prizmkit 1.1.10 → 1.1.13

package/bundled/dev-pipeline/scripts/generate-bootstrap-prompt.py

@@ -26,7 +26,7 @@ import os
 import re
 import sys
 
-from utils import enrich_global_context, load_json_file, setup_logging
+from utils import enrich_global_context, load_json_file, read_platform_conventions, setup_logging
 
 
 DEFAULT_MAX_RETRIES = 3
@@ -407,27 +407,25 @@ def resolve_project_root(script_dir):
 
 
 def process_conditional_blocks(content, resume_phase):
-    """Handle conditional blocks based on resume_phase and pipeline mode.
+    """Handle conditional blocks based on resume_phase.
 
     Supports:
     - {{IF_FRESH_START}} / {{END_IF_FRESH_START}}
     - {{IF_RESUME}} / {{END_IF_RESUME}}
-    - {{IF_INIT_NEEDED}} / {{END_IF_INIT_NEEDED}}
-    - {{IF_INIT_DONE}} / {{END_IF_INIT_DONE}}
-    - {{IF_MODE_LITE}} / {{END_IF_MODE_LITE}}
-    - {{IF_MODE_STANDARD}} / {{END_IF_MODE_STANDARD}}
-    - {{IF_MODE_FULL}} / {{END_IF_MODE_FULL}}
+    - {{IF_RETRY}} / {{END_IF_RETRY}}
     """
     is_resume = resume_phase != "null"
 
     if is_resume:
-        content = re.sub(r"\{\{IF_RESUME\}\}\n?", "", content)
-        content = re.sub(r"\{\{END_IF_RESUME\}\}\n?", "", content)
+        # Remove fresh-start blocks, keep resume blocks
         content = re.sub(
             r"\{\{IF_FRESH_START\}\}.*?\{\{END_IF_FRESH_START\}\}\n?",
             "", content, flags=re.DOTALL,
         )
+        content = re.sub(r"\{\{IF_RESUME\}\}\n?", "", content)
+        content = re.sub(r"\{\{END_IF_RESUME\}\}\n?", "", content)
     else:
+        # Keep fresh-start blocks, remove resume blocks
         content = re.sub(r"\{\{IF_FRESH_START\}\}\n?", "", content)
         content = re.sub(r"\{\{END_IF_FRESH_START\}\}\n?", "", content)
         content = re.sub(
@@ -848,7 +846,7 @@ def _tier_reminders(pipeline_mode, critic_enabled=False):
 
 
 def assemble_sections(pipeline_mode, sections_dir, init_done, is_resume,
-                      critic_enabled, browser_enabled):
+                      critic_enabled, browser_enabled, retry_count=0):
     """Assemble prompt sections based on tier and conditions.
 
     Uses Python code for conditional logic instead of regex-based
@@ -933,15 +931,6 @@ def assemble_sections(pipeline_mode, sections_dir, init_done, is_resume,
             sections.append(("phase0-skip",
                               "### Phase 0: SKIP (already initialized)\n"))
 
-    # --- Resume header (if resuming) ---
-    if is_resume:
-        sections.append(("resume-header",
-                          load_section(sections_dir, "resume-header.md")))
-
-    # --- Failure log check ---
-    sections.append(("failure-log-check",
-                      load_section(sections_dir, "failure-log-check.md")))
-
     # --- Context Snapshot + Plan (tier-dependent) ---
     if pipeline_mode == "full":
         # Tier 3: full specify + plan workflow
@@ -995,9 +984,15 @@ def assemble_sections(pipeline_mode, sections_dir, init_done, is_resume,
                           load_section(sections_dir,
                                        "phase-implement-agent.md")))
 
-    # --- Test Failure Recovery Protocol (all tiers) ---
-    sections.append(("test-failure-recovery",
-                      load_section(sections_dir, "test-failure-recovery.md")))
+    # --- Test Failure Recovery Protocol (tier-specific) ---
+    if pipeline_mode == "lite":
+        sections.append(("test-failure-recovery",
+                          load_section(sections_dir,
+                                       "test-failure-recovery-lite.md")))
+    else:
+        sections.append(("test-failure-recovery",
+                          load_section(sections_dir,
+                                       "test-failure-recovery-agent.md")))
 
