Fix MCP server name

Currently, Claude Code complains that it won't use the MCP server
because of the incorrect name and it falls back to the fall back
option, which is less efficient.

Because the skills reference tools that don't exist
(mcp__content-mcp__*), when a skill runs it will never find the MCP
tools and will always fall back to the slower, less precise
filesystem-based alternatives. The skills still work due to the fallback
mechanism, but they miss the benefits of the MCP server (structured
results, semantic search via find_similar_requirements, PDF parsing,
etc.).

The correct name is `content-agent`.
See https://github.com/ComplianceAsCode/content-mcp/blob/main/README.md

Author: jan-cerny

.claude/skills/build-product/SKILL.md

@@ -11,16 +11,16 @@ Build a ComplianceAsCode product.
 
 ## Tool Strategy
 
-This skill uses `mcp__content-mcp__*` tools when available (preferred — deterministic, structured results). When the MCP server is not configured, fall back to filesystem-based alternatives noted as **Fallback** in each step. See `.claude/skills/shared/mcp_fallbacks.md` for detailed fallback procedures. The skill must complete successfully either way.
+This skill uses `mcp__content-agent__*` tools when available (preferred — deterministic, structured results). When the MCP server is not configured, fall back to filesystem-based alternatives noted as **Fallback** in each step. See `.claude/skills/shared/mcp_fallbacks.md` for detailed fallback procedures. The skill must complete successfully either way.
 
 ## Phase 1: Validate Product
 
 1. **Check if product is valid**:
-   Use `mcp__content-mcp__get_product_details` with `product_id=$ARGUMENTS` to validate the product exists and get its metadata.
+   Use `mcp__content-agent__get_product_details` with `product_id=$ARGUMENTS` to validate the product exists and get its metadata.
    **Fallback**: Read `products/$ARGUMENTS/product.yml` directly. If the file doesn't exist, the product is invalid.
 
 2. **If product not found**, list available products:
-   Use `mcp__content-mcp__list_products` to get all available products.
+   Use `mcp__content-agent__list_products` to get all available products.
    **Fallback**: Run `ls products/` to list available product directories.
 
 3. **If no product specified**, ask user using AskUserQuestion:
@@ -72,7 +72,7 @@ Expected artifacts in `build/`:
    - Non-zero = Build failed
 
 2. **Verify key artifacts exist**:
-   Use `mcp__content-mcp__get_datastream_info` with `product=$ARGUMENTS` to verify the datastream was built successfully and get artifact details.
+   Use `mcp__content-agent__get_datastream_info` with `product=$ARGUMENTS` to verify the datastream was built successfully and get artifact details.
    **Fallback**: Check files directly:
    ```bash
    ls -la build/ssg-$ARGUMENTS-ds.xml

.claude/skills/create-rule/SKILL.md

@@ -11,7 +11,7 @@ Create a new security rule for ComplianceAsCode. This skill handles both templat
 
 ## Tool Strategy
 
-This skill uses `mcp__content-mcp__*` tools when available (preferred — deterministic, structured results). When the MCP server is not configured, fall back to filesystem-based alternatives noted as **Fallback** in each step. See `.claude/skills/shared/mcp_fallbacks.md` for detailed fallback procedures. The skill must complete successfully either way.
+This skill uses `mcp__content-agent__*` tools when available (preferred — deterministic, structured results). When the MCP server is not configured, fall back to filesystem-based alternatives noted as **Fallback** in each step. See `.claude/skills/shared/mcp_fallbacks.md` for detailed fallback procedures. The skill must complete successfully either way.
 
 ## Phase 1: Validate Input
 
@@ -20,7 +20,7 @@ This skill uses `mcp__content-mcp__*` tools when available (preferred — determ
    - Example valid IDs: `sshd_max_auth_tries`, `accounts_password_minlen`
 
 2. **Check if rule already exists**:
-   Use `mcp__content-mcp__search_rules` with `query=$ARGUMENTS` to check if a rule with this ID already exists.
+   Use `mcp__content-agent__search_rules` with `query=$ARGUMENTS` to check if a rule with this ID already exists.
    **Fallback**: Use `Glob` to search for `**/$ARGUMENTS/rule.yml`.
    - If rule exists, inform user and ask if they want to modify it instead
 
@@ -42,14 +42,14 @@ Use AskUserQuestion to ask the user:
 
 If user chose templated rule:
 
-1. **List available templates** using `mcp__content-mcp__list_templates` to get all available templates with their descriptions.
+1. **List available templates** using `mcp__content-agent__list_templates` to get all available templates with their descriptions.
    **Fallback**: Run `ls shared/templates/` to list available template directories.
 
