cc-dev-template 0.1.95 → 0.1.96

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "cc-dev-template",
3
- "version": "0.1.95",
3
+ "version": "0.1.96",
4
4
  "description": "Structured AI-assisted development framework for Claude Code",
5
5
  "bin": {
6
6
  "cc-dev-template": "./bin/install.js"
@@ -4,23 +4,21 @@ Orchestrate implementation using spec-implementer and spec-validator sub-agents.
4
4
 
5
5
  Read `{spec_dir}/spec.md` and list all task files in `{spec_dir}/tasks/`.
6
6
 
7
- ## Create Tasks
7
+ ## Step 1: Hydrate Task System
8
8
 
9
- Create these tasks and work through them in order:
9
+ **This step is mandatory — do not skip it.** The task system is how you track dependency order, implementation status, and validation status throughout this phase. Without it, you will lose track of what's done and what's blocked.
10
10
 
11
- 1. "Hydrate task system" load task files into Claude Code's task system
12
- 2. "Implement tasks" dispatch implementer agents
13
- 3. "Validate tasks" dispatch validator agents
14
- 4. "Handle failures" — triage and re-attempt if needed
15
- 5. "Finalize" — present results
11
+ Read each task file in `{spec_dir}/tasks/`. For each one, create a Claude Code task (TaskCreate) with:
12
+ - The task title and description from the task file
13
+ - `blockedBy` relationships based on the `depends_on` field in each task file's frontmatter
16
14
 
17
- ## Task 1: Hydrate
15
+ After hydration, use TaskList to verify all tasks are created with correct dependencies before proceeding.
18
16
 
19
- Read each task file in `{spec_dir}/tasks/`. For each one, create a Claude Code task (TaskCreate) with the task title and description. Set up `blockedBy` relationships based on the `depends_on` field in each task file's frontmatter.
17
+ ## Step 2: Implement and Validate
20
18
 
21
- ## Task 2: Implement
19
+ Work through tasks in dependency order. For each task that is ready (no blockers):
22
20
 
23
- For each task that is ready (no blockers), dispatch an implementer:
21
+ **Implement** dispatch an implementer agent:
24
22
 
25
23
  ```
26
24
  Agent tool:
@@ -28,11 +26,9 @@ Agent tool:
28
26
  prompt: "Implement the task described in {task_file_path}. Read the task file for requirements, files to modify, and verification steps. Also read {spec_dir}/spec.md for overall context. After implementation, run the verification steps described in the task file."
29
27
  ```
30
28
 
31
- Run independent tasks in parallel when possible. Mark each Claude Code task as completed when its implementer finishes successfully.
29
+ When the implementer returns, use TaskUpdate to mark the task as `in_progress` (implementation done, not yet validated).
32
30
 
33
- ## Task 3: Validate
34
-
35
- For each completed task, dispatch a validator:
31
+ **Validate** dispatch a validator agent for the same task:
36
32
 
37
33
  ```
38
34
  Agent tool:
@@ -40,15 +36,12 @@ Agent tool:
40
36
  prompt: "Validate the task described in {task_file_path}. Review the code changes, run tests, and verify against the acceptance criteria in {spec_dir}/spec.md. Report pass/fail with details."
41
37
  ```
42
38
 
43
- ## Task 4: Handle Failures
44
-
45
- If any validation fails:
39
+ - **If pass**: Use TaskUpdate to mark the task as `completed`. This unblocks downstream tasks.
40
+ - **If fail**: Re-dispatch the spec-implementer with the validation feedback appended to the prompt, then re-validate. If a task fails validation twice, flag it for the user and ask how to proceed.
46
41
 
47
- - Re-dispatch the spec-implementer with the validation feedback appended to the prompt
48
- - Re-validate after fixes
49
- - If a task fails validation twice, flag it for the user with the error details and ask how to proceed
42
+ Run independent tasks (no dependency between them) in parallel when possible. Always check TaskList before dispatching to confirm which tasks are ready.
50
43
 
51
- ## Task 5: Finalize
44
+ ## Step 3: Finalize
52
45
 
53
46
  Update `{spec_dir}/state.yaml` — set `phase: complete`.
54
47