation occurs: only print "included in system prompt" when CLAUDE.md was NOT truncated (was_truncated flag) - Fix CI test failures: update subprocess-spawn tests to expect projectPath as cwd instead of autoBuildSource (matches the #1661 CWD change) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
There was a problem hiding this comment.
Actionable comments posted: 1
🤖 Fix all issues with AI agents
In `@apps/backend/core/client.py`:
- Around line 843-848: The truncation appends WINDOWS_TRUNCATION_MESSAGE after
slicing to max_claude_md_chars, which causes the final claude_md_content to
exceed the WINDOWS_MAX_SYSTEM_PROMPT_CHARS budget; adjust the budget calculation
by subtracting len(WINDOWS_TRUNCATION_MESSAGE) as well. Change the logic around
max_claude_md_chars/claude_md_content to compute allowed =
WINDOWS_MAX_SYSTEM_PROMPT_CHARS - len(base_prompt) -
len(WINDOWS_TRUNCATION_MESSAGE), ensure allowed > 0 before slicing, and then set
claude_md_content = claude_md_content[:allowed] + WINDOWS_TRUNCATION_MESSAGE (or
if allowed <= 0, set claude_md_content = WINDOWS_TRUNCATION_MESSAGE or an
appropriate fallback) so the final length always respects the system prompt
limit.
apps/backend/core/client.py
Outdated
| max_claude_md_chars = WINDOWS_MAX_SYSTEM_PROMPT_CHARS - len(base_prompt) | ||
| if len(claude_md_content) > max_claude_md_chars > 0: | ||
| claude_md_content = ( | ||
| claude_md_content[:max_claude_md_chars] | ||
| + WINDOWS_TRUNCATION_MESSAGE | ||
| ) |
There was a problem hiding this comment.
🧹 Nitpick | 🔵 Trivial
Truncation message length not included in the budget calculation.
After truncation, the actual CLAUDE.md content length is max_claude_md_chars + len(WINDOWS_TRUNCATION_MESSAGE) (~70 extra chars). This slightly overshoots the budget. With ~12 KB headroom it's not a practical problem, but for correctness:
🔧 Suggested fix
- max_claude_md_chars = WINDOWS_MAX_SYSTEM_PROMPT_CHARS - len(base_prompt)
+ max_claude_md_chars = WINDOWS_MAX_SYSTEM_PROMPT_CHARS - len(base_prompt) - len(WINDOWS_TRUNCATION_MESSAGE)📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
| max_claude_md_chars = WINDOWS_MAX_SYSTEM_PROMPT_CHARS - len(base_prompt) | |
| if len(claude_md_content) > max_claude_md_chars > 0: | |
| claude_md_content = ( | |
| claude_md_content[:max_claude_md_chars] | |
| + WINDOWS_TRUNCATION_MESSAGE | |
| ) | |
| max_claude_md_chars = WINDOWS_MAX_SYSTEM_PROMPT_CHARS - len(base_prompt) - len(WINDOWS_TRUNCATION_MESSAGE) | |
| if len(claude_md_content) > max_claude_md_chars > 0: | |
| claude_md_content = ( | |
| claude_md_content[:max_claude_md_chars] | |
| WINDOWS_TRUNCATION_MESSAGE | |
| ) |
🤖 Prompt for AI Agents
In `@apps/backend/core/client.py` around lines 843 - 848, The truncation appends
WINDOWS_TRUNCATION_MESSAGE after slicing to max_claude_md_chars, which causes
the final claude_md_content to exceed the WINDOWS_MAX_SYSTEM_PROMPT_CHARS
budget; adjust the budget calculation by subtracting
len(WINDOWS_TRUNCATION_MESSAGE) as well. Change the logic around
max_claude_md_chars/claude_md_content to compute allowed =
WINDOWS_MAX_SYSTEM_PROMPT_CHARS - len(base_prompt) -
len(WINDOWS_TRUNCATION_MESSAGE), ensure allowed > 0 before slicing, and then set
claude_md_content = claude_md_content[:allowed] + WINDOWS_TRUNCATION_MESSAGE (or
if allowed <= 0, set claude_md_content = WINDOWS_TRUNCATION_MESSAGE or an
appropriate fallback) so the final length always respects the system prompt
limit.
AndyMik90
left a comment
AndyMik90
left a comment
There was a problem hiding this comment.
