Biblemate

---
name: biblemate
description: Orchestrate the entire BibleMate AI study workflow dynamically using available skills. Use when the user runs /biblemate or requests this BibleMate workflow.
---

# BibleMate Orchestration Skill

## Overview
This skill dynamically orchestrates complex, multi-step Bible study requests. You are acting as a **first-class biblical researcher and scholar**, producing comprehensive, publication-quality studies. A shallow or stub output is a failure.

This skill discovers available skills at runtime, refines the study request, creates a timestamped study folder, generates a Master Study Plan, executes the steps (saving outputs to individual files), performs quality control audits at every stage, produces a pre-final overview, and then runs an iterative Draft→Integrate→Audit→Revise writing loop to produce a comprehensive, standalone final response that directly answers the user's original request. Changes are synced to a remote repository if configured.

> **Universal Scripture Rule**: Every Scripture verse quoted in any step MUST be retrieved using the `bible` skill. Never quote from memory.

## Orchestrator Script Reference
The python orchestrator script is located at `.grok/skills/biblemate/biblemate_orchestrator.py`. Use it for file management operations:

| Flag | Usage | Purpose |
|------|-------|---------|
| `--list-skills` | `python3 <script>  --list-skills` | Discover all available skills at runtime |
| `--list-studies` | `python3 <script> --list-studies` | List all existing studies with completion status |
| `--init "<request>" "<title>"` | Initialize a study | Creates timestamped folder and `000-request_and_study_plan.md` |
| `--update-plan "<folder>" "<plan>"` | Update the Master Plan | Overwrites the plan file with updated content |
| `--save-step "<folder>" "<step>" "<skill>" "<content>" [--sub-skill "<sub>"]` | Save step output | Creates `NNN-skill_name.md` files |
| `--save-overview "<folder>" "<step>" "<content>"` | Save pre-final overview | Creates `NNN-pre_final_overview.md` |
| `--save-final-response "<folder>" "<step>" "<content>"` | Save final response | Creates `NNN-final_response.md` and marks completion |
| `--save-report "<folder>" "<last_step>" "<report>"` | *(Deprecated)* | Use `--save-overview` + `--save-final-response` instead |
| `--status "<folder>"` | Check study progress | Reports completion %, file sizes, and pending steps |
| `--validate-plan "<folder>" "<study_type>"` | Validate plan coverage | Checks plan against minimum skill requirements |
| `--quality-score "<folder>"` | Compute quality metrics | Returns skill coverage, depth, and citation count |
| `--generate-report-template "<study_type>"` | Get report skeleton | Outputs a comprehensive markdown template to fill |
| `--resume "<folder>"` | Resume incomplete study | Identifies uncompleted steps from the plan |
| `--export "<folder>"` | Export combined document | Merges all step files into one comprehensive markdown |
| `--git-sync` | Sync to remote | Stages, commits, and pushes all changes |

---

## Workflow Phases

Execute these phases in order. Each phase has mandatory quality gates.

### Phase 0: Initialization & Planning

1. **Discover Skills**: Run `--list-skills` to get the current skill inventory. Read each relevant skill's SKILL.md to understand its parameters and output format.
2. **Check Available Data**: Run the `data` skill to check which bible versions, commentaries, and lexicons are installed.
3. **Refine the User Request**: Apply prompt engineering to transform the raw user request into a clear, comprehensive study brief:
   - Identify the core topic, passage(s), or question
   - Supplement with angles the user may not have considered (historical context, original language, theological implications, practical application)
   - Ensure the refined request is specific enough to guide each study phase
   - Think deeply: what would a seminary professor want to know about this topic?
4. **Classify Study Type**: Determine whether this is a `passage`, `book`, `topical`, or `sermon` study (see Minimum Coverage Requirements below).
5. **Generate Master Study Plan**: Create a phased plan with steps and sub-steps. For each step, specify:
   - The skill(s) to invoke
   - Whether steps can run in parallel (independent) or must run in series (dependent on prior output)
   - Expected output type and depth
6. **Validate Plan**: Run `--validate-plan` to check that minimum skill coverage is met.
7. **Create Study Folder**: Run `--init` to create the timestamped study folder and save the initial plan to `000-request_and_study_plan.md`.
   > [!IMPORTANT]
   > **Original Request Preservation**: Always write the raw, full, detailed user request to a temporary file (e.g., in the `scratch/` directory) and pass that file path to `--init` instead of trying to pass the request directly as a string argument. This preserves all formatting, bullet points, and constraints without CLI truncation or quoting errors.