     # --- Critic: Code Challenge (only if critic enabled, agent tiers) ---
     if critic_enabled and pipeline_mode in ("standard", "full"):
@@ -1213,8 +1208,6 @@ def build_replacements(args, feature, features, global_context, script_dir):
     # Make it absolute from project root
     session_status_abs = os.path.join(project_root, session_status_path)
 
-    prev_status = get_prev_session_status(args.state_dir, args.feature_id)
-
     # Compute feature slug for per-feature directory naming
     feature_slug = compute_feature_slug(
         args.feature_id, feature.get("title", "")
@@ -1223,7 +1216,10 @@ def build_replacements(args, feature, features, global_context, script_dir):
     # Detect project state
     init_done = detect_init_status(project_root)
     artifacts = detect_existing_artifacts(project_root, feature_slug)
-    complexity = feature.get("estimated_complexity", "medium")
+    complexity = feature.get(
+        "estimated_complexity",
+        feature.get("complexity", "medium"),
+    )
     if args.mode:
         pipeline_mode = args.mode
     else:
@@ -1273,8 +1269,6 @@ def build_replacements(args, feature, features, global_context, script_dir):
     # Browser interaction - extract from feature if present and playwright-cli available
     browser_interaction = feature.get("browser_interaction")
     browser_enabled = False
-    browser_url = ""
-    browser_setup_command = ""
     browser_verify_steps = ""
 
     browser_verify_env = os.environ.get("BROWSER_VERIFY", "").lower()
@@ -1282,18 +1276,27 @@ def build_replacements(args, feature, features, global_context, script_dir):
         browser_interaction = None
 
     if browser_interaction and isinstance(browser_interaction, dict):
-        browser_url = browser_interaction.get("url", "")
-        if browser_url:
+        # browser_interaction only needs verify_steps — AI auto-detects
+        # dev server command, URL, and port from project config
+        steps = browser_interaction.get("verify_steps", [])
+        if isinstance(browser_interaction, bool) and browser_interaction:
+            # Simple boolean: browser verification enabled, no specific goals
             browser_enabled = True
-            browser_setup_command = browser_interaction.get("setup_command", "# no setup needed")
-            steps = browser_interaction.get("verify_steps", [])
-            if steps:
-                browser_verify_steps = "\n".join(
-                    "   # Goal {}: {}".format(i + 1, step)
-                    for i, step in enumerate(steps)
-                )
-            else:
-                browser_verify_steps = "   # (no specific verify goals — just open and screenshot)"
+            browser_verify_steps = (
+                "   # (no specific verify goals — explore the app and "
+                "verify the feature works as expected)")
+        elif steps:
+            browser_enabled = True
+            browser_verify_steps = "\n".join(
+                "   # Goal {}: {}".format(i + 1, step)
+                for i, step in enumerate(steps)
+            )
+        elif browser_interaction.get("url") or browser_interaction.get("enabled", True):
+            # Backward compat: old format had url/setup_command fields
+            browser_enabled = True
+            browser_verify_steps = (
+                "   # (no specific verify goals — explore the app and "
+                "verify the feature works as expected)")
 
     # Auto-detect test commands from project structure
     test_cmd = detect_test_commands(project_root)
@@ -1318,10 +1321,6 @@ def build_replacements(args, feature, features, global_context, script_dir):
         "{{FEATURE_ID}}": args.feature_id,
         "{{FEATURE_LIST_PATH}}": os.path.abspath(args.feature_list),
         "{{FEATURE_TITLE}}": feature.get("title", ""),
-        "{{RETRY_COUNT}}": str(args.retry_count),
-        "{{MAX_RETRIES}}": str(DEFAULT_MAX_RETRIES),
-        "{{PREV_SESSION_STATUS}}": prev_status,
-        "{{RESUME_PHASE}}": args.resume_phase,
         "{{FEATURE_DESCRIPTION}}": feature.get("description", ""),
         "{{ACCEPTANCE_CRITERIA}}": format_acceptance_criteria(
             feature.get("acceptance_criteria", [])
@@ -1331,6 +1330,7 @@ def build_replacements(args, feature, features, global_context, script_dir):
         ),
         "{{GLOBAL_CONTEXT}}": format_global_context(global_context, project_root),
         "{{PROJECT_BRIEF}}": _read_project_brief(project_root),
+        "{{PLATFORM_CONVENTIONS}}": read_platform_conventions(project_root),
         "{{TEAM_CONFIG_PATH}}": team_config_path,
         "{{DEV_SUBAGENT_PATH}}": dev_subagent,
         "{{REVIEWER_SUBAGENT_PATH}}": reviewer_subagent,
@@ -1349,8 +1349,6 @@ def build_replacements(args, feature, features, global_context, script_dir):
         "{{INIT_DONE}}": "true" if init_done else "false",
         "{{HAS_SPEC}}": "true" if artifacts["has_spec"] else "false",
         "{{HAS_PLAN}}": "true" if artifacts["has_plan"] else "false",
-        "{{BROWSER_URL}}": browser_url,
-        "{{BROWSER_SETUP_COMMAND}}": browser_setup_command,
         "{{BROWSER_VERIFY_STEPS}}": browser_verify_steps,
         "{{AC_CHECKLIST}}": format_ac_checklist(
             feature.get("acceptance_criteria", [])
@@ -1438,7 +1436,6 @@ def main():
     replacements, effective_resume, browser_enabled = build_replacements(
         args, feature, features, global_context, script_dir
     )
-    replacements["{{RESUME_PHASE}}"] = effective_resume
 