🤖 Auto Claude PR Review
🟠 Follow-up Review: Needs Revision
🟠 Needs revision - 2 blocking issue(s) require fixes.
Resolution Status
- ✅ Resolved: 0 previous findings addressed
- ❌ Unresolved: 0 previous findings remain
- 🆕 New Issues: 4 new findings in recent changes
Finding Validation
- 🔍 Dismissed as False Positives: 0 findings were re-investigated and found to be incorrect
- ✓ Confirmed Valid: 3 findings verified as genuine issues
- 👤 Needs Human Review: 0 findings require manual verification
🚨 Blocking Issues
- quality: PATH merging selects wrong PATH key on Windows due to casing
- quality: Duplicate PATH keys with different casings passed to child process on Windows
Verdict
CI status: ✅ All 20 CI checks passing (resolved from previous 4 failures — good progress). However, 2 medium-severity findings (NEW-001, NEW-002) in the new PATH merging code were confirmed valid by the finding-validator. Both relate to Windows PATH casing: getAugmentedEnv() can produce env objects with both Path (original) and PATH (augmented) keys on Windows. The new merge logic at line 689 uses Object.keys().find() which selects the non-augmented Path first (insertion order), and the final spread produces duplicate PATH keys with different values in the child process environment. The fix is straightforward: prefer uppercase PATH key and normalize to a single key before merging, similar to how python-env-manager.ts already handles this at lines 737-738. NEW-003 (low severity, truncation budget ~116 char overshoot) is confirmed valid but practically harmless given ~12KB headroom. CodeRabbit's concern about insufficient headroom (CMT-001) was already addressed by reducing the constant from 24000 to 20000.
Review Process
Agents invoked: new-code-reviewer, comment-analyzer, finding-validator
This is an AI-generated follow-up review using parallel specialist analysis with finding validation.
Findings (4 selected of 4 total)
🟡 [NEW-001] [MEDIUM] PATH merging selects wrong PATH key on Windows due to casing
📁 apps/frontend/src/main/agent/agent-process.ts:689
On Windows, getAugmentedEnv() produces an env object with both 'Path' (original system PATH from process.env spread) and 'PATH' (augmented paths only). The Object.keys(env).find(k => k.toUpperCase() === 'PATH') call returns 'Path' first (insertion order), so the merge logic uses the non-augmented system PATH as its base. This means augmented entries (npm globals, Claude CLI path, tool-specific paths) added by getAugmentedEnv() are discarded from the merged result, potentially causing 'Claude Code not found' on Windows setups where the claude CLI is only reachable via an augmented PATH entry.
Suggested fix:
Prefer the uppercase 'PATH' key since getAugmentedEnv() always writes to it:
const envPathKey = 'PATH' in env ? 'PATH' : (Object.keys(env).find(k => k.toUpperCase() === 'PATH') || 'PATH');
Alternatively, normalize the env object to a single PATH key before merging, similar to what getPythonEnv() already does at lines 737-738 of python-env-manager.ts.
🟡 [NEW-002] [MEDIUM] Duplicate PATH keys with different casings passed to child process on Windows
📁 apps/frontend/src/main/agent/agent-process.ts:711
When envPathKey (e.g., 'Path' from env) and pythonPathKey (e.g., 'PATH' from mergedPythonEnv) differ in casing, the final spread {...env, ...mergedPythonEnv} produces an env object with BOTH 'Path' and 'PATH' keys. mergedPythonEnv's 'PATH' overwrites env's 'PATH' but env's 'Path' persists. On Windows, CreateProcessW treats env vars case-insensitively, so which PATH value the child process actually uses is undefined. getPythonEnv() in python-env-manager.ts (lines 737-738) already handles this correctly by deleting the non-normalized key, but this new merge code does not.
Suggested fix:
After finding envPathKey, normalize to a single PATH key by deleting the stale variant from env before merging:
if (envPathKey !== 'PATH' && envPathKey in env) {
env['PATH'] = env[envPathKey];
delete env[envPathKey];
}
This follows the same pattern used in python-env-manager.ts lines 737-738.