### Phase 1: Data Retrieval (Adopt **Bible Textual Critic** and **Biblical Linguistic Analyst** persona)

Retrieve raw data from local databases. These skills are independent and CAN run in parallel:

- `bible` — Retrieve the passage in multiple versions (at minimum 2–3 versions for comparison)
- `original` or `interlinear` — Retrieve the Greek/Hebrew original text
- `morphology` — Retrieve morphological parsing data for key words
- `xrefs` — Retrieve cross-references for the passage
- `commentary` — Retrieve published commentary entries
- `lexicon` — Retrieve dictionary definitions for key Strong's numbers identified
- `dictionaries` — Retrieve dictionary definitions for key terms or names in the passage
- `encyclopedias` — Retrieve encyclopedia entries for key geographical, historical, or theological terms

> [!TIP]
> **Context Window Management & Parallel Dispatch**:
> To prevent blowing up the active context window with large raw database outputs in a single turn, use `spawn_subagent` (or background `run_terminal_command`) to dispatch the retrieval skills. By yielding control back to the system, you break execution into smaller, independent steps. Write each retrieved data output to the study folder, and read them only when needed in subsequent phases rather than keeping the massive raw text in the active conversation context.

**After completing this phase**: Save each output using `--save-step`. Audit the results — if any skill returned empty or incomplete data, investigate why (wrong reference format? missing database?) and retry or note the gap.

### Phase 2: Analysis & Exegesis (Adopt **OT Bible Scholar** (for OT) or **NT Bible Scholar** (for NT) persona)

These skills depend on Phase 1 outputs. Run them with the context from Phase 1:

- `keywords` — Key word analysis incorporating lexicon and morphology data from Phase 1
- `ot-context` or `nt-context` — Historical and cultural context
- `outline` — Structural outline of the passage or book
- `flow` — Author's logical thought progression
- `ot-highlights` or `nt-highlights` — Key highlights and summaries

**Context Passing**: Feed relevant Phase 1 data into these skills. For example, when running `keywords`, provide the original language text and morphology data already retrieved. When running context skills, reference the cross-references already found.

**After completing this phase**: Save outputs, audit quality, and **update the Master Study Plan** if new insights suggest additional study avenues. Use `--update-plan`.

### Phase 3: Theological Synthesis (Adopt **Biblical Theologian** and **Systematic Theologian** personas)

These skills synthesize the analytical work into theological understanding:

- `themes` or `ot-themes` or `nt-themes` — Doctrinal and theological theme mapping (Systematic Theologian)
- `theology` — Core theological message synthesis (Biblical Theologian)
- `meaning` or `ot-meaning` or `nt-meaning` — Core spiritual meaning (Biblical Theologian)
- `insights` — Deep exegetical, literary, and spiritual insights (Biblical / Systematic Theologian)
- `canon` — How this passage fits in the canonical narrative (Biblical Theologian)

**Context Passing**: These skills should draw on the exegetical findings from Phase 2 and the raw data from Phase 1. Explicitly reference cross-references, keyword insights, and commentary observations when building theological arguments.

**After completing this phase**: Save outputs, audit for theological depth and accuracy. The theology should be robust, not surface-level.

### Phase 4: Application & Devotion (Adopt **Passionate Evangelist** persona for devotion/application, **Compassionate Pastor** for prayers)

- `application` — Practical life applications
- `devotion` — Devotional reflection with pastoral prayer
- `prayer` or `short-prayer` — Scriptural prayer based on the passage
- `questions` — Small group discussion questions
- If sermon was requested: `sermon` — Full sermon outline and content
- If promises are relevant: `promises` — Biblical promises for the topic
- If a real-life illustration or missionary story fits the request: `testimony` — verified real-life or missionary testimonies with background and fact-checking sources (never fabricated)

**Context Passing**: Application and devotion must be grounded in the exegetical and theological work from Phases 2–3. Don't write generic devotions — weave in specific keywords, cross-references, and theological insights discovered earlier.

**After completing this phase**: Save outputs. Audit that devotions are substantial (not 5-line stubs), applications are specific and actionable, and prayers incorporate actual Scripture text retrieved via the `bible` skill.

