ma-agents 2.19.0 → 2.19.2

package/lib/installer.js

@@ -144,6 +144,13 @@ Do not load skills that are not relevant to the current task.
       } else {
         content += wrappedInstruction;
       }
+    } else if (agent.category === 'bmad') {
+      // BMAD agent instruction files ARE the agent definitions (persona, menu, activation).
+      // Never create them from scratch — they must be deployed by applyCustomizations().
+      // Creating them here would produce a file with ONLY the MA-AGENTS block,
+      // wiping the entire agent definition.
+      console.log(chalk.gray(`    Skipped ${fileName} (BMAD agent file not yet deployed)`));
+      continue;
     } else {
       content = wrappedInstruction;
     }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "ma-agents",
-  "version": "2.19.0",
+  "version": "2.19.2",
   "description": "NPX tool to install skills for AI coding agents (Claude Code, Gemini, Copilot, Kilocode, Cline, Cursor)",
   "main": "index.js",
   "bin": {

package/skills/ai-audit-trail/SKILL.md

@@ -1,148 +1,19 @@
 # AI Audit Trail
 
-Track AI agent session activity in a project-level `AiAudit.md` file. This provides accountability and visibility into AI agent work — what was done, when, and the resources consumed.
+At the end of every significant agent session, append an entry to `{project_root}/AiAudit.md`. If the file does not exist, create it with a `# AI Audit Trail` header first.
 
-## Policy
-
-### 1. Always Maintain the Audit Log
-
-At the **end of every significant agent session** (document generation, multi-step workflow, code changes), the agent MUST append an entry to `AiAudit.md` in the project root.
-
-A "significant session" is any task that:
-- Generates or substantially modifies a document
-- Executes a multi-step workflow
-- Makes changes across multiple files
-- Takes more than a trivial amount of interaction
-
-Single-question answers or trivial lookups do NOT require an audit entry.
-
-### 2. File Location
-
-The audit log is always at:
-```
-{project_root}/AiAudit.md
-```
-
-If the file does not exist, create it with the header (see format below).
-
-### 3. File Format
+## Entry Format
 
 ```markdown
-# AI Audit Trail
-
-This file tracks AI agent sessions, time spent, and resources consumed.
-
----
-
-## Session: {session_title}
-
-| Field | Value |
-|-------|-------|
-| **Date** | {YYYY-MM-DD} |
-| **Agent** | {agent_name} ({model_name}) |
-| **Task** | {brief_task_description} |
-| **Duration** | {estimated_duration} |
-| **Tokens (est.)** | Input: ~{input_tokens} / Output: ~{output_tokens} |
-| **Files Modified** | {count} files |
-| **Documents Generated** | {list_or_count} |
-| **Status** | {Completed / Partial / Failed} |
-
-### Summary
-{2-3 sentence summary of what was accomplished}
-
-### Files Touched
-- `path/to/file1.md` — {action: created/updated/deleted}
-- `path/to/file2.md` — {action}
-
----
+| Date | Task | Tokens (est.) |
+|------|------|---------------|
+| YYYY-MM-DD | Brief task description | In: ~X / Out: ~Y |
 ```
 
-### 4. Field Definitions
-
-| Field | Description | How to Determine |
-|-------|-------------|-----------------|
-| **Date** | Session date | Current date in `YYYY-MM-DD` format |
-| **Agent** | Agent identity | The agent name and model (e.g., `Joseph (Claude Opus 4.6)`, `Claude Code (Claude Sonnet 4.6)`) |
-| **Task** | What was requested | Brief description of the user's request |
-| **Duration** | Estimated wall-clock time | Estimate from session start to end. Use ranges if uncertain (e.g., `~15-20 minutes`) |
-| **Tokens (est.)** | Estimated token consumption | Provide best estimate of input/output tokens. Use `~` prefix to indicate approximation. If unknown, write `Not available` |
-| **Files Modified** | Number of files changed | Count of files created, updated, or deleted |
-| **Documents Generated** | Documents produced | List document names or count |
-| **Status** | Outcome | `Completed` if all tasks done, `Partial` if some remain, `Failed` if blocked |
-
-### 5. Rules
-
-1. **Append only** — Never delete or modify previous session entries. The audit log is an append-only record.
-2. **One entry per session** — Do not create multiple entries for the same continuous work session. If a session spans multiple tasks, combine them into one entry.
-3. **Estimate honestly** — Token estimates are approximations. Round to the nearest thousand. If you cannot estimate, say so explicitly.
-4. **Session title** — Use a descriptive, unique title (e.g., `Generate SSDD v1.0`, `Fix SRS interface requirements`, `MIL-498 workflow audit`).
-5. **End of session** — Write the audit entry as the **last action** before presenting final results to the user.
-6. **Chronological order** — New entries go at the **bottom** of the file, maintaining chronological order.
+## Rules
 
-### 6. Duration Estimation
-
-Since AI agents don't have built-in timers, use these guidelines:
-- Note the time when you start the task (from system context or user message timestamps)
-- Note the time when you finish
-- If timestamps aren't available, estimate based on the complexity: quick task (~5 min), medium task (~15-30 min), large task (~1-2 hours)
-- Always prefix with `~` to indicate it's an estimate
-
-### 7. Token Estimation
-
-Provide your best estimate:
-- **Input tokens**: Roughly proportional to the amount of context you consumed (files read, conversation history)
-- **Output tokens**: Roughly proportional to the amount of text you generated (documents, code, messages)
-- If precise counts are available from the environment, use them
-- If not, estimate and note `(estimated)` next to the values
-
-## Example
-
-```markdown
-# AI Audit Trail
-
-This file tracks AI agent sessions, time spent, and resources consumed.
-
----
-
-## Session: Generate SSDD v1.0
-
-| Field | Value |
-|-------|-------|
-| **Date** | 2026-03-01 |
-| **Agent** | Joseph (Claude Opus 4.6) |
-| **Task** | Generate MIL-STD-498 SSDD document from project artifacts |
-| **Duration** | ~25 minutes |
-| **Tokens (est.)** | Input: ~45,000 / Output: ~12,000 |
-| **Files Modified** | 1 file |
-| **Documents Generated** | SSDD.md |
-| **Status** | Completed |
-
-### Summary
-Generated the System/Subsystem Design Description from PRD, architecture, and product brief artifacts. Conducted CSCI discovery interview identifying 5 CSCIs. All 6 DID sections populated with full requirements traceability.
-
-### Files Touched
-- `_bmad-output/planning-artifacts/SSDD.md` — created
-
----
-
-## Session: Update SRS per user feedback
-
-| Field | Value |
-|-------|-------|
-| **Date** | 2026-03-02 |
-| **Agent** | Joseph (Claude Opus 4.6) |
-| **Task** | Update SRS interface requirements based on user review feedback |
-| **Duration** | ~10 minutes |
-| **Tokens (est.)** | Input: ~20,000 / Output: ~5,000 |
-| **Files Modified** | 1 file |
-| **Documents Generated** | SRS.md (updated) |
-| **Status** | Completed |
-
-### Summary
-Updated Section 3.3 interface requirements to add 4 new external interfaces identified during user review. Updated traceability matrix in Section 5.
-
-### Files Touched
-- `_bmad-output/planning-artifacts/SRS.md` — updated
-
----
-```
+- **Append only** — never delete or modify previous entries
+- **One row per session** — combine multiple tasks in the same session into one entry
+- **Estimate tokens** — round to nearest thousand, prefix with `~`. Write `N/A` if unknown
+- Write the entry as the **last action** before presenting results to the user
+- Skip trivial interactions (single-question answers, quick lookups)