Rename --tool orchestration flag to --non-interactive

CHANGELOG.md

@@ -8,6 +8,10 @@ Please choose versions by [Semantic Versioning](http://semver.org/).
 * MINOR version when you add functionality in a backwards-compatible manner, and
 * PATCH version when you make backwards-compatible bug fixes.
 
+## Unreleased
+
+- refactor(plugin): rename orchestration-mode flag `--tool` → `--non-interactive` and internal label `MODE=tool` → `MODE=non_interactive` across create-task, create-goal, complete-task, complete-goal, defer-task, defer-goal commands and task-creator, goal-creator, goal-manager-agent agents; the command/agent instructions still accept `--tool` as a deprecated alias for one release. These flags are Claude Code slash-command arguments interpreted by the agent instructions — not vault-cli binary flags.
+
 ## v0.96.0
 
 - feat: Config loading follows XDG Base Directory spec — new FindConfigDir prefers ~/.config/vault-cli/ over legacy ~/.vault-cli/ with no forced migration

agents/goal-creator.md

@@ -29,9 +29,9 @@ You create one goal file in the configured vault. You read the vault config via
 
 ## 1. Parse arguments
 
-Input form: `[goal description] [--tool] [--vault NAME] [--objective NAME]`.
+Input form: `[goal description] [--non-interactive] [--vault NAME] [--objective NAME]`.
 
-- `--tool` → MODE = `tool` (machine-readable JSON output, no AskUserQuestion calls, no audit, hard-fail on collision)
+- `--non-interactive` (deprecated alias: `--tool`) → MODE = `non_interactive` (machine-readable JSON output, no AskUserQuestion calls, no audit, hard-fail on collision)
 - `--vault NAME` → target vault override
 - `--objective NAME` → parent objective name (string match against objectives_dir)
 - Remaining tokens → goal description / title
@@ -52,7 +52,7 @@ Find the vault entry matching the requested vault name (or `default_vault` from
 - `objectives_dir` — parent objective lookup folder (or empty)
 - `themes_dir` — theme folder (or empty)
 
-If the vault is not found, report the error and stop. In MODE=tool, return `{"success": false, "error": "..."}`.
+If the vault is not found, report the error and stop. In MODE=non_interactive, return `{"success": false, "error": "..."}`.
 
 ## 3. Compose the title
 
@@ -68,8 +68,8 @@ Final filename: `{Title}.md` (no Jira prefix; goals are higher-level than tasks)
 If `--objective NAME` was provided:
 
 - Glob `{vault.path}/{objectives_dir}/*.md` and find a name match (case-insensitive, partial OK)
-- If 0 matches → MODE=tool fails with `objective not found`; MODE=interactive AskUserQuestion to pick from list or skip
-- If >1 matches → MODE=tool fails; MODE=interactive AskUserQuestion to pick
+- If 0 matches → MODE=non_interactive fails with `objective not found`; MODE=interactive AskUserQuestion to pick from list or skip
+- If >1 matches → MODE=non_interactive fails; MODE=interactive AskUserQuestion to pick
 - Store as `OBJECTIVE_LINK` (e.g. `[[Objective Name]]`)
 
 If `--objective` not provided, leave `OBJECTIVE_LINK` empty. Do not auto-detect.
@@ -86,14 +86,14 @@ In MODE=interactive, AskUserQuestion for timeline:
 
 If the user picks a duration, compute `start_date = today` and `end_date = today + N weeks`. Store as `TIMELINE = "{start_date} to {end_date}"`.
 
-In MODE=tool, leave timeline empty.
+In MODE=non_interactive, leave timeline empty.
 
 ## 6. Determine category and priority
 
 - **Category**: infer from keywords or description (e.g. `trading`, `health`, `learning`, `personal`); leave empty if unclear
 - **Priority**: default 3; raise to 2 if user description signals importance; 1 reserved for top-priority goals
 
-In MODE=tool, default to category=`""` and priority=3, do not block.
+In MODE=non_interactive, default to category=`""` and priority=3, do not block.
 