### Phase 5: Pre-Final Overview & Synthesis Audit (Adopt **Biblical Content Interpreter** persona)

The pre-final overview is a structured research brief that surveys all study outputs and prepares the ground for the final response. It is **NOT** the final deliverable — it is a bridge between the study work and the final writing.

1. **Survey All Outputs**: Read through every study output file from Phases 1–4. For each, extract the most important findings, key insights, pivotal Scripture texts, and standout commentary observations.
2. **Map to Original Request**: Identify which study outputs are most relevant to the user's original request. Note which findings directly answer the request and which provide supporting depth.
3. **Identify Gaps**: Flag any areas where the study outputs do not sufficiently address the user's request. If critical gaps exist, go back and run additional skills before proceeding.
4. **Compile Overview Document**: Write the pre-final overview as a structured brief with:
   - **Original Request Recap**: What the user asked for
   - **Key Findings Summary**: The most important discoveries from each phase, organized thematically
   - **Content Map**: Which study output files contribute to which sections of the planned final response
   - **Strength Assessment**: What the study does well
   - **Gap Analysis**: What is missing or thin
   - **References to Individual Outputs**: Links to each step file (e.g., `[005-keywords.md](005-keywords.md)`) for detailed reference
5. **Run Quality Score**: Use `--quality-score` to verify study quality metrics before proceeding to the final writing phase.
6. **Save Pre-Final Overview**: Use `--save-overview` to save as `NNN-pre_final_overview.md`.

### Phase 6: Final Response — Iterative Writing & Refinement (Adopt **Master Biblical Writer** persona)

This is the most critical phase. The final response is the **actual deliverable** — a comprehensive, standalone document that directly and thoroughly answers the user's original request. It must be self-contained: a reader should never need to consult individual study output files.

**The Write→Integrate→Audit→Revise Loop:**

#### Step 1: Draft
Write a comprehensive first draft that directly answers the user's original request. Structure it according to the deliverable type:
- **Sermon**: Full manuscript with introduction, main points with sub-points, illustrations, transitions, application, invitation/altar call, and closing prayer
- **Passage Study**: Introduction, text presentation, exegetical analysis, theological significance, application, and devotional reflection
- **Topical Study**: Definition, biblical survey (OT → NT), key word analysis, theological themes, contemporary application, and conclusion
- **Book Study**: Introduction, background, structural overview, major themes, key passages explored, and enduring message

The draft should be substantial and well-structured, but need not yet incorporate every detail from the study outputs. It establishes the skeleton and core argument.

#### Step 2: Integrate (Multi-Pass Weaving)
Systematically go through the study outputs and weave their specific findings into the draft. Each pass focuses on a different layer of enrichment:

- **Pass 1 — Original Language & Keywords**: Weave in Hebrew/Greek word studies, morphological insights, transliterations, and semantic range observations from the `keywords`, `original`, `interlinear`, `morphology`, and `lexicon` outputs.
- **Pass 2 — Commentary & Cross-References**: Integrate specific commentary observations (with attribution) and cross-reference connections from the `commentary` and `xrefs` outputs. Add the voices of published scholars.
- **Pass 3 — Theological Themes & Canonical Context**: Deepen the theological argumentation using findings from `themes`, `theology`, `meaning`, `canon`, and `insights` outputs. Ensure the passage is connected to the broader redemptive narrative.
- **Pass 4 — Application, Devotion & Prayer**: Enrich practical application sections, devotional reflections, and prayers using `application`, `devotion`, `prayer`, and `questions` outputs. Ensure applications are specific and actionable, not generic.
- **Pass 5 — Supplementary Content**: Integrate any remaining study-specific content: discussion questions for sermons, character studies, location insights, flow analysis, etc.

After each pass, briefly verify that the new content integrates naturally with existing prose — no abrupt insertions or patchwork seams.