     # Load agent prompt templates and merge into replacements
     agent_prompt_replacements = load_agent_prompts(templates_dir)
@@ -1460,6 +1457,7 @@ def main():
             sections = assemble_sections(
                 pipeline_mode, sections_dir, init_done, is_resume,
                 critic_enabled, browser_enabled,
+                retry_count=int(args.retry_count),
             )
             rendered = render_from_sections(sections, replacements)
         except FileNotFoundError as exc:
@@ -1474,7 +1472,10 @@ def main():
         if args.template:
             template_path = args.template
         else:
-            complexity = feature.get("estimated_complexity", "medium")
+            complexity = feature.get(
+                "estimated_complexity",
+                feature.get("complexity", "medium"),
+            )
             _mode = args.mode or determine_pipeline_mode(complexity)
             _tier_file_map = {
                 "lite": "bootstrap-tier1.md",

package/bundled/dev-pipeline/scripts/generate-bugfix-prompt.py

@@ -19,7 +19,7 @@ import os
 import re
 import sys
 
-from utils import enrich_global_context, load_json_file, setup_logging
+from utils import enrich_global_context, load_json_file, read_platform_conventions, setup_logging
 
 
 DEFAULT_MAX_RETRIES = 3
@@ -223,8 +223,6 @@ def build_replacements(args, bug, global_context, script_dir):
         "sessions", args.session_id, "session-status.json"
     )
 
-    prev_status = get_prev_session_status(args.state_dir, args.bug_id)
-
     # Error source
     error_source = bug.get("error_source", {})
     error_type = error_source.get("type", "unknown") if isinstance(error_source, dict) else "unknown"
@@ -246,10 +244,6 @@ def build_replacements(args, bug, global_context, script_dir):
         "{{BUG_TITLE}}": bug.get("title", ""),
         "{{SEVERITY}}": bug.get("severity", "medium"),
         "{{VERIFICATION_TYPE}}": vtype,
-        "{{RETRY_COUNT}}": str(args.retry_count),
-        "{{MAX_RETRIES}}": str(DEFAULT_MAX_RETRIES),
-        "{{PREV_SESSION_STATUS}}": prev_status,
-        "{{RESUME_PHASE}}": args.resume_phase,
         "{{BUG_DESCRIPTION}}": bug.get("description", ""),
         "{{ERROR_SOURCE_TYPE}}": error_type,
         "{{ERROR_SOURCE_DETAILS}}": format_error_source_details(error_source),
@@ -258,6 +252,7 @@ def build_replacements(args, bug, global_context, script_dir):
         ),
         "{{ENVIRONMENT}}": format_environment(bug.get("environment")),
         "{{GLOBAL_CONTEXT}}": format_global_context(global_context, project_root),
+        "{{PLATFORM_CONVENTIONS}}": read_platform_conventions(project_root),
         "{{TEAM_CONFIG_PATH}}": team_config_path,
         "{{DEV_SUBAGENT_PATH}}": dev_subagent,
         "{{REVIEWER_SUBAGENT_PATH}}": reviewer_subagent,
@@ -331,13 +326,13 @@ def emit_failure(message):
 