 ## 7. Resolve template body
 
@@ -129,7 +129,7 @@ If a template body was loaded in step 7, use it. Otherwise, generate the standar
 
 1. `Tags: [[Goal]]` (plus theme tags if set)
 2. `---` separator
-3. Short summary paragraph (1–2 sentences). **First sentence MUST state the outcome** — what's true when the goal is done, not what work happens. E.g. "Reduce X from Y to Z" not "Build Z by doing W". **Before writing, rephrase any mechanism-phrased input into outcome form** — this rule applies in both interactive and tool mode; do NOT rely on step 12 audit to catch it (audit only runs in interactive mode). Avoid `via X` / `by doing Y` / `through Z` / `Split X / Build Y / Refactor Z / Set up Z` openings — these are mechanism leaks. Mechanism details belong in `# Impact` as an "Approach" lead paragraph. See `docs/goal-writing.md#summary-first-sentence` for the sniff test.
+3. Short summary paragraph (1–2 sentences). **First sentence MUST state the outcome** — what's true when the goal is done, not what work happens. E.g. "Reduce X from Y to Z" not "Build Z by doing W". **Before writing, rephrase any mechanism-phrased input into outcome form** — this rule applies in both interactive and non_interactive mode; do NOT rely on step 12 audit to catch it (audit only runs in interactive mode). Avoid `via X` / `by doing Y` / `through Z` / `Split X / Build Y / Refactor Z / Set up Z` openings — these are mechanism leaks. Mechanism details belong in `# Impact` as an "Approach" lead paragraph. See `docs/goal-writing.md#summary-first-sentence` for the sniff test.
 4. `# Impact` — why this goal matters strategically. If a brief mechanism description is needed, lead with an "**Approach**:" paragraph before the strategic bullets.
 5. `# Status Summary` — Progress / Current / Next / Blockers placeholders
 6. `# Success Criteria` — measurable outcomes as checkboxes
@@ -145,7 +145,7 @@ Glob: {vault.path}/{goals_dir}/{filename}
 
 If the file already exists:
 
-- MODE=tool → return `{"success": false, "error": "goal file already exists: ..."}`
+- MODE=non_interactive → return `{"success": false, "error": "goal file already exists: ..."}`
 - MODE=interactive → AskUserQuestion: 1. Pick a different name  2. Cancel
 
 ## 11. Write the file
@@ -170,7 +170,7 @@ Run a light self-audit against the file:
 - If `timeline` is set, validate it is ≤ 4 weeks
 - No accidental empty sections
 
-Skip in MODE=tool.
+Skip in MODE=non_interactive.
 
 ## 13. Return
 
@@ -187,13 +187,13 @@ Next steps:
 3. Verify: /vault-cli:verify-goal "{title}"
 ```
 
-MODE=tool output (single JSON object on stdout, nothing else):
+MODE=non_interactive output (single JSON object on stdout, nothing else):
 
 ```json
 {"success": true, "path": "{absolute_path}", "filename": "{filename}"}
 ```
 
-On error in MODE=tool:
+On error in MODE=non_interactive:
 
 ```json
 {"success": false, "error": "<message>"}
@@ -205,7 +205,7 @@ On error in MODE=tool:
 - Vault not found → fail with the requested vault name in the error
 - `goals_dir` directory does not exist on disk → create it (`mkdir -p`) before writing
 - `goal_template` configured but file missing → fail with "template not found: {path}"
-- `--objective` not resolvable → MODE=tool fails; MODE=interactive prompts
+- `--objective` not resolvable → MODE=non_interactive fails; MODE=interactive prompts
 - Filename collision → see step 10
 - Timeline > 4 weeks → fail with "goals must be ≤ 4 weeks; use an objective for longer horizons"
 </error_handling>

agents/goal-manager-agent.md

@@ -83,7 +83,7 @@ Show goal status. Accepts an explicit goal name/path, or detects from conversati
      - MODE=interactive: parse conversation for file paths, wiki links, goal mentions
        - 0 matches → error "No active goal detected; pass a goal name explicitly"
        - >1 matches → AskUserQuestion to select
-     - MODE=tool: return `{"success": false, "error": "goal name required in tool mode"}` and STOP
+     - MODE=non_interactive: return `{"success": false, "error": "goal name required in non_interactive mode"}` and STOP
 
 2. **Find goal file:**
    ```

agents/task-creator.md

@@ -30,9 +30,9 @@ You create one task file in the configured vault. You read the vault config via
 
 ## 1. Parse arguments
 
-Input form: `[task description] [--tool] [--vault NAME]`.
+Input form: `[task description] [--non-interactive] [--vault NAME]`.
 