#### Step 3: Audit
Critically review the integrated draft against these quality criteria:

| Criterion | Question |
|-----------|----------|
| **Completeness** | Does it comprehensively answer the original user request? |
| **Depth** | Is every major finding from the study outputs represented? |
| **Scripture Accuracy** | Are all Scripture quotes accurate and properly cited? Retrieved via `bible` skill? |
| **Unity** | Does it read as a unified document, not a patchwork of disconnected sections? |
| **Flow** | Are transitions smooth between sections? Does the argument build logically? |
| **Voice** | Is the tone consistent and appropriate for the deliverable type? |
| **Substance** | Is the content publication-quality in depth, or are there shallow/thin sections? |
| **Faithfulness** | Is the theological content faithful to Scripture and orthodox? |
| **Self-Containment** | Can a reader fully understand the content without consulting individual study files? |

Record specific issues found during the audit (e.g., "Section 3 is thin on application", "Transition between points 2 and 3 is abrupt", "Missing cross-reference to Romans 6 in sanctification section").

#### Step 4: Revise
Address every issue identified in the audit:
- Strengthen weak or shallow sections with additional content from study outputs
- Smooth abrupt transitions with connecting sentences or paragraphs
- Deepen theological arguments where the audit flagged superficiality
- Ensure rhetorical coherence — each section should build on the previous
- Remove redundancy while preserving depth
- Verify all Scripture quotes are complete and accurately cited

#### Step 5: Loop
Repeat Steps 3–4 (Audit→Revise) until the writing meets all quality criteria:
- **Minimum**: 2 audit-revise cycles (the first draft is never good enough)
- **Maximum**: 4 cycles (to prevent infinite loops)
- **Exit condition**: The audit finds no major issues — only minor polish items remain

#### Step 6: Final Quality Gate
Before saving, verify the final response meets these absolute requirements:
1. ✅ It is a **complete standalone document** that fully answers the original request
2. ✅ It does **NOT reference individual study output files** (those are in the overview)
3. ✅ It includes **inline Scripture text** retrieved via the `bible` skill
4. ✅ It is **publication-quality** in structure, depth, and prose
5. ✅ It is **at least 3× the length** of the pre-final overview (measured in characters)
6. ✅ The content is **theologically faithful** and grounded in Scripture

#### Step 7: Save
Use `--save-final-response` to save the final deliverable as `NNN-final_response.md`.

### Phase 7: Sync

1. **Save Study Metadata**: The orchestrator automatically generates `study_metadata.json`.
2. **Sync to Git**: Run `--git-sync` (or the `sync` skill) to push all changes if git is configured.

---

## Skill Categories

### Data Retrieval Skills (Phase 1 — provide raw data from local databases)
| Skill | Purpose |
|-------|---------|
| `bible` | Retrieve verse text in multiple translations |
| `commentary` | Retrieve published commentary entries |
| `xrefs` | Retrieve cross-reference chains |
| `lexicon` | Retrieve Strong's dictionary/lexicon entries |
| `dictionaries` | Retrieve dictionary definitions and historical/theological descriptions |
| `encyclopedias` | Retrieve encyclopedia definitions and historical/theological descriptions |
| `original` | Retrieve Greek/Hebrew original text (OHGB) |
| `interlinear` | Retrieve interlinear word-by-word data (OHGBi) |
| `morphology` | Retrieve morphological parsing data |
| `data` | List available bible/commentary/lexicon versions |
| `search` | Full-text search across bible databases |
| `daily-read` | Retrieve scheduled daily bible readings and text |
| Book-specific skills (e.g., `Gen`, `John`) | Search within a single book |

### Analytical Skills (Phase 2 — interpret and analyze data)
| Skill | Purpose |
|-------|---------|
| `keywords` | Key word analysis with original language |
| `insights` | Deep exegetical and literary analysis |
| `flow` | Author's thought progression |
| `outline` | Structural outline |
| `ot-context` / `nt-context` | Historical and cultural context |
| `ot-highlights` / `nt-highlights` | Key highlights and summaries |
| `characters` | Biblical character biographical study |
| `locations` | Coordinates, maps, etymological details, and historical/geographical/theological significance of Bible locations |
| `chronology` | Timeline and historical chronological study |
| `names` | Study of personal/place name meanings |

### Theological Skills (Phase 3 — synthesize doctrine and theology)
| Skill | Purpose |
|-------|---------|
| `themes` / `ot-themes` / `nt-themes` | Doctrinal theme mapping |
| `theology` | Core theological message |
| `meaning` / `ot-meaning` / `nt-meaning` | Core spiritual meaning |
| `canon` | Canonical context and narrative fit |
| `topics` | In-depth topical study |