 2. **Present available templates** from the list obtained in step 1 and help the user select the right one based on their use case.
 
 3. **Ask for template selection** using AskUserQuestion
 
-4. **Get template parameter schema** using `mcp__content-mcp__get_template_schema` with `template_name=<selected_template>` to get the full parameter schema, supported languages, and documentation.
+4. **Get template parameter schema** using `mcp__content-agent__get_template_schema` with `template_name=<selected_template>` to get the full parameter schema, supported languages, and documentation.
    **Fallback**: Read `shared/templates/<template_name>/template.yml` or `template.py` for parameter definitions. Check existing rules using this template for usage examples.
 
 5. **Collect template variables**:
@@ -110,7 +110,7 @@ Collect the following information using AskUserQuestion or prompts:
 
 ### For Templated Rules
 
-Use `mcp__content-mcp__generate_rule_from_template` with:
+Use `mcp__content-agent__generate_rule_from_template` with:
 - `template_name`: The selected template name
 - `parameters`: Template variables collected from the user
 - `rule_id`: $ARGUMENTS
@@ -122,7 +122,7 @@ This generates the rule directory and rule.yml with the template configuration a
 
 ### For Non-Templated Rules
 
-Use `mcp__content-mcp__generate_rule_boilerplate` with:
+Use `mcp__content-agent__generate_rule_boilerplate` with:
 - `rule_id`: $ARGUMENTS
 - `title`: The rule title
 - `description`: The rule description
@@ -323,7 +323,7 @@ Most profiles in the project reference control files rather than listing rules d
 
 ### Step 1: List Available Profiles
 
-Use `mcp__content-mcp__list_profiles` with `product=<product>` to list all available profiles for each target product.
+Use `mcp__content-agent__list_profiles` with `product=<product>` to list all available profiles for each target product.
    **Fallback**: Run `ls products/<product>/profiles/*.profile` to list profiles.
 
 ### Step 2: Ask User Which Profile(s)
@@ -472,7 +472,7 @@ ls tests/data/profile_stability/
    cat <rule_directory>/rule.yml
    ```
 
-2. **Validate rule YAML** using `mcp__content-mcp__validate_rule_yaml` with the content of the generated rule.yml. This validates syntax, structure, and reference formats.
+2. **Validate rule YAML** using `mcp__content-agent__validate_rule_yaml` with the content of the generated rule.yml. This validates syntax, structure, and reference formats.
    **Fallback**: Validate against the JSON schema:
    ```bash
    python3 -c "
@@ -488,7 +488,7 @@ ls tests/data/profile_stability/
    " <path_to_rule.yml>
    ```
 
-   For control files, use `mcp__content-mcp__validate_control_file` with the control file path.
+   For control files, use `mcp__content-agent__validate_control_file` with the control file path.
    **Fallback**: Validate against the control schema:
    ```bash
    python3 -c "

.claude/skills/draft-pr/SKILL.md

@@ -9,7 +9,7 @@ Generate a pull request description for the current branch by analyzing commits
 
 ## Tool Strategy
 
-This skill uses `mcp__content-mcp__*` tools when available (preferred — deterministic, structured results). When the MCP server is not configured, fall back to filesystem-based alternatives noted as **Fallback** in each step. See `.claude/skills/shared/mcp_fallbacks.md` for detailed fallback procedures. The skill must complete successfully either way.
+This skill uses `mcp__content-agent__*` tools when available (preferred — deterministic, structured results). When the MCP server is not configured, fall back to filesystem-based alternatives noted as **Fallback** in each step. See `.claude/skills/shared/mcp_fallbacks.md` for detailed fallback procedures. The skill must complete successfully either way.
 