-- `--tool` → MODE = `tool` (machine-readable JSON output, no AskUserQuestion calls, no audit, hard-fail on collision)
+- `--non-interactive` (deprecated alias: `--tool`) → MODE = `non_interactive` (machine-readable JSON output, no AskUserQuestion calls, no audit, hard-fail on collision)
 - `--vault NAME` → target vault override
 - Remaining tokens → task description / title
 
@@ -51,7 +51,7 @@ Find the vault entry matching the requested vault name (or `default_vault` from
 - `task_template` — absolute path to the task template (or empty/absent)
 - `goals_dir` — for optional parent-goal linking (or empty)
 
-If the vault is not found, report the error and stop. In MODE=tool, return `{"success": false, "error": "..."}`.
+If the vault is not found, report the error and stop. In MODE=non_interactive, return `{"success": false, "error": "..."}`.
 
 ## 3. Detect Jira / issue key in the description
 
@@ -60,7 +60,7 @@ Scan the description for patterns like `BRO-18665`, `TRADE-4304`, `OC-2042`, etc
 - If found, store as `JIRA_KEY` and remove from the title text.
 - If a single key is found, prepend it to the final filename: `{JIRA_KEY} {Title}.md`.
 
-In MODE=interactive, optionally fetch the Jira issue details via `mcp__atlassian__getJiraIssue` to enrich the title and description (only if a Jira key was detected and the MCP tool is available). Skip in MODE=tool.
+In MODE=interactive, optionally fetch the Jira issue details via `mcp__atlassian__getJiraIssue` to enrich the title and description (only if a Jira key was detected and the MCP tool is available). Skip in MODE=non_interactive.
 
 ## 4. Detect incident-shaped task (interactive only)
 
@@ -77,7 +77,7 @@ If matched, AskUserQuestion with a SEV-1..SEV-4 picker:
 
 If the user picks a severity, store as `SEVERITY` for inclusion in frontmatter.
 
-Skip this step entirely in MODE=tool.
+Skip this step entirely in MODE=non_interactive.
 
 ## 5. Compose the title
 
@@ -90,7 +90,7 @@ Final filename: `{JIRA_KEY }{Title}.md` (Jira key prefix only when detected).
 
 ## 6. Determine category and priority
 
-- **Category**: infer from keywords or task content (e.g. `octopus`, `trading`, `personal`); if unclear, leave empty in tool mode or AskUserQuestion in interactive mode
+- **Category**: infer from keywords or task content (e.g. `octopus`, `trading`, `personal`); if unclear, leave empty in non_interactive mode or AskUserQuestion in interactive mode
 - **Priority**: default 3 (low/normal); raise to 2 if SEV-2/3 incident or if user description signals urgency; 1 only on SEV-1
 
 ## 7. Resolve template body
@@ -145,7 +145,7 @@ Glob: {vault.path}/{tasks_dir}/{filename}
 
 If the file already exists:
 
-- MODE=tool → return `{"success": false, "error": "task file already exists: ..."}`
+- MODE=non_interactive → return `{"success": false, "error": "task file already exists: ..."}`
 - MODE=interactive → AskUserQuestion: 1. Pick a different name  2. Cancel
 
 ## 11. Write the file
@@ -165,7 +165,7 @@ Run a light self-audit against the file:
 - Body has Success Criteria + Tasks sections (or template body)
 - No accidental empty sections
 
-Skip in MODE=tool.
+Skip in MODE=non_interactive.
 
 ## 13. Return
 
@@ -182,13 +182,13 @@ Next steps:
 3. Defer: /defer-task "{title}" <date>
 ```
 
-MODE=tool output (single JSON object on stdout, nothing else):
+MODE=non_interactive output (single JSON object on stdout, nothing else):
 
 ```json
 {"success": true, "path": "{absolute_path}", "filename": "{filename}"}
 ```
 
-On error in MODE=tool:
+On error in MODE=non_interactive:
 
