opencode-team-lead 0.3.0 → 0.3.1

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/README.md CHANGED
@@ -59,6 +59,7 @@ The agent has a minimal permission set:
59
59
  | `memoai_*` | allow |
60
60
  | `sequential-thinking_*` | allow |
61
61
  | `bash` (git only) | allow |
62
+ | `read` / `edit` (`.opencode/scratchpad.md` only) | allow |
62
63
  | Everything else | deny |
63
64
 
64
65
  ## Customization
package/index.js CHANGED
@@ -41,6 +41,14 @@ export const TeamLeadPlugin = async ({ directory, worktree }) => {
41
41
  distill: "allow",
42
42
  prune: "allow",
43
43
  compress: "allow",
44
+ read: {
45
+ "*": "deny",
46
+ ".opencode/scratchpad.md": "allow",
47
+ },
48
+ edit: {
49
+ "*": "deny",
50
+ ".opencode/scratchpad.md": "allow",
51
+ },
44
52
  "memoai_*": "allow",
45
53
  "sequential-thinking_*": "allow",
46
54
  bash: {
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "opencode-team-lead",
3
- "version": "0.3.0",
3
+ "version": "0.3.1",
4
4
  "description": "Team-lead orchestrator agent for opencode — delegates work, reviews quality, manages context",
5
5
  "type": "module",
6
6
  "main": "index.js",
package/prompt.md CHANGED
@@ -51,6 +51,7 @@ If you catch yourself about to use `read`, `edit`, `bash`, `glob`, `grep`, or `w
51
51
  - Specify what the agent should RETURN so you can synthesize results
52
52
  - **Parallelize independent tasks** — launch multiple agents simultaneously when possible
53
53
  - Never assume an agent knows project context — be explicit
54
+ - **Update the scratchpad** after each delegation — add agent result summaries to the Agent Results section
54
55
 
55
56
  ### 4. Review
56
57
  - **Every code, architecture, infra, or security change MUST be reviewed before reporting success**
@@ -61,12 +62,14 @@ If you catch yourself about to use `read`, `edit`, `bash`, `glob`, `grep`, or `w
61
62
  - If the reviewer returns **BLOCKED**: escalate immediately to the user with the reviewer's reasoning
62
63
  - **Maximum 2 review rounds** — if still not approved after 2 iterations, escalate to the user
63
64
  - Parallelize reviews when possible (e.g., code review + security review simultaneously)
65
+ - **Update the scratchpad** after each review — update task statuses and record review outcomes
64
66
 
65
67
  ### 5. Synthesize & Report
66
68
  - **Self-evaluate first** — before reporting anything, run through the Self-Evaluation checklist below. If something doesn't pass, loop back to the appropriate phase.
67
69
  - Collect outputs from all agents
68
70
  - Summarize results concisely for the user
69
71
  - Flag any issues, conflicts, or failures
72
+ - **Update the scratchpad** — final state capture before reporting to the user
70
73
  - Propose next steps if applicable
71
74
  - **Record learnings in `memoai_memo_record`** — don't just offer, do it systematically (see Memory Protocol below)
72
75