 BUGFIX_STEPS = [
     ("prizmkit-init", "Initialize", []),
-    ("bug-diagnosis", "Bug Diagnosis & Fix Plan",
-     [".prizmkit/bugfix/{slug}/fix-plan.md"]),
-    ("bug-reproduce", "Write Reproduction Test", []),
-    ("bug-fix", "Implement Fix", []),
+    ("bug-diagnosis-and-plan", "Diagnose & Plan",
+     [".prizmkit/bugfix/{slug}/spec.md",
+      ".prizmkit/bugfix/{slug}/plan.md"]),
+    ("prizmkit-implement", "Implement Fix", []),
     ("prizmkit-code-review", "Code Review", []),
     ("prizmkit-committer", "Commit", []),
-    ("bug-report", "Generate Fix Report & Update TRAPS",
+    ("bug-report", "Generate Fix Report",
      [".prizmkit/bugfix/{slug}/fix-report.md"]),
 ]
 

package/bundled/dev-pipeline/scripts/generate-refactor-prompt.py

@@ -19,7 +19,7 @@ import os
 import re
 import sys
 
-from utils import enrich_global_context, load_json_file, setup_logging
+from utils import enrich_global_context, load_json_file, read_platform_conventions, setup_logging
 
 
 DEFAULT_MAX_RETRIES = 3
@@ -27,6 +27,97 @@ DEFAULT_MAX_RETRIES = 3
 LOGGER = setup_logging("generate-refactor-prompt")
 
 
+# Refactor pipeline checkpoint steps (skill_key, display_name, required_artifacts)
+# Artifacts use {slug} placeholder, replaced with refactor_id at runtime.
+REFACTOR_STEPS = [
+    ("prizmkit-init", "Initialize",
+     [".prizmkit/refactor/{slug}"]),
+    ("prizmkit-plan", "Plan — Specification & Plan Generation",
+     [".prizmkit/refactor/{slug}/spec.md",
+      ".prizmkit/refactor/{slug}/plan.md"]),
+    ("prizmkit-implement", "Implement — Behavior-Preserving Refactoring",
+     [".prizmkit/refactor/{slug}/plan.md"]),
+    ("prizmkit-code-review", "Review — Code Review & Behavior Verification",
+     [".prizmkit/refactor/{slug}/review-report.md"]),
+    ("prizmkit-committer", "Commit",
+     []),
+    ("refactor-report", "Generate Refactor Report",
+     [".prizmkit/refactor/{slug}/refactor-report.md"]),
+]
+
+
+def generate_refactor_checkpoint(refactor_id, session_id):
+    """Generate a checkpoint definition for refactor pipeline.
+
+    Returns a dict suitable for writing as workflow-checkpoint.json.
+    """
+    steps = []
+    prev_id = None
+    for i, (skill, name, artifacts) in enumerate(REFACTOR_STEPS, 1):
+        step_id = "S{:02d}".format(i)
+        steps.append({
+            "id": step_id,
+            "skill": skill,
+            "name": name,
+            "status": "pending",
+            "required_artifacts": [a.replace("{slug}", refactor_id) for a in artifacts],
+            "depends_on": prev_id,
+        })
+        prev_id = step_id
+
+    return {
+        "version": 1,
+        "workflow_type": "refactor-pipeline",
+        "pipeline_mode": "standard",
+        "item_id": refactor_id,
+        "item_slug": refactor_id,
+        "session_id": session_id,
+        "steps": steps,
+    }
+
+
+def merge_refactor_checkpoint_state(existing, fresh, project_root):
+    """Merge existing refactor checkpoint state into fresh definition.
+
+    Same logic as feature/bugfix pipelines: validate artifacts, break chain
+    on first invalid step.
+    """
+    existing_status = {}
+    existing_artifacts = {}
+    for step in existing.get("steps", []):
+        existing_status[step["skill"]] = step["status"]
+        existing_artifacts[step["skill"]] = step.get("required_artifacts", [])
+
+    valid_completed = set()
+    for skill_key, status in existing_status.items():
+        if status == "completed":
+            artifacts = existing_artifacts.get(skill_key, [])
+            if all(os.path.exists(os.path.join(project_root, a))
+                   for a in artifacts):
+                valid_completed.add(skill_key)
+            else:
+                LOGGER.warning(
+                    "Step '%s' was completed but artifacts missing — "
+                    "resetting to pending", skill_key
+                )
+        elif status == "skipped":
+            valid_completed.add(skill_key)
+
+    chain_broken = False
+    for step in fresh["steps"]:
+        if chain_broken:
+            step["status"] = "pending"
+            continue
+        prev = existing_status.get(step["skill"])
+        if step["skill"] in valid_completed:
+            step["status"] = prev
+        else:
+            chain_broken = True
+            step["status"] = "pending"
+
+    return fresh
+
+
 def parse_args():
     parser = argparse.ArgumentParser(
         description=(
@@ -281,8 +372,6 @@ def build_replacements(args, refactor, refactors, global_context, script_dir):
         "sessions", args.session_id, "session-status.json"
     )
 