 ```json
 {"success": false, "error": "<message>"}
@@ -201,5 +201,5 @@ On error in MODE=tool:
 - `tasks_dir` directory does not exist on disk → create it (`mkdir -p`) before writing
 - `task_template` configured but file missing → fail with "template not found: {path}"
 - Filename collision → see step 10
-- Unable to detect category/priority and MODE=tool → use category=`""` and priority=3, do not block
+- Unable to detect category/priority and MODE=non_interactive → use category=`""` and priority=3, do not block
 </error_handling>

commands/complete-goal.md

@@ -1,6 +1,6 @@
 ---
 description: Mark goal as complete (checks subtasks and success criteria)
-argument-hint: "<goal-name-or-path> [--tool] [--force]"
+argument-hint: "<goal-name-or-path> [--non-interactive] [--force]"
 ---
 
 <objective>
@@ -9,7 +9,7 @@ Mark goal as complete using vault-cli. Verifies success criteria and linked subt
 
 <process>
 1. Parse arguments:
-   - If contains `--tool` → MODE=tool, remove flag from args
+   - If contains `--non-interactive` (or deprecated `--tool`) → MODE=non_interactive, remove flag from args
    - If contains `--force` → FORCE=true, remove flag from args
    - Otherwise → MODE=interactive
    - Extract goal name from remaining args
@@ -41,7 +41,7 @@ Mark goal as complete using vault-cli. Verifies success criteria and linked subt
       ```
       - If warnings in output, show them
 
-3. **MODE=tool (--tool flag):**
+3. **MODE=non_interactive (--non-interactive flag):**
 
    a. Read goal file to check completion state
    b. If incomplete success criteria or open subtasks (and FORCE not set):
@@ -60,7 +60,7 @@ Mark goal as complete using vault-cli. Verifies success criteria and linked subt
 
 <success_criteria>
 - vault-cli goal complete invoked (NOT Edit tool for frontmatter)
-- **MODE=tool**: Returns JSON only, never forces unless `--force` explicitly passed
+- **MODE=non_interactive**: Returns JSON only, never forces unless `--force` explicitly passed
 - **MODE=interactive**: Shows progress, asks if incomplete, reports result
 - Parent objective updated (by vault-cli)
 - `--force` only used when user explicitly approves or passes the flag

commands/complete-task.md

@@ -1,6 +1,6 @@
 ---
 description: Mark task as complete (normal or recurring)
-argument-hint: "<task-name-or-path> [--tool] [--force]"
+argument-hint: "<task-name-or-path> [--non-interactive] [--force]"
 allowed-tools:
   - Bash(vault-cli:*)
   - Read
@@ -12,9 +12,9 @@ Mark task as complete using vault-cli. Handles normal and recurring tasks approp
 
 <process>
 1. Parse arguments:
-   - If contains `--tool` → MODE=tool, remove flag from args
+   - If contains `--non-interactive` (or deprecated `--tool`) → MODE=non_interactive, remove flag from args
    - Otherwise → MODE=interactive
-   - If contains `--force` → FORCE=true, remove flag from args (interactive mode only — `--tool` overrides as defined in step 3)
+   - If contains `--force` → FORCE=true, remove flag from args (interactive mode only — `--non-interactive` overrides as defined in step 3)
    - Extract task name from remaining args
 
 2. **MODE=interactive (default):**
@@ -49,9 +49,9 @@ Mark task as complete using vault-cli. Handles normal and recurring tasks approp
 
       Rationale for this closer (one-task-per-session contract, why never `/vault-cli:next-task` here, why the "no end-of-day suggestions" global rule does NOT override): see [`sync-progress.md` Phase 6](sync-progress.md) — same contract applies. Single source of truth lives there to prevent drift.
 
-      MODE=tool MUST NOT emit this panel (see step 3 — JSON only).
+      MODE=non_interactive MUST NOT emit this panel (see step 3 — JSON only).
 
-3. **MODE=tool (--tool flag):**
+3. **MODE=non_interactive (--non-interactive flag):**
 
    a. Read task file to check completion state
    b. If incomplete items:
@@ -70,7 +70,7 @@ Mark task as complete using vault-cli. Handles normal and recurring tasks approp
 
    d. Never ask questions, never prompt.
 
-   e. **Never emit the `⚪ DONE` closer panel** — MODE=tool output is JSON only. The closer panel from step 2e is interactive-mode only.
+   e. **Never emit the `⚪ DONE` closer panel** — MODE=non_interactive output is JSON only. The closer panel from step 2e is interactive-mode only.
 