🔵 [NEW-003] [LOW] Truncation budget does not account for appended overhead (~116 chars)
📁 apps/backend/core/client.py:843
The budget max_claude_md_chars = WINDOWS_MAX_SYSTEM_PROMPT_CHARS - len(base_prompt) does not subtract the WINDOWS_TRUNCATION_MESSAGE length (~72 chars) or the header template '\n\n# Project Instructions (from CLAUDE.md)\n\n' (~44 chars). The final system prompt overshoots the 20,000-char target by ~116 characters. With ~12KB headroom to the 32,768 Windows CreateProcessW limit, this is practically inconsequential, but the budget calculation is subtly incorrect.
Suggested fix:
Subtract the overhead from the budget:
overhead = len(WINDOWS_TRUNCATION_MESSAGE) + len('\n\n# Project Instructions (from CLAUDE.md)\n\n')
max_claude_md_chars = WINDOWS_MAX_SYSTEM_PROMPT_CHARS - len(base_prompt) - overhead
🔵 [CMT-001] [LOW] [FROM COMMENTS] CodeRabbit concern about insufficient CLI overhead reserve - already addressed
📁 apps/backend/core/client.py:39
CodeRabbit flagged that the original 8KB reserve (when constant was 24000) may be insufficient for complex MCP configurations. This was already addressed in commit d646c84 by reducing the constant from 24000 to 20000, increasing headroom from ~8KB to ~12KB. No further action needed.
This review was generated by Auto Claude.
- Normalize env objects to a single uppercase 'PATH' key before merging to prevent duplicate PATH keys on Windows where process.env has 'Path' and getAugmentedEnv() writes 'PATH'. Without this, Object.keys().find() returns 'Path' first (insertion order), discarding augmented entries, and the final spread produces both 'Path' and 'PATH' keys. Follows the same pattern used in python-env-manager.ts. (#1661) - Subtract WINDOWS_TRUNCATION_MESSAGE length from the truncation budget so the final system prompt stays within WINDOWS_MAX_SYSTEM_PROMPT_CHARS. Addresses PR #1843 review findings NEW-001, NEW-002, NEW-003. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
github-actions
bot
added
size/M
There was a problem hiding this comment.
Actionable comments posted: 2
🤖 Fix all issues with AI agents
In `@apps/backend/core/client.py`:
- Around line 836-859: The current Windows truncation logic fails when
max_claude_md_chars <= 0 and ends up appending the full claude_md_content;
update the block around is_windows()/max_claude_md_chars so that if
max_claude_md_chars <= 0 you do not append CLAUDE.md at all (set was_truncated =
True), print a clear message (e.g., "CLAUDE.md: omitted (no budget under Windows
command-line limit)"), and only append claude_md_content when
max_claude_md_chars > 0 (truncating when necessary with
WINDOWS_TRUNCATION_MESSAGE); reference is_windows,
WINDOWS_MAX_SYSTEM_PROMPT_CHARS, base_prompt, WINDOWS_TRUNCATION_MESSAGE,
claude_md_content, and was_truncated to locate where to change the logic and
messages.
In `@apps/frontend/src/main/agent/agent-process.ts`:
- Around line 726-733: The spawn call currently spreads env then mergedPythonEnv
so mergedPythonEnv unconditionally overrides keys from env (including PATH which
you purposely merged earlier), and setupProcessEnvironment() still injects
duplicate Python vars; remove the redundant PYTHON-specific assignments (e.g.,
PYTHONUNBUFFERED, PYTHONIOENCODING, PYTHONUTF8) from setupProcessEnvironment()
so getPythonEnv()/getMergedPythonEnv (the code producing mergedPythonEnv) is the
canonical source, or alternatively change the spawn merge to explicitly merge
Python keys the same way PATH is merged; also update the inline comment near the
spawn (where env, mergedPythonEnv, oauthModeClearVars, apiProfileEnv are
combined) to clearly state that mergedPythonEnv has final precedence for
Python-related vars and PATH merging is preserved.
