Preflight Checklist
Type of Behavior Issue
Claude ignored my instructions or configuration
What You Asked Claude to Do
~/.claude/CLAUDE.md includes @~/.claude/AGENTS.md and states "Follow all rules in AGENTS.md above." The model confirms it sees AGENTS.md content when asked.
What Claude Actually Did
Instructions defined inline in CLAUDE.md are consistently followed. Identical instructions in @-included files are selectively ignored under the same conditions. Moving an instruction verbatim from AGENTS.md into CLAUDE.md improves compliance immediately, with no other change.
Observed examples (both instructions were already in AGENTS.md before being added to CLAUDE.md):
- "Always branch from
origin/<target>" — ignored from AGENTS.md; followed after adding to CLAUDE.md
- "NEVER add Co-Authored-By trailers referencing Claude or Anthropic" — ignored from
AGENTS.md; followed after adding to CLAUDE.md
Expected Behavior
Instructions in @-included files should carry the same authority as inline instructions, especially when the including file explicitly states they must be followed.
Why This Is Not a Duplicate of #28158
#28158 reports intermittent general non-compliance suspected to be caused by model substitution or load balancing, with Opus 4.6 on the direct API.
This issue is a controlled differential experiment: the same instruction, moved verbatim from an @-included file into inline CLAUDE.md, is followed reliably every time with no other variable changed. The failure mode is not "instructions ignored generally" but specifically that @-included content is treated with lower authority than inline content. This points to a defect in how @-include context is weighted or surfaced, not a general compliance regression.
#34043 was incorrectly auto-closed as a duplicate of #28158 and cannot be re-opened (external repo). This issue continues that report.
Can You Reproduce This?
Yes, every time with the same prompt
Claude Model
Sonnet 4.6
Claude Code Version
2.1.74 (Claude Code)
Platform
AWS Bedrock
Preflight Checklist
Type of Behavior Issue
Claude ignored my instructions or configuration
What You Asked Claude to Do
~/.claude/CLAUDE.mdincludes@~/.claude/AGENTS.mdand states "Follow all rules in AGENTS.md above." The model confirms it seesAGENTS.mdcontent when asked.What Claude Actually Did
Instructions defined inline in
CLAUDE.mdare consistently followed. Identical instructions in@-included files are selectively ignored under the same conditions. Moving an instruction verbatim fromAGENTS.mdintoCLAUDE.mdimproves compliance immediately, with no other change.Observed examples (both instructions were already in
AGENTS.mdbefore being added toCLAUDE.md):origin/<target>" — ignored fromAGENTS.md; followed after adding toCLAUDE.mdAGENTS.md; followed after adding toCLAUDE.mdExpected Behavior
Instructions in
@-included files should carry the same authority as inline instructions, especially when the including file explicitly states they must be followed.Why This Is Not a Duplicate of #28158
#28158 reports intermittent general non-compliance suspected to be caused by model substitution or load balancing, with Opus 4.6 on the direct API.
This issue is a controlled differential experiment: the same instruction, moved verbatim from an
@-included file into inlineCLAUDE.md, is followed reliably every time with no other variable changed. The failure mode is not "instructions ignored generally" but specifically that@-included content is treated with lower authority than inline content. This points to a defect in how@-include context is weighted or surfaced, not a general compliance regression.#34043 was incorrectly auto-closed as a duplicate of #28158 and cannot be re-opened (external repo). This issue continues that report.
Can You Reproduce This?
Yes, every time with the same prompt
Claude Model
Sonnet 4.6
Claude Code Version
2.1.74 (Claude Code)
Platform
AWS Bedrock