opencode-discipline 0.1.3 → 0.1.5

package/agents/README.md

@@ -1,34 +1,43 @@
 # opencode-discipline Agent Suite
 
-This package ships seven agent definitions for disciplined plan-first execution.
+This package ships eight agent definitions for disciplined plan-first execution.
 
 | Agent | Model | Mode | Purpose |
 | --- | --- | --- | --- |
 | `plan` | `anthropic/claude-opus-4-6` | primary | Runs the 4-wave planning workflow and coordinates accept/revise handoff |
 | `build` | `anthropic/claude-opus-4-6` | primary | Executes approved plans phase-by-phase with verification gates |
 | `oracle` | `openai/gpt-5.4` | subagent | Read-only architecture and tradeoff advisor |
-| `librarian` | `openai/gpt-4.1-mini` | subagent | Read-only research and documentation specialist |
+| `librarian` | `openai/gpt-5.4-mini` | subagent | Read-only research and documentation specialist |
 | `reviewer` | `openai/gpt-5.4` | subagent | Read-only quality critic for plans and code |
 | `designer` | `anthropic/claude-sonnet-4-6` | subagent | Read-only UI/UX and accessibility advisor |
 | `deep` | `openai/gpt-5.3-codex` | subagent | Advanced implementation subagent for complex coding work |
+| `explore` | `openai/gpt-5.4-mini` | subagent | Fast codebase search and file discovery |
 
 ## Installation
 
-Manual install:
+Enable the plugin in `opencode.json`:
 
-```bash
-mkdir -p ~/.config/opencode/agents
-cp agents/*.md ~/.config/opencode/agents/
+```json
+{
+  "plugin": ["opencode-discipline"]
+}
 ```
 
-Then enable the plugin in `opencode.json`:
+The plugin automatically injects all agent configurations via the `config` hook — no manual file copying required.
+
+To override a specific field for any agent, add it to `opencode.json`:
 
 ```json
 {
-  "plugin": ["opencode-discipline"]
+  "plugin": ["opencode-discipline"],
+  "agent": {
+    "plan": { "model": "anthropic/claude-sonnet-4-6" }
+  }
 }
 ```
 
+User overrides take precedence per field; non-overridden fields use the plugin defaults.
+
 ## Delegation flow
 
 ```text

package/agents/deep.md

@@ -16,6 +16,7 @@ permission:
   bash:
     "*": ask
     "rtk *": allow
+    "echo *": allow
     "git log *": allow
     "git diff *": allow
     "git status": allow

package/agents/designer.md

@@ -17,6 +17,7 @@ permission:
   bash:
     "*": ask
     "rtk *": allow
+    "echo *": allow
   edit: deny
   write: deny
 ---

package/agents/explore.md

@@ -1,6 +1,6 @@
 ---
 description: Fast codebase search. Finds files, patterns, and structure quickly. Cheap and parallel-friendly.
-model: openai/gpt-4.1-mini
+model: openai/gpt-5.4-mini
 temperature: 0
 mode: subagent
 color: "#78909C"
@@ -24,6 +24,7 @@ permission:
     "wc *": allow
     "git log *": allow
     "git show *": allow
+    "echo *": allow
 ---
 
 You are a fast codebase explorer. Your job is to find files, patterns, and structure as quickly as possible. You never implement or modify anything.

package/agents/librarian.md

@@ -1,6 +1,6 @@
 ---
 description: Research specialist. Gathers context from codebase, external docs, and reference implementations before work begins. Returns structured findings.
-model: openai/gpt-4.1-mini
+model: openai/gpt-5.4-mini
 temperature: 0
 mode: subagent
 color: "#1D9E75"
@@ -17,6 +17,7 @@ permission:
   bash:
     "*": ask
     "rtk *": allow
+    "echo *": allow
 ---
 
 You are a research specialist. Your job is to gather, organize, and present information. You never implement.

package/agents/oracle.md

@@ -23,6 +23,7 @@ permission:
     "cargo clippy *": allow
     "npm run type-check *": allow
     "tsc --noEmit *": allow
+    "echo *": allow
 ---
 
 You are a senior staff engineer acting as an architecture consultant. You are the "last resort" advisor — consulted when facing unfamiliar patterns, tricky tradeoffs, or repeated failures.

package/agents/plan.md

@@ -37,6 +37,7 @@ permission:
     "wc *": allow
     "mkdir *": allow
     "ls *": allow
+    "echo *": allow
   task:
     "*": deny
     "explore": allow
@@ -73,10 +74,10 @@ Before generating the plan, perform a Metis-style gap check. Ask yourself:
 4. **Ambiguity** — Are there terms or requirements that could be interpreted two ways?
 5. **Breaking changes** — Will this plan touch shared interfaces, public APIs, or migration paths?
 
-If any gaps are found:
-- Ask the user to clarify (1-2 questions max)
-- Delegate to @oracle if the gap involves an architectural tradeoff
-- Document the gap and resolution in the plan under "## Decisions"
+Wave 2 policy (always required):
+- Ask the user to clarify gaps when needed (1-2 questions max)
+- Delegate to @oracle once in every Wave 2 pass, even for simple tasks
+- Summarize oracle feedback and document the resolution in the plan under "## Decisions"
 
 ### Wave 3: Plan generation
 
@@ -157,14 +158,15 @@ For high-stakes plans (production changes, data migrations, auth refactors):
 After writing and completing Wave 4 review:
 1. Tell the user: "Plan written to `tasks/plans/{filename}.md`"
 2. Summarize in 3-5 sentences
-3. Present exactly three choices and ask the user to pick one:
-   - **Accept plan & start work** — Call `accept_plan(action="accept")` to create a new Build session with the plan registered. The Build agent will start implementing immediately.
-   - **Start work later** — Do nothing. The plan is saved and the user can start a Build session manually whenever they're ready.
-   - **I have modifications to make** — Call `accept_plan(action="revise")` and stay in Wave 4 to incorporate the user's feedback. After revisions, present these three choices again.
+3. The plugin will inject a handoff prompt into your system context with three choices. Follow those instructions exactly — present the choices and wait for the user's response before acting.
+   - Choice 1 label must be: **Accept plan, clear context and build.**
+   - Choice 2 label: **Start work later**
+   - Choice 3 label: **I have modifications to make**
 
 ## Rules
 
 - NEVER write implementation code. Only markdown files in `tasks/`.
+- Immediately after Wave 1 starts, call `todowrite` to create the 5-item planning checklist (Wave 1-4 plus Step 5 handoff) and keep it updated.
 - Every phase MUST have a verify step that is a runnable command.
 - Every phase with destructive changes MUST have a rollback step.
 - Reference specific file paths, function names, and line numbers — never "the auth module."

package/agents/reviewer.md

@@ -23,6 +23,7 @@ permission:
     "cargo clippy *": allow
     "npm test *": allow
     "npm run type-check *": allow
+    "echo *": allow
 ---
 
 You are a ruthless reviewer. You review plans and code changes with zero tolerance for vagueness, missing tests, or scope creep. Your job is to catch what others miss.

package/dist/agents.d.ts

@@ -0,0 +1,9 @@
+/**
+ * Reads all agent `.md` files from the given directory, parses their YAML
+ * frontmatter and markdown body, and returns a map of agent name to config.
+ *
+ * The returned config objects have the same shape as `AgentConfig` in the
+ * OpenCode SDK — frontmatter keys map directly to config fields, and the
+ * markdown body becomes the `prompt` field.
+ */
+export declare function loadAgentConfigs(agentsDir: string): Record<string, Record<string, unknown>>;