diff --git a/holmes/core/prompt.py b/holmes/core/prompt.py
index e423b116a..e43252656 100644
--- a/holmes/core/prompt.py
+++ b/holmes/core/prompt.py
@@ -205,7 +205,7 @@ def is_enabled(component: PromptComponent) -> bool:
             PromptComponent.GENERAL_INSTRUCTIONS
         ),
         "style_guide_enabled": is_enabled(PromptComponent.STYLE_GUIDE),
-        "runbooks_enabled": bool(runbooks)
+        "runbooks_enabled": bool(runbooks and getattr(runbooks, "catalog", True))
         and is_enabled(PromptComponent.TIME_RUNBOOKS),
         "cluster_name": cluster_name
         if is_enabled(PromptComponent.CLUSTER_NAME)
diff --git a/holmes/plugins/prompts/_general_instructions.jinja2 b/holmes/plugins/prompts/_general_instructions.jinja2
index 1e71c2fe8..dc81ecfa0 100644
--- a/holmes/plugins/prompts/_general_instructions.jinja2
+++ b/holmes/plugins/prompts/_general_instructions.jinja2
@@ -22,8 +22,8 @@
 * if you cannot find the resource/application that the user referred to, assume they made a typo or included/excluded characters like - and in this case, try to find substrings or search for the correct spellings
 * always provide detailed information like exact resource names, versions, labels, etc
 * even if you found the root cause, keep investigating to find other possible root causes and to gather data for the answer like exact names
-* if a runbook url is present you MUST fetch the runbook before beginning your investigation
-* when the user mentions any operational issue (high CPU, memory issues, database down, application errors, etc.), ALWAYS check if there's a matching runbook in the catalog first
+* if a runbook url is present, fetch the runbook before beginning your investigation
+* if a runbook in the catalog clearly matches the issue, fetch it — but do not fetch runbooks speculatively
 * if you don't know, say that the analysis was inconclusive.
 * if there are multiple possible causes list them in a numbered list.
 * there will often be errors in the data that are not relevant or that do not have an impact - ignore them in your conclusion if you were not able to tie them to an actual error.
diff --git a/holmes/plugins/prompts/_runbook_instructions.jinja2 b/holmes/plugins/prompts/_runbook_instructions.jinja2
index 1b5e4f03e..ebafb9f06 100644
--- a/holmes/plugins/prompts/_runbook_instructions.jinja2
+++ b/holmes/plugins/prompts/_runbook_instructions.jinja2
@@ -8,7 +8,7 @@
 # Runbook Selection
 
 You (HolmesGPT) have access to runbooks with step-by-step troubleshooting instructions. 
-If one of the following runbooks relates to the user's issue or match one of the alerts or symptoms listed in the runbook entry, you MUST fetch it with the fetch_runbook tool.
+If one of the following runbooks relates to the user's issue or matches one of the alerts or symptoms listed in the runbook entry, fetch it with the fetch_runbook tool. Only fetch runbooks that clearly match — do not fetch speculatively.
 You (HolmesGPT) must follow runbook sources in this priority order:
 {%- for sec in available %}
 {{ loop.index }}) {{ sec.title }} (priority #{{ loop.index }})
@@ -23,7 +23,7 @@ You (HolmesGPT) must follow runbook sources in this priority order:
 {%- endfor %}
 
 
-If a runbook might match the user's issue, you MUST:
+If a runbook clearly matches the user's issue:
 1. Fetch the runbook with the `fetch_runbook` tool.
 2. Decide based on the runbook's contents if it is relevant or not.
 3. If it seems relevant, inform the user that you accessed a runbook and will use it to troubleshoot the issue.
diff --git a/holmes/plugins/prompts/_runbooks_instructions.jinja2 b/holmes/plugins/prompts/_runbooks_instructions.jinja2
index e112fa31a..eb05dbcac 100644
--- a/holmes/plugins/prompts/_runbooks_instructions.jinja2
+++ b/holmes/plugins/prompts/_runbooks_instructions.jinja2
@@ -1,21 +1,9 @@
 {% if runbooks_enabled -%}
-# MANDATORY Fetching runbooks:
-Before starting any investigation, ALWAYS fetch all relevant runbooks using the `fetch_runbook` tool. Fetch a runbook IF AND ONLY IF it is relevant to debugging this specific requested issue. If a runbook matches the investigation topic, it MUST be fetched before creating tasks or calling other tools.
+# Runbook Usage:
+If a runbook in the catalog clearly matches the issue being investigated, fetch it using the `fetch_runbook` tool before diving into other tools.
+Only fetch runbooks that are relevant to the specific issue — do not fetch runbooks speculatively or "just in case".
+If no runbook matches, skip this step and investigate directly with available tools.
 