-    prev_status = get_prev_session_status(args.state_dir, args.refactor_id)
-
     # Scope
     scope = refactor.get("scope", {})
 
@@ -321,10 +410,6 @@ def build_replacements(args, refactor, refactors, global_context, script_dir):
         "{{NEW_TESTS_NEEDED}}": new_tests_str,
         "{{PRIORITY}}": refactor.get("priority", "medium"),
         "{{COMPLEXITY}}": refactor.get("complexity", "medium"),
-        "{{RETRY_COUNT}}": str(args.retry_count),
-        "{{MAX_RETRIES}}": str(DEFAULT_MAX_RETRIES),
-        "{{PREV_SESSION_STATUS}}": prev_status,
-        "{{RESUME_PHASE}}": args.resume_phase,
         "{{REFACTOR_DESCRIPTION}}": refactor.get("description", ""),
         "{{ACCEPTANCE_CRITERIA}}": format_acceptance_criteria(
             refactor.get("acceptance_criteria", [])
@@ -342,6 +427,7 @@ def build_replacements(args, refactor, refactors, global_context, script_dir):
             ".prizmkit", "refactor", args.refactor_id, "workflow-checkpoint.json",
         ),
         "{{TIMESTAMP}}": "",  # Placeholder — agent fills in timestamp
+        "{{PLATFORM_CONVENTIONS}}": read_platform_conventions(project_root),
     }
 
     return replacements
@@ -350,33 +436,20 @@ def build_replacements(args, refactor, refactors, global_context, script_dir):
 def process_conditional_blocks(content, resume_phase):
     """Handle conditional blocks based on resume_phase.
 
-    - {{IF_RESUME}}...{{END_IF_RESUME}} — include only when resuming (resume_phase != "null")
     - {{IF_FRESH_START}}...{{END_IF_FRESH_START}} — include only on fresh start (resume_phase == "null")
     """
     is_resume = resume_phase != "null"
 
     if is_resume:
-        # Keep IF_RESUME content, strip markers
-        content = content.replace("{{IF_RESUME}}\n", "")
-        content = content.replace("{{IF_RESUME}}", "")
-        content = content.replace("{{END_IF_RESUME}}\n", "")
-        content = content.replace("{{END_IF_RESUME}}", "")
-        # Remove IF_FRESH_START blocks entirely
         content = re.sub(
             r"\{\{IF_FRESH_START\}\}.*?\{\{END_IF_FRESH_START\}\}\n?",
             "", content, flags=re.DOTALL,
         )
     else:
-        # Keep IF_FRESH_START content, strip markers
         content = content.replace("{{IF_FRESH_START}}\n", "")
         content = content.replace("{{IF_FRESH_START}}", "")
         content = content.replace("{{END_IF_FRESH_START}}\n", "")
         content = content.replace("{{END_IF_FRESH_START}}", "")
-        # Remove IF_RESUME blocks entirely
-        content = re.sub(
-            r"\{\{IF_RESUME\}\}.*?\{\{END_IF_RESUME\}\}\n?",
-            "", content, flags=re.DOTALL,
-        )
 
     return content
 
@@ -470,6 +543,37 @@ def main():
     if err:
         emit_failure(err)
 