| # On Windows, the SDK passes system_prompt as a --system-prompt CLI argument. | ||
| # Windows CreateProcessW has a 32,768 character limit for the entire command line. | ||
| # When CLAUDE.md is very large, the command can exceed this limit, causing Windows | ||
| # to return ERROR_FILE_NOT_FOUND which the SDK misreports as "Claude Code not found". | ||
| # Cap CLAUDE.md content to keep total command line under the limit. (#1661) | ||
| was_truncated = False | ||
| if is_windows(): | ||
| max_claude_md_chars = ( | ||
| WINDOWS_MAX_SYSTEM_PROMPT_CHARS | ||
| - len(base_prompt) | ||
| - len(WINDOWS_TRUNCATION_MESSAGE) | ||
| ) | ||
| if len(claude_md_content) > max_claude_md_chars > 0: | ||
| claude_md_content = ( | ||
| claude_md_content[:max_claude_md_chars] | ||
| + WINDOWS_TRUNCATION_MESSAGE | ||
| ) | ||
| print( | ||
| " - CLAUDE.md: truncated (exceeded Windows command-line limit)" | ||
| ) | ||
| was_truncated = True | ||
| base_prompt = f"{base_prompt}\n\n# Project Instructions (from CLAUDE.md)\n\n{claude_md_content}" | ||
| print(" - CLAUDE.md: included in system prompt") | ||
| if not was_truncated: | ||
| print(" - CLAUDE.md: included in system prompt") |
There was a problem hiding this comment.
Edge case: when max_claude_md_chars <= 0, full CLAUDE.md is included untruncated.
If base_prompt + WINDOWS_TRUNCATION_MESSAGE already exceeds WINDOWS_MAX_SYSTEM_PROMPT_CHARS, the > 0 guard skips truncation and the entire CLAUDE.md is appended — which will definitely exceed the Windows limit. While base_prompt is currently short (~400 chars), this is a latent bug if the prompt grows.
Proposed fix: omit CLAUDE.md when budget is exhausted
if is_windows():
max_claude_md_chars = (
WINDOWS_MAX_SYSTEM_PROMPT_CHARS
- len(base_prompt)
- len(WINDOWS_TRUNCATION_MESSAGE)
)
- if len(claude_md_content) > max_claude_md_chars > 0:
+ if max_claude_md_chars <= 0:
+ claude_md_content = ""
+ print(
+ " - CLAUDE.md: omitted (base prompt already near Windows limit)"
+ )
+ was_truncated = True
+ elif len(claude_md_content) > max_claude_md_chars:
claude_md_content = (
claude_md_content[:max_claude_md_chars]
+ WINDOWS_TRUNCATION_MESSAGE
)🤖 Prompt for AI Agents
In `@apps/backend/core/client.py` around lines 836 - 859, The current Windows
truncation logic fails when max_claude_md_chars <= 0 and ends up appending the
full claude_md_content; update the block around is_windows()/max_claude_md_chars
so that if max_claude_md_chars <= 0 you do not append CLAUDE.md at all (set
was_truncated = True), print a clear message (e.g., "CLAUDE.md: omitted (no
budget under Windows command-line limit)"), and only append claude_md_content
when max_claude_md_chars > 0 (truncating when necessary with
WINDOWS_TRUNCATION_MESSAGE); reference is_windows,
WINDOWS_MAX_SYSTEM_PROMPT_CHARS, base_prompt, WINDOWS_TRUNCATION_MESSAGE,
claude_md_content, and was_truncated to locate where to change the logic and
messages.
| childProcess = spawn(pythonCommand, [...pythonBaseArgs, ...args], { | ||
| cwd, | ||
| env: { | ||
| ...env, // Already includes process.env, extraEnv, profileEnv, PYTHONUNBUFFERED, PYTHONUTF8 | ||
| ...pythonEnv, // Include Python environment (PYTHONPATH for bundled packages) | ||
| ...mergedPythonEnv, // Python env with merged PATH (preserves augmented PATH entries) | ||
| ...oauthModeClearVars, // Clear stale ANTHROPIC_* vars when in OAuth mode | ||
| ...apiProfileEnv // Include active API profile config (highest priority for ANTHROPIC_* vars) | ||
| } |
There was a problem hiding this comment.