 4. Task types (handled by vault-cli internally):
    - Normal tasks: status→completed, goals updated, daily note checked
@@ -79,7 +79,7 @@ Mark task as complete using vault-cli. Handles normal and recurring tasks approp
 
 <success_criteria>
 - vault-cli task complete invoked (NOT Edit tool for frontmatter)
-- **MODE=tool**: Returns JSON only, sets phase=human_review if incomplete
+- **MODE=non_interactive**: Returns JSON only, sets phase=human_review if incomplete
 - **MODE=interactive**: Shows completion %, aborts with `--force` hint if incomplete (no prompts), reports result on success
 - Goal files updated (by vault-cli)
 - Daily note updated (by vault-cli)

commands/create-goal.md

@@ -1,13 +1,13 @@
 ---
 description: Create a new goal in the configured vault with guided prompts
-argument-hint: "[goal description] [--tool] [--vault NAME] [--objective NAME]"
+argument-hint: "[goal description] [--non-interactive] [--vault NAME] [--objective NAME]"
 allowed-tools: [Task, Read, Write, Glob, Grep, Bash, AskUserQuestion]
 ---
 
 Invoke the goal-creator agent to create a goal file in the vault.
 
 Parse `$ARGUMENTS`:
-- `--tool` → MODE=tool (orchestration mode, JSON output, no prompts)
+- `--non-interactive` (deprecated alias: `--tool`) → MODE=non_interactive (orchestration mode, JSON output, no prompts)
 - `--vault NAME` → target a specific vault (otherwise use the default vault from `~/.vault-cli/config.yaml`)
 - `--objective NAME` → link to parent objective by name (interactive mode resolves via Glob/AskUserQuestion if ambiguous)
 - Remaining text → goal description / title

commands/create-task.md

@@ -1,13 +1,13 @@
 ---
 description: Create a new task in the configured vault with guided prompts
-argument-hint: "[task description] [--tool] [--vault NAME]"
+argument-hint: "[task description] [--non-interactive] [--vault NAME]"
 allowed-tools: [Task, Read, Write, Glob, Grep, Bash, AskUserQuestion]
 ---
 
 Invoke the task-creator agent to create a task file in the vault.
 
 Parse `$ARGUMENTS`:
-- `--tool` → MODE=tool (orchestration mode, JSON output, no prompts)
+- `--non-interactive` (deprecated alias: `--tool`) → MODE=non_interactive (orchestration mode, JSON output, no prompts)
 - `--vault NAME` → target a specific vault (otherwise use the default vault from `~/.vault-cli/config.yaml`)
 - Remaining text → task description / title
 

commands/defer-goal.md

@@ -1,6 +1,6 @@
 ---
 description: Defer goal to specific date
-argument-hint: "<goal-name> <YYYY-MM-DD|+Nd|weekday> [--tool]"
+argument-hint: "<goal-name> <YYYY-MM-DD|+Nd|weekday> [--non-interactive]"
 allowed-tools: Bash(vault-cli goal defer:*)
 ---
 
@@ -10,7 +10,7 @@ Defer goal to specific date using vault-cli. Supports absolute (YYYY-MM-DD), rel
 
 <process>
 1. Parse arguments:
-   - If contains `--tool` → MODE=tool, remove flag from args
+   - If contains `--non-interactive` (or deprecated `--tool`) → MODE=non_interactive, remove flag from args
    - Otherwise → MODE=interactive
    - Split remaining args: goal name = all but last, date = last word
 
@@ -30,7 +30,7 @@ Defer goal to specific date using vault-cli. Supports absolute (YYYY-MM-DD), rel
       ```
       - If warnings in output, show them
 
-3. **MODE=tool (--tool flag):**
+3. **MODE=non_interactive (--non-interactive flag):**
 
    a. Run vault-cli:
       ```bash
@@ -50,6 +50,6 @@ Defer goal to specific date using vault-cli. Supports absolute (YYYY-MM-DD), rel
 - vault-cli goal defer invoked (NOT Edit tool for frontmatter)
 - Date parsed and validated (by vault-cli)
 - Goal defer_date updated (by vault-cli)
-- MODE=tool: Returns JSON only
+- MODE=non_interactive: Returns JSON only
 - MODE=interactive: Full report
 </success_criteria>