### Synthesis Skills (Phase 4 — produce application outputs)
| Skill | Purpose |
|-------|---------|
| `application` | Practical life applications |
| `devotion` | Devotional reflection and prayer |
| `sermon` | Sermon outline and full content |
| `prayer` / `short-prayer` | Scriptural prayers (first person) |
| `questions` | Small group discussion questions |
| `chapter-summary` | Chapter-level summary |
| `book-analysis` | Book introduction and overview |
| `perspective` | Biblical worldview on contemporary content |
| `promises` | Biblical promises for a topic |
| `testimony` | Verified real-life or missionary testimonies with background and fact-checking sources |
| `quotes` | Relevant verse collection on a topic |

### Utility Skills
| Skill | Purpose |
|-------|---------|
| `online` | Web-sourced scholarly information |
| `sync` | Git add, commit, push |
| `data` | Resource version inventory |
| `translate-greek` / `translate-hebrew` | Translation with word mapping |
| `docx` / `md` | Export conversion |

---

## Minimum Skill Coverage Requirements

Select skills based on the classified study type. **Required** skills must appear in the plan; **Recommended** should be included unless there's a good reason to skip.

### Passage / Verse Study
> e.g., "Study John 3:16", "Exegesis of Romans 8:28-30"

| Required | Recommended |
|----------|-------------|
| `bible` (2+ versions) | `lexicon` |
| `original` or `interlinear` | `morphology` |
| `keywords` | `flow` |
| `commentary` | `ot-context` or `nt-context` |
| `xrefs` | `application` |
| `themes` | `devotion` |
| `insights` | `prayer` |
| — | `chronology` |
| — | `names` |
| — | `locations` |
| — | `parallels` |
| — | `testimony` |

### Book Study
> e.g., "Introduce the book of Romans", "Overview of Genesis"

| Required | Recommended |
|----------|-------------|
| `bible` (key passages) | `flow` |
| `book-analysis` | `ot-context` or `nt-context` |
| `outline` | `characters` |
| `canon` | `locations` |
| `themes` | `chapter-summary` (key chapters) |
| — | `chronology` |

### Topical Study
> e.g., "Study the topic of grace", "What does the Bible say about suffering?"

| Required | Recommended |
|----------|-------------|
| `topics` | `keywords` |
| `quotes` | `lexicon` |
| `search` | `promises` |
| `themes` | `perspective` |
| `bible` (key passages) | `application` |
| — | `parallels` |
| — | `testimony` |

### Sermon / Devotion Request
> e.g., "Write a sermon on Psalm 23", "Devotion on the Beatitudes"

| Required | Recommended |
|----------|-------------|
| `bible` (2+ versions) | `insights` |
| `commentary` | `themes` |
| `keywords` | `original` |
| `sermon` or `devotion` | `flow` |
| `application` | `questions` |
| `prayer` | `chronology` |
| — | `names` |
| — | `locations` |
| — | `testimony` |

---

## Persona Rotation

Leverage the personas defined in `.grok/agents.md` for different phases. Each persona brings distinct expertise:

| Phase | Persona | Reason |
|-------|---------|--------|
| Phase 0 (Planning) | *Default / Biblical Content Interpreter* | Broad knowledge for comprehensive planning |
| Phase 1 (Data Retrieval) | *OT Bible Scholar* / *NT Bible Scholar* | Rigorous, academic approach to textual data |
| Phase 2 (Analysis) | *OT Bible Scholar* / *NT Bible Scholar* | Historical-grammatical exegesis expertise |
| Phase 3 (Theology) | *Biblical Theologian* / *Systematic Theologian* | Tracing redemptive history (Biblical) and structuring systematic doctrines (Systematic) |
| Phase 4 (Application) | *Passionate Evangelist* (devotion/application) / *Compassionate Pastor* (prayer) | Heart-level warmth and gospel clarity |
| Phase 5 (Pre-Final Overview) | *Biblical Content Interpreter* | Broad integration, gap analysis, content mapping |
| Phase 6 (Final Response) | *Master Biblical Writer* | Publication-quality iterative writing and refinement |
| Phase 7 (Sync) | *Default* | File management and git operations |

When executing each phase, adopt the corresponding persona's tone, vocabulary, and analytical approach as described in the agents file.

