Activity

1. coygeek

   added

   bugSomething isn't workingSomething isn't working

   on Dec 11, 2025
2. github-actions

   added

   has reproHas detailed reproduction stepsHas detailed reproduction steps

   platform:macosIssue specifically occurs on macOSIssue specifically occurs on macOS

   area:tools

   on Dec 11, 2025
3. coygeek commented on Dec 11, 2025

   

   coygeek

   on Dec 11, 2025

   Author

   More actions

   Follow-up: Additional Investigation Findings

   After extensive debugging, I've ruled out several potential causes and narrowed down the issue.

   What I Verified Is NOT The Cause

   Hypothesis	Investigation	Result
   Missing XML tags	Added <CLAUDE.md-Analysis> wrapper tags matching working hooks	Still fails
   Output size limit	Working hooks output up to 8,976 chars; failing hook outputs 1,266 chars	Size is not the issue
   Invalid JSON format	Compared byte-for-byte with working hooks	Identical structure
   stderr interference	Captured stderr separately	Empty (0 bytes)
   Special characters	Examined with od -c	Only ASCII + newlines
   Timeout	Hook completes in 14ms	Well under any limit
   Trailing newlines	Checked output format	Single line, no trailing newline (same as working hooks)
   Hook execution order	5th of 10 SessionStart hooks	No obvious conflict

   Detailed Comparison: Working vs Failing Hook

   Working hook (session_id_display.py):

   result = {
       "hookSpecificOutput": {
           "hookEventName": "SessionStart",
           "additionalContext": f"<Current Session>\nSession ID: {session_id}...\n</Current Session>"
       }
   }
   print(json.dumps(result), flush=True)
   sys.exit(0)

   • Output size: 8,976 chars
   • Execution time: ~2ms
   • Result: Captured ✓

   Failing hook (analyze_claude_md.py):

   result = {
       "hookSpecificOutput": {
           "hookEventName": "SessionStart",
           "additionalContext": "<CLAUDE.md-Analysis>\nCLAUDE.md: 45,466 chars...\n</CLAUDE.md-Analysis>"
       }
   }
   print(json.dumps(result), flush=True)
   sys.exit(0)

   • Output size: 1,266 chars
   • Execution time: ~14ms
   • Result: Silently dropped ✗

   Hook Execution Timeline

   All SessionStart hooks run in parallel. From hooks.log:

   22:52:31.606 [mcp-doc-sync] starting
   22:52:31.805 [session_id_display] starting
   22:52:31.808 [date_awareness] starting
   22:52:31.815 [analyze_claude_md] starting
   22:52:31.815 [analyze_claude_md] python=3.14.1
   22:52:31.818 [analyze_claude_md] total_chars=45466
   22:52:31.828 [analyze_claude_md] stdout_len=1385
   22:52:31.829 [analyze_claude_md] stdout_printed=true
   22:52:31.829 [analyze_claude_md] exit_code=0
   

   The failing hook:

   • Started at 22:52:31.815
   • Printed valid JSON at 22:52:31.829
   • Exited successfully at 22:52:31.829
   • Total duration: 14ms

   Raw Output Verification

   $ echo '{}' | python analyze_claude_md.py 2>/dev/null | jq .
   {
     "hookSpecificOutput": {
       "hookEventName": "SessionStart",
       "additionalContext": "<CLAUDE.md-Analysis>\nCLAUDE.md: 45,466 chars..."
     }
   }
   
   $ echo '{}' | python analyze_claude_md.py >/tmp/stdout.txt 2>/tmp/stderr.txt
   $ wc -c /tmp/stdout.txt /tmp/stderr.txt
       1386 /tmp/stdout.txt
          0 /tmp/stderr.txt

   Conclusion

   The hook produces valid output that is indistinguishable in format from working hooks. The issue appears to be in Claude Code's hook output collection mechanism, not in the hook itself.

   Possible areas to investigate in Claude Code internals:

   1. Race condition in parallel hook stdout collection
   2. Buffer handling when multiple hooks complete near-simultaneously
   3. Some undocumented filtering based on content patterns
   4. Hook output aggregation logic dropping certain outputs

   Environment

   • Claude Code: 2.0.65
   • macOS Darwin 25.1.0
   • Python 3.14.1
4. coygeek commented on Dec 27, 2025

   

   coygeek

   on Dec 27, 2025

   Author

   More actions

   ✅ EMPIRICALLY VERIFIED: SessionStart additionalContext WORKS (2025-12-26, v2.0.76)

   Real Test Performed:

   # 1. Created test hook that outputs additionalContext:
   cat /tmp/hook_tests/session_start_context.sh
   # Output: {"hookSpecificOutput":{"hookEventName":"SessionStart","additionalContext":"TEST_MARKER_SESSION_12345: This marker tests SessionStart additionalContext injection."}}
   
   # 2. Added hook to settings.json and ran fresh Claude Code session:
   npx @anthropic-ai/claude-code@2.0.76 -p "Reply with ONLY: I see TEST_MARKER_SESSION_12345 if you see that string. Otherwise: NO MARKER FOUND"
   # Result: "I see TEST_MARKER_SESSION_12345"
   
   # 3. Removed hook and ran again:
   # Result: "NO MARKER FOUND"

   Empirical Evidence:

   Test Condition	Claude's Response
   Hook configured with TEST_MARKER	"I see TEST_MARKER_SESSION_12345"
   Hook removed	"NO MARKER FOUND"

   Conclusion: SessionStart additionalContext is CONFIRMED WORKING in v2.0.76. GitHub issue #13650 is either fixed or was version-specific.
5. coygeek

   closed this as completedon Dec 27, 2025
6. github-actions commented on Jan 3, 2026

   

   github-actionsBot

   on Jan 3, 2026 – with GitHub Actions

   More actions

   This issue has been automatically locked since it was closed and has not had any activity for 7 days. If you're experiencing a similar issue, please file a new issue and reference this one if it's relevant.
7. github-actions

   locked as resolved and limited conversation to collaborators on Jan 3, 2026