+    # Generate checkpoint file
+    project_root = resolve_project_root(script_dir)
+    checkpoint_rel = os.path.join(
+        ".prizmkit", "refactor", args.refactor_id, "workflow-checkpoint.json",
+    )
+    checkpoint_path = os.path.join(project_root, checkpoint_rel)
+    checkpoint_dir = os.path.dirname(checkpoint_path)
+    os.makedirs(checkpoint_dir, exist_ok=True)
+
+    checkpoint = generate_refactor_checkpoint(args.refactor_id, args.session_id)
+
+    is_resume = args.resume_phase != "null"
+    if is_resume and os.path.exists(checkpoint_path):
+        try:
+            with open(checkpoint_path, "r", encoding="utf-8") as f:
+                existing = json.load(f)
+            checkpoint = merge_refactor_checkpoint_state(
+                existing, checkpoint, project_root,
+            )
+            LOGGER.info("Merged existing refactor checkpoint from %s",
+                        checkpoint_path)
+        except (json.JSONDecodeError, KeyError) as exc:
+            LOGGER.warning(
+                "Existing refactor checkpoint corrupted (%s) — generating fresh",
+                exc,
+            )
+
+    with open(checkpoint_path, "w", encoding="utf-8") as f:
+        json.dump(checkpoint, f, indent=2, ensure_ascii=False)
+    LOGGER.info("Wrote refactor checkpoint to %s", checkpoint_path)
+
     # Resolve critic and mode
     refactor_critic = refactor.get("critic", False)
     if args.critic is not None:

package/bundled/dev-pipeline/scripts/utils.py

@@ -518,3 +518,23 @@ def enrich_global_context(global_context, project_root):
             global_context[ctx_key] = detected[det_key] + " (auto-detected)"
 
     return global_context
+
+
+def read_platform_conventions(project_root):
+    """Resolve the path to CLAUDE.md or CODEBUDDY.md for project-level conventions.
+
+    Returns a path reference for the AI agent to read at runtime,
+    rather than inlining the full file content into the prompt.
+    """
+    platform = os.environ.get("PRIZMKIT_PLATFORM", "claude")
+    if platform == "codebuddy":
+        candidates = ["CODEBUDDY.md", "CLAUDE.md"]
+    else:
+        candidates = ["CLAUDE.md", "CODEBUDDY.md"]
+
+    for filename in candidates:
+        filepath = os.path.join(project_root, filename)
+        if os.path.isfile(filepath):
+            return "`{}`".format(filename)
+
+    return "(No project conventions file found — CLAUDE.md or CODEBUDDY.md)"

package/bundled/dev-pipeline/templates/agent-prompts/dev-implement.md

@@ -14,14 +14,20 @@ Update the AC Verification Checklist in context-snapshot.md by marking each item
 - All [ ] must become [x] — if any AC remains unverified, the feature is incomplete
 - Document any AC that cannot be verified due to test failures
 
-## Test Failure Recovery (Max 2 Fix Rounds)
+## Test Failure Recovery (Convergence-Based)
 
-If tests fail during implementation:
-- **Round 1**: Analyze failure (code bug vs. brittle test vs. environment), apply fix, re-run tests
-- **Round 2 (if still failing)**: Try alternate fix approach, re-run tests
-- **After Round 2**: Document failure in Implementation Log with root cause, category, and fix attempted
-- **Do NOT block completion** if unable to resolve — only NEW REGRESSIONS (not in baseline) require fixing
-- **If any AC cannot be verified** due to test failure: the feature is incomplete, add to failure notes
+If tests fail during implementation, use convergence-based recovery — keep fixing as long as progress is being made:
+
+1. **Run tests, record results**: count failures, note which tests failed (exclude baseline failures)
+2. **Check termination**:
+   - All tests pass → Done
+   - Plateau: same failures for 3 consecutive rounds → Cannot resolve, document and stop
+   - Failures decreased → Continue fixing
+3. **Fix and iterate**: analyze, apply fix, re-run `$TEST_CMD`, go back to step 1
+
+**Key rule**: If failures decrease (even by 1), the plateau counter resets to 0.
+**Do NOT block completion** if unable to resolve — only NEW REGRESSIONS (not in baseline) require fixing.
+**If any AC cannot be verified** due to test failure: the feature is incomplete, add to failure notes.
 
 4. Do NOT execute any git commands (no git add/commit/reset/push).
 Do NOT exit until all tasks are [x], the '## Implementation Log' section is written, and AC Verification Checklist is 100% complete in context-snapshot.md."