[MODEL] Claude uses Bash (cat, grep, head) instead of dedicated tools despite system prompt prohibition

[VoxCore84]

Preflight Checklist

• I have searched existing issues and this hasn't been reported yet
• This report does NOT contain sensitive information

Type of Behavior Issue

Claude ignored my instructions or configuration

What You Asked Claude to Do

Use dedicated tools (Read, Grep, Glob, Edit, Write) instead of Bash equivalents, as specified in the system prompt.

What Claude Actually Did

The system prompt contains explicit instructions:

Do NOT use the Bash to run commands when a relevant dedicated tool is provided.

• To read files use Read instead of cat, head, tail, or sed
• To edit files use Edit instead of sed or awk
• To search for files use Glob instead of find or ls
• To search the content of files, use Grep instead of grep or rg

Despite this, Claude Code routinely:

• Runs cat file.cpp | head -50 instead of Read(file, limit=50)
• Runs grep -r "pattern" src/ instead of Grep("pattern", "src/")
• Runs find . -name "*.h" instead of Glob("**/*.h")
• Runs sed -i 's/old/new/' file instead of using Edit

Reproduction

This happens in virtually every session longer than 30 minutes. It's especially common:

1. After context compaction — the dedicated-tool preference gets compressed away
2. In subagents — spawned agents seem to have weaker adherence to system prompt rules
3. When the model is "thinking fast" — quick one-off lookups default to Bash

Why This Matters

1. Bash output is verbose and wastes tokens — cat dumps raw content without line numbers; Read provides structured output
2. Bypasses tool UX — dedicated tools show the user what's being read/searched; Bash calls are opaque
3. Breaks permission model — users configure tool-level permissions; Bash bypasses these
4. The system prompt already prohibits this — the model is violating its own instructions

Evidence

Over 200+ sessions on a large C++ codebase (TrinityCore fork, ~2M LOC), this happens in roughly 40% of sessions. The CLAUDE.md reinforces the system prompt with additional rules, and we still see the model reach for cat and grep regularly.

Environment

• Claude Code v2.1.85
• Opus 4.6
• Windows 11 Pro
• Platform: CLI (bash shell)

Cross-References

• [Bug] Claude reads files but ignores actionable instructions contained in them #32290 — Reads files but ignores actionable instructions
• [BUG] CLAUDE.md instructions completely ignored after context compaction #19471 — CLAUDE.md instructions ignored after compaction
• [Bug] Context amnesia in long sessions — constraints silently dropped as context grows #32659 — Context amnesia in long sessions

[github-actions]

Found 3 possible duplicate issues:

1. [MODEL] Frequently uses Bash tools (sed/grep/etc) when use-case is well aligned to other builtin tools (Read/Grep/etc) #19649
2. [MODEL] Explore subagent system prompt has contradictory instructions for Bash vs built-in tool usage #34992
3. [BUG] claude tends to get stuck using find/grep/sed to READ files during Explore() tool causing many permission prompts #36297

This issue will be automatically closed as a duplicate in 3 days.

• If your issue is a duplicate, please close it and 👍 the existing issue instead
• To prevent auto-closure, add a comment or 👎 this comment

🤖 Generated with Claude Code

[yurukusa]

A PreToolUse hook can enforce the "use dedicated tools" rule mechanically:

CMD=$(cat | jq -r ".tool_input.command // empty" 2>/dev/null)
[ -z "$CMD" ] && exit 0
echo "$CMD" | grep -qE "^\s*cat\s+[^|]+$" && echo "BLOCKED: Use Read tool instead of cat." >&2 && exit 2
echo "$CMD" | grep -qE "^\s*(head|tail)\s+[^|]+$" && echo "BLOCKED: Use Read tool with offset/limit." >&2 && exit 2
echo "$CMD" | grep -qE "^\s*grep\s+[^|]+$" && ! echo "$CMD" | grep -qE "\|" && echo "BLOCKED: Use Grep tool." >&2 && exit 2
echo "$CMD" | grep -qE "^\s*find\s+\S+\s+-name" && echo "BLOCKED: Use Glob tool." >&2 && exit 2
exit 0

This blocks standalone cat README.md but allows cat file | grep pattern (which has no dedicated tool equivalent). The model is forced to use Read/Grep/Glob where they exist, reducing token waste from shell overhead.

[mimuelas]

Not a duplicate — here's why:

• [MODEL] Frequently uses Bash tools (sed/grep/etc) when use-case is well aligned to other builtin tools (Read/Grep/etc) #19649: Filed as an enhancement, reports the general preference for Bash over built-in tools. My issue specifically documents this as a system prompt violation — Claude Code's own instructions explicitly say "Do NOT use Bash to run commands when a relevant dedicated tool is provided", yet the model ignores it. That's a bug, not a missing feature.
• [MODEL] Explore subagent system prompt has contradictory instructions for Bash vs built-in tool usage #34992 and [BUG] claude tends to get stuck using find/grep/sed to READ files during Explore() tool causing many permission prompts #36297: Both are specific to Explore subagents, not the main model. My issue is about the primary Claude Code model in the main conversation context.

The distinguishing factor of this issue is the system prompt contradiction: the prohibition exists in the instructions Claude receives, it's not a user preference — it's Claude Code's own rule that the model fails to follow.

Happy to close if maintainers disagree, but I believe this framing (system prompt violation vs. enhancement request) is distinct enough to track separately.

[mimuelas]

Closing in favor of #19649 which tracks the same core problem. Added my evidence there.

[github-actions]

Closing for now — inactive for too long. Please open a new issue if this is still relevant.