Improve promise tag format to show policy name with checkmark

Change the promise tag body from generic "addressed" text to a more
compelling "✓ Policy Name" format that clearly shows which policy was
addressed. This provides better visibility for end users when viewing
the LLM's response.

Author: claude

.claude/commands/deepwork_policy.define.md

@@ -203,7 +203,7 @@ Policies are evaluated automatically when you finish working on a task. The syst
 3. Skips policies where safety patterns also matched
 4. Prompts you with instructions for any triggered policies
 
-You can mark a policy as addressed by including `<promise policy="Policy Name">considered</promise>` in your response. This tells the system you've already handled that policy's requirements.
+You can mark a policy as addressed by including `<promise policy="Policy Name">✓ Policy Name</promise>` in your response (replace Policy Name with the actual policy name). This tells the system you've already handled that policy's requirements.
 
 
 ## Inputs

.deepwork/jobs/deepwork_policy/steps/define.md

@@ -171,4 +171,4 @@ Policies are evaluated automatically when you finish working on a task. The syst
 3. Skips policies where safety patterns also matched
 4. Prompts you with instructions for any triggered policies
 
-You can mark a policy as addressed by including `<promise policy="Policy Name">considered</promise>` in your response. This tells the system you've already handled that policy's requirements.
+You can mark a policy as addressed by including `<promise policy="Policy Name">✓ Policy Name</promise>` in your response (replace Policy Name with the actual policy name). This tells the system you've already handled that policy's requirements.

.gemini/commands/deepwork_policy/define.toml

@@ -208,7 +208,7 @@ Policies are evaluated automatically when you finish working on a task. The syst
 3. Skips policies where safety patterns also matched
 4. Prompts you with instructions for any triggered policies
 
-You can mark a policy as addressed by including `<promise policy="Policy Name">considered</promise>` in your response. This tells the system you've already handled that policy's requirements.
+You can mark a policy as addressed by including `<promise policy="Policy Name">✓ Policy Name</promise>` in your response (replace Policy Name with the actual policy name). This tells the system you've already handled that policy's requirements.
 
 
 ## Inputs

README.md

@@ -297,7 +297,7 @@ Policies automatically enforce team guidelines when files change:
 1. When you start a Claude Code session, the baseline git state is captured
 2. When the agent finishes, changed files are compared against policy triggers
 3. If policies fire (trigger matches, no safety match), Claude is prompted to address them
-4. Use `<promise policy="Policy Name">addressed</promise>` to mark policies as handled
+4. Use `<promise policy="Policy Name">✓ Policy Name</promise>` to mark policies as handled
 
 **Use cases**:
 - Keep documentation in sync with code changes

doc/architecture.md

@@ -1033,7 +1033,7 @@ Policies are defined in `.deepwork.policy.yml` at the project root:
      - The agent hasn't marked it with a `<promise>` tag
      - → The policy fires
    - If policies fire, Claude is prompted to address them
-4. **Promise Tags**: Agents can mark policies as addressed by including `<promise policy="Policy Name">addressed</promise>` in their response
+4. **Promise Tags**: Agents can mark policies as addressed by including `<promise policy="Policy Name">✓ Policy Name</promise>` in their response
 
 ### Hook Integration
 

src/deepwork/hooks/evaluate_policies.py

@@ -63,8 +63,8 @@ def format_policy_message(policies: list) -> str:
     lines = ["## DeepWork Policies Triggered", ""]
     lines.append(
         "Comply with the following policies. "
-        'To mark a policy as addressed, include `<promise policy="Policy Name">addressed</promise>` '
-        "in your response."
+        'To mark a policy as addressed, include `<promise policy="Policy Name">✓ Policy Name</promise>` '
+        "in your response (replace Policy Name with the actual policy name)."
     )
     lines.append("")
 

src/deepwork/standard_jobs/deepwork_policy/steps/define.md

@@ -171,4 +171,4 @@ Policies are evaluated automatically when you finish working on a task. The syst
 3. Skips policies where safety patterns also matched
 4. Prompts you with instructions for any triggered policies
 
-You can mark a policy as addressed by including `<promise policy="Policy Name">considered</promise>` in your response. This tells the system you've already handled that policy's requirements.
+You can mark a policy as addressed by including `<promise policy="Policy Name">✓ Policy Name</promise>` in your response (replace Policy Name with the actual policy name). This tells the system you've already handled that policy's requirements.

tests/unit/test_evaluate_policies.py

@@ -9,7 +9,7 @@ class TestExtractPromiseTags:
 
     def test_extracts_double_quote_promise(self) -> None:
         """Test extracting promise with double quotes."""
-        text = '<promise policy="Update Docs">addressed</promise>'
+        text = '<promise policy="Update Docs">✓ Update Docs</promise>'
         result = extract_promise_tags(text)
         assert result == {"Update Docs"}