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash
# Find pythonEnvManager.getPythonEnv() calls and implementation
rg -n "getPythonEnv" --type=ts -A5 -B5 | head -100Repository: AndyMik90/Auto-Claude
Length of output: 10363
🏁 Script executed:
#!/bin/bash
# Look for pythonEnvManager definition and initialization
rg -n "pythonEnvManager\s*=" --type=ts -A3 -B3Repository: AndyMik90/Auto-Claude
Length of output: 615
🏁 Script executed:
#!/bin/bash
# Check the file where pythonEnvManager is defined
fd pythonEnvManager --type=tsRepository: AndyMik90/Auto-Claude
Length of output: 234
🏁 Script executed:
#!/bin/bash
# Read the getPythonEnv() method implementation
sed -n '690,750p' apps/frontend/src/main/python-env-manager.tsRepository: AndyMik90/Auto-Claude
Length of output: 3204
🏁 Script executed:
#!/bin/bash
# Read more of the getPythonEnv() implementation to see all returned keys
sed -n '690,800p' apps/frontend/src/main/python-env-manager.tsRepository: AndyMik90/Auto-Claude
Length of output: 4692
🏁 Script executed:
#!/bin/bash
# Look at agent-process.ts to see mergedPythonEnv construction and context
sed -n '700,750p' apps/frontend/src/main/agent/agent-process.tsRepository: AndyMik90/Auto-Claude
Length of output: 2573
🏁 Script executed:
#!/bin/bash
# Look at env construction before it's passed to spawn
sed -n '650,730p' apps/frontend/src/main/agent/agent-process.ts | head -100Repository: AndyMik90/Auto-Claude
Length of output: 3992
🏁 Script executed:
#!/bin/bash
# Check what extraEnv, profileEnv, and other env components contain
rg -n "extraEnv|profileEnv|oauthModeClearVars" apps/frontend/src/main/agent/agent-process.ts -B2 -A2 | head -80Repository: AndyMik90/Auto-Claude
Length of output: 3624
🏁 Script executed:
#!/bin/bash
# Check setupProcessEnvironment to see what's set in env
sed -n '176,260p' apps/frontend/src/main/agent/agent-process.tsRepository: AndyMik90/Auto-Claude
Length of output: 4017
Clarify redundant Python environment variable overrides and document override semantics.
mergedPythonEnv spreads after env (line 730), causing all keys to override env values. PATH is explicitly merged (lines 706-721) to preserve augmented entries, but other keys override unconditionally: PYTHONUNBUFFERED, PYTHONIOENCODING, and PYTHONUTF8 are redundantly set in both env and mergedPythonEnv. While new keys like PYTHONDONTWRITEBYTECODE and PYTHONNOUSERSITE should override consistently, remove the redundant duplication from setupProcessEnvironment() (lines 242-244) since getPythonEnv() now provides the canonical Python configuration. Update the comment on line 730 to document this override behavior or consider an explicit merge strategy for all Python-related keys to match PATH handling.
🤖 Prompt for AI Agents
In `@apps/frontend/src/main/agent/agent-process.ts` around lines 726 - 733, The
spawn call currently spreads env then mergedPythonEnv so mergedPythonEnv
unconditionally overrides keys from env (including PATH which you purposely
merged earlier), and setupProcessEnvironment() still injects duplicate Python
vars; remove the redundant PYTHON-specific assignments (e.g., PYTHONUNBUFFERED,
PYTHONIOENCODING, PYTHONUTF8) from setupProcessEnvironment() so
getPythonEnv()/getMergedPythonEnv (the code producing mergedPythonEnv) is the
canonical source, or alternatively change the spawn merge to explicitly merge
Python keys the same way PATH is merged; also update the inline comment near the
spawn (where env, mergedPythonEnv, oauthModeClearVars, apiProfileEnv are
combined) to clearly state that mergedPythonEnv has final precedence for
Python-related vars and PATH merging is preserved.
1 participant
Summary
Fixes three root causes behind "Claude code not found" errors on Windows Kanban tasks (#1661):
pythonEnv.PATH(containing onlypywin32_system32) was spread after the augmentedenv.PATH(containing npm globals, Homebrew, etc.), overwriting it. Now merges PATH entries instead - prepends python-specific paths while preserving all augmented entries.system_promptas a--system-promptCLI argument. LargeCLAUDE.mdfiles exceed WindowsCreateProcessW's 32,768 char limit, causingERROR_FILE_NOT_FOUNDwhich the SDK misreports as "Claude Code not found". Now caps CLAUDE.md content on Windows to stay under the limit.autoBuildSource(backend dir) as cwd. On Windows cross-drive setups (e.g., backend on C:, project on D:), this caused file access issues. Now usesprojectPathas cwd since all script paths are absolute.Changes
apps/frontend/src/main/agent/agent-process.tspythonEnvwith augmented PATH instead of overwritingapps/backend/core/client.pyapps/frontend/src/main/agent/agent-manager.tsprojectPathas cwd for spec creation, task execution, and QA spawnsTest plan
Closes #1661
🤖 Generated with Claude Code
Summary by CodeRabbit
Bug Fixes
Tests