 ## Phase 1: Gather Branch Data
 
@@ -64,10 +64,10 @@ Look for product indicators:
 ### 2.3 Read Key Changed Files
 
 For significant changed files, use MCP functions to get structured metadata:
-- **Rules**: Use `mcp__content-mcp__get_rule_details` with the rule ID to get title, description, rationale, template, severity, references, and platform info
-- **Profiles**: Use `mcp__content-mcp__get_profile_details` with product and profile ID to get profile structure and rule selections
-- **Controls**: Use `mcp__content-mcp__get_control_details` to understand control framework structure
-- **Templates**: Use `mcp__content-mcp__get_template_schema` to get template parameter info
+- **Rules**: Use `mcp__content-agent__get_rule_details` with the rule ID to get title, description, rationale, template, severity, references, and platform info
+- **Profiles**: Use `mcp__content-agent__get_profile_details` with product and profile ID to get profile structure and rule selections
+- **Controls**: Use `mcp__content-agent__get_control_details` to understand control framework structure
+- **Templates**: Use `mcp__content-agent__get_template_schema` to get template parameter info
 
 **Fallback**: Read the files directly — `rule.yml`, `.profile`, control YAML, and template files in `shared/templates/<name>/`.
 

.claude/skills/inspect-control/SKILL.md

@@ -11,14 +11,14 @@ Analyze a control file and report its current state: how many requirements are m
 
 ## Tool Strategy
 
-This skill uses `mcp__content-mcp__*` tools when available (preferred — deterministic, structured results). When the MCP server is not configured, fall back to filesystem-based alternatives noted as **Fallback** in each step. See `.claude/skills/shared/mcp_fallbacks.md` for detailed fallback procedures. The skill must complete successfully either way.
+This skill uses `mcp__content-agent__*` tools when available (preferred — deterministic, structured results). When the MCP server is not configured, fall back to filesystem-based alternatives noted as **Fallback** in each step. See `.claude/skills/shared/mcp_fallbacks.md` for detailed fallback procedures. The skill must complete successfully either way.
 
 ## Phase 1: Validate and Load Control File
 
 1. **Parse arguments**: Extract control_id from `$ARGUMENTS`.
 
-2. **Validate control exists**: Call `mcp__content-mcp__get_control_stats` with `control_id=$ARGUMENTS`.
-   - If the tool returns an error (control not found), call `mcp__content-mcp__list_controls` and present available control files via `AskUserQuestion`:
+2. **Validate control exists**: Call `mcp__content-agent__get_control_stats` with `control_id=$ARGUMENTS`.
+   - If the tool returns an error (control not found), call `mcp__content-agent__list_controls` and present available control files via `AskUserQuestion`:
      - "Control file '$ARGUMENTS' not found. Which control file do you want to inspect?"
      - Options: top 4 most relevant from the list + Other
    - **Fallback**: Look for `controls/$ARGUMENTS.yml` or `products/**/controls/$ARGUMENTS.yml`. If not found, run `ls controls/*.yml` and `ls products/*/controls/*.yml` to list available control files. To compute stats, read the control YAML and count requirements grouped by `status` field.
@@ -51,7 +51,7 @@ Only show status rows that have non-zero counts.
 
 ## Phase 3: List Unmapped Requirements
 
-1. Call `mcp__content-mcp__list_unmapped_requirements` with `control_id` and default `status_filter` (pending).
+1. Call `mcp__content-agent__list_unmapped_requirements` with `control_id` and default `status_filter` (pending).
    **Fallback**: Read the control YAML and filter requirements where `status` is `pending` or where `rules:` is empty/missing.
 
 2. Present the unmapped work queue:
@@ -86,7 +86,7 @@ Present actionable next steps based on the analysis:
 
 - To map all unmapped requirements interactively: `/map-controls {control_id} --product <product>`
 - To map a single requirement: `/map-requirement {control_id} <requirement_id> --product <product>`
-- To view full control details: use `mcp__content-mcp__get_control_details`
+- To view full control details: use `mcp__content-agent__get_control_details`
 
 **Tip**: If you have the source security policy document (PDF, Markdown, or HTML), pass it with `--policy <path>` to enrich mapping with the original requirement text. This improves cross-framework matching by using the full policy context instead of just the control file's summary. Example:
   `/map-controls {control_id} --product <product> --policy security_policies/<file>`