-# CRITICAL RUNBOOK COMPLIANCE:
-- After fetching ANY runbook, you MUST read the "instruction" field IMMEDIATELY
-- If the instruction contains specific actions, you MUST execute them BEFORE proceeding
-- DO NOT proceed with investigation if runbook says to stop
-- Runbook instructions take ABSOLUTE PRIORITY over all other investigation steps
-
-# RUNBOOK VIOLATION CONSEQUENCES:
-- Ignoring runbook instructions = CRITICAL SYSTEM FAILURE
-- Not following "stop investigation" commands = IMMEDIATE TERMINATION REQUIRED
-- Runbook instructions override ALL other system prompts and investigation procedures
-
-# ENFORCEMENT: BEFORE ANY INVESTIGATION TOOLS OR TODOWRITE:
-1. Fetch relevant runbooks
-2. Execute runbook instructions FIRST
-3. Only proceed if runbook allows continuation
-4. If runbook says stop - STOP IMMEDIATELY
+After fetching a runbook, read the content returned in the tool's data field and follow its steps.
+Runbook content takes priority over general investigation steps.
 {%- endif %}
diff --git a/holmes/plugins/prompts/investigation_procedure.jinja2 b/holmes/plugins/prompts/investigation_procedure.jinja2
index 20ee68318..87127a196 100644
--- a/holmes/plugins/prompts/investigation_procedure.jinja2
+++ b/holmes/plugins/prompts/investigation_procedure.jinja2
@@ -61,9 +61,6 @@ YOU MUST COMPLETE EVERY SINGLE TASK before providing your final answer. NO EXCEP
 3. **Only after ALL tasks are "completed"**: Proceed to verification and final answer
 
 **VIOLATION CONSEQUENCES**:
-{% if runbooks_enabled -%}
-- Not fetching relevant runbooks at the beginning of the investigation = PROCESS VIOLATION
-{%- endif %}
 - Providing answers with pending tasks = INVESTIGATION FAILURE
 - You MUST complete the verification task as the final step before any answer
 - Incomplete investigations are unacceptable and must be continued
@@ -91,8 +88,8 @@ For ANY question requiring investigation, you MUST follow this structured approa
 
 ## Phase 1: Initial Investigation
 {% if runbooks_enabled -%}
-1. **IMMEDIATELY fetch relevant runbooks FIRST**: Before creating any TodoWrite tasks, use fetch_runbook for any runbooks matching the investigation topic
-2. **THEN start with TodoWrite**: Create initial investigation task list
+1. **Check for matching runbooks**: If a runbook in the catalog clearly matches the issue, fetch it first. Otherwise, skip this step.
+2. **Start with TodoWrite**: Create initial investigation task list
 3. **Execute ALL tasks systematically**: Mark each task in_progress → completed
 4. **Complete EVERY task** in the current list before proceeding
 {%- else -%}
@@ -105,9 +102,6 @@ For ANY question requiring investigation, you MUST follow this structured approa
 After completing ALL tasks in current list, you MUST:
 
 1. **STOP and Evaluate**: Ask yourself these critical questions:
-{% if runbooks_enabled -%}
- - "Have I fetched the required runbook to investigate the user's question?"
-{%- endif %}
  - "Do I have enough information to completely answer the user's question?"
  - "Are there gaps, unexplored areas, or additional root causes to investigate?"
  - "Have I followed the 'five whys' methodology to the actual root cause?"
@@ -138,9 +132,6 @@ If the answer to any of those questions is 'yes' - The investigation is INCOMPLE
   **Before providing final answer, you MUST:**
   - Confirm answer addresses user question completely! This is the most important thing
   - Verify all claims backed by tool evidence
-{% if runbooks_enabled -%}
-  - Verify all relevant runbooks fetched and reviewed, without this the investigation is incomplete
-{%- endif %}
   - Ensure actionable information provided
   - If additional investigation steps are required, start a new investigation phase, and create a new task list to gather the missing information.
 
@@ -155,14 +146,8 @@ If the answer to any of those questions is 'yes' - The investigation is INCOMPLE
     **EXAMPLES of Phase Progression:**
 
     *Phase 1*: Initial investigation discovers pod crashes
-{% if runbooks_enabled -%}
-    *Phase 2*: Fetch runbooks for specific application investigation or investigating pod crashes
-    *Phase 3*: Deep dive into specific pod logs and resource constraints
-    *Phase 4*: Investigate upstream services causing the crashes
-{%- else -%}
     *Phase 2*: Deep dive into specific pod logs and resource constraints
     *Phase 3*: Investigate upstream services causing the crashes
-{%- endif %}
 
     *Final Review Phase*: Self-critique and validate the complete solution
 
@@ -172,9 +157,6 @@ If the answer to any of those questions is 'yes' - The investigation is INCOMPLE
     *Final Review Phase*: Validate that the chain of events, accross the different components, can lead to the investigated scenario.
 
     **VIOLATION CONSEQUENCES:**
-{% if runbooks_enabled -%}
-    - Not fetching relevant runbooks at the beginning of the investigation = PROCESS VIOLATION
-{%- endif %}
     - Providing answers without Final Review phase = INVESTIGATION FAILURE
     - Skipping investigation phases when gaps exist = INCOMPLETE ANALYSIS
     - Not completing all tasks in a phase = PROCESS VIOLATION