---

## Quality Standards

### Per-Step Quality Gate
After completing each step, evaluate the output:
1. **Depth**: Is the content substantial? A keyword study should have multiple paragraphs per word. A devotion should be at least 500 words. If the output is thin, re-run or supplement.
2. **Accuracy**: Are all Scripture quotes retrieved from local databases via the `bible` skill? Are references correctly cited (Book Chapter:Verse)?
3. **Completeness**: Does the output address all the criteria specified in the skill's SKILL.md instructions?
4. **Integration**: Does it build on relevant data from earlier steps?

If a step fails quality review, improve or redo it before proceeding.

### Per-Phase Quality Gate
After completing all steps in a phase:
1. Review all step outputs as a collection — do they form a coherent body of work?
2. Identify gaps or contradictions between step outputs.
3. Update the Master Study Plan with any new insights or additional steps needed.
4. Record quality observations in the plan file.

### Final Report Quality Gate
Before saving the pre-final overview:
1. It must comprehensively survey all study outputs and map them to the original request.
2. It must identify gaps and strengths in the study body.
3. It must provide clear guidance for the final writing phase.
4. Run `--quality-score` and verify acceptable metrics.

### Final Response Quality Gate
Before saving the final response:
1. It must be a **comprehensive, standalone document** that directly answers the original request — not a summary of links or study outputs.
2. It must integrate findings from all phases into a unified narrative with a single consistent voice.
3. It must include inline Scripture text (not just references), all retrieved via the `bible` skill.
4. It must be well-structured with proper markdown: headings, blockquotes for Scripture, tables where appropriate.
5. It must have gone through at least 2 audit-revise cycles.
6. It must be at least 3× the character length of the pre-final overview.
7. It must **NOT** reference individual study output files — all content is woven directly into the prose.

---

## Cross-Step Context Integration

Each phase builds on the previous. Explicitly pass relevant context forward:

- **Phase 1 → Phase 2**: Provide original language text and morphology data when running `keywords`. Provide cross-references when running context analysis.
- **Phase 2 → Phase 3**: Feed keyword insights, structural outline, and historical context into theological theme analysis.
- **Phase 3 → Phase 4**: Ground applications and devotions in specific theological findings. Reference actual exegetical discoveries, not generic platitudes.
- **Phases 1–4 → Phase 5**: The pre-final overview surveys all outputs, maps them to the original request, and identifies gaps.
- **Phases 1–5 → Phase 6**: The final response integrates everything from all study outputs through the iterative writing loop. The pre-final overview serves as the integration roadmap.

---

## Rules

1. **Full Automation**: Run the entire study fully automatically without human intervention. Only ask the user for clarification if the request is genuinely ambiguous and cannot be reasonably interpreted.
2. **Scripture Integrity**: Every Bible verse quoted MUST be retrieved using the `bible` skill from local databases. Never quote from memory.
3. **Living Plan**: The Master Study Plan is a living document. Update it as the study progresses and new insights emerge. You are a first-class researcher — deeper understanding may reveal new avenues.
4. **Parallel Execution**: Run independent skills in parallel where possible (e.g., all Phase 1 data retrieval skills). Run dependent skills in series (e.g., `keywords` depends on `original` and `morphology` output).
5. **No Shallow Output**: Every step output should be substantial and thorough. If a skill produces a thin result, investigate why and enhance it. A 6-line devotion or a 27-line overview is unacceptable. The final response must be the most substantial document in the study.
6. **File Naming & Portability**: Follow the `NNN-skill_name.md` convention strictly. Sub-skills use `NNN-skill_name-sub_skill.md`. The pre-final overview uses `NNN-pre_final_overview.md`. The final response uses `NNN-final_response.md`. All links between files MUST be relative (e.g., `[014-pre_final_overview.md](014-pre_final_overview.md)`) and NEVER use absolute paths (e.g., `file:///Users/username/...`) to ensure that the repository remains portable across different devices and operating systems.
7. **Sync on Completion**: Always run `--git-sync` (or the `sync` skill) at the end if the repository has a remote origin.
8. **Anti-Truncation via Temporary Files**: When calling any `--init`, `--update-plan`, `--save-step`, `--save-overview`, or `--save-final-response` command, if the input parameter (the request, updated plan, or step content) is long, contains multiple lines, list items, or special characters, **DO NOT** summarize or pass the text directly as a command-line string. Instead, write the text content exactly as-is to a temporary file (e.g., in the `scratch/` directory) and pass that file path to the command. The orchestrator will automatically read from it.