# Claude Templates — On-Demand Reference > **Do NOT read this file at session start.** Read it only when you need to write a summary, handoff, decision record, or subagent output. This file is referenced from CLAUDE.md. --- ## Template 1: Source Document Summary **Use when:** Processing any input document (client brief, research report, requirements doc, existing proposal) **Write to:** `./docs/summaries/source-[filename].md` ```markdown # Source Summary: [Original Document Name] **Processed:** [YYYY-MM-DD] **Source Path:** [exact file path] **Archived From:** [original path, if moved to docs/archive/] **Document Type:** [brief / requirements / research / proposal / transcript / other] **Confidence:** [high = I understood everything / medium = some interpretation needed / low = significant gaps] ## Exact Numbers & Metrics - [metric]: [exact value] (page/section reference if available) - [metric]: [exact value] ## Key Facts (Confirmed) - [fact] — stated in [section/page] - [fact] — stated in [section/page] ## Requirements & Constraints - REQUIREMENT: [what is needed] - CONDITION: [when/if this applies] - CONSTRAINT: [limitation or exception] - PRIORITY: [must-have / should-have / nice-to-have / stated by whom] ## Decisions Referenced - DECISION: [what was decided] - RATIONALE: [why, as stated in document] - ALTERNATIVES MENTIONED: [what else was considered] - DECIDED BY: [who, if stated] ## Relationships to Other Documents - SUPPORTS: [other document/decision it reinforces] - CONTRADICTS: [other document/decision it conflicts with] - DEPENDS ON: [other document/decision it requires] - UPDATES: [other document/decision it supersedes] ## Open Questions & Ambiguities - UNCLEAR: [what is ambiguous] — needs clarification from [whom] - ASSUMED: [interpretation made] — verify with [whom] - MISSING: [information referenced but not provided] ## Verbatim Quotes Worth Preserving - "[exact quote]" — [speaker/author], [context] ``` --- ## Template 2: Analysis / Research Summary **Use when:** Completing competitive analysis, market research, technical evaluation **Write to:** `./docs/summaries/analysis-[topic].md` ```markdown # Analysis Summary: [Topic] **Completed:** [YYYY-MM-DD] **Analysis Type:** [competitive / market / technical / financial / feasibility] **Sources Used:** [list source paths or URLs] **Confidence:** [high / medium / low — and WHY this confidence level] ## Core Finding (One Sentence) [Single sentence: the most important conclusion] ## Evidence Base | Data Point | Value | Source | Date of Data | |-----------|-------|--------|-------------| | [metric] | [exact value] | [source] | [date] | ## Detailed Findings ### Finding 1: [Name] - WHAT: [the finding] - SO WHAT: [why it matters for this project] - EVIDENCE: [specific supporting data] - CONFIDENCE: [high/medium/low] ### Finding 2: [Name] [same structure] ## Conditional Conclusions - IF [condition], THEN [conclusion], BECAUSE [evidence] - IF [alternative condition], THEN [different conclusion] ## What This Analysis Does NOT Cover - [topic not addressed and why] - [data not available] ## Recommended Next Steps 1. [action] — priority [high/medium/low], depends on [what] 2. [action] ``` --- ## Template 3: Decision Record **Use when:** Any significant decision is made during a session **Write to:** `./docs/summaries/decision-[number]-[topic].md` ```markdown # Decision Record: [Short Title] **Decision ID:** DR-[sequential number] **Date:** [YYYY-MM-DD] **Status:** CONFIRMED / PROVISIONAL / REQUIRES_VALIDATION ## Decision [One clear sentence: what was decided] ## Context [2-3 sentences: what situation prompted this decision] ## Rationale - CHOSE [option] BECAUSE: [specific reasons with data] - REJECTED [alternative 1] BECAUSE: [specific reasons] - REJECTED [alternative 2] BECAUSE: [specific reasons] ## Quantified Impact - [metric affected]: [expected change with numbers] - [cost/time/resource implication]: [specific figures] ## Conditions & Constraints - VALID IF: [conditions under which this decision holds] - REVISIT IF: [triggers that should cause reconsideration] - DEPENDS ON: [upstream decisions or facts this relies on] ## Stakeholder Input - [name/role]: [their stated position, if known] ## Downstream Effects - AFFECTS: [what other decisions, documents, or deliverables this impacts] - REQUIRES UPDATE TO: [specific files or deliverables that need revision] ``` --- ## Template 4: Session Handoff **Use when:** A session is ending (context limit approaching OR phase complete) **Write to:** `./docs/summaries/handoff-[YYYY-MM-DD]-[topic].md` **LIFECYCLE**: After writing a new handoff, move the PREVIOUS handoff to `docs/archive/handoffs/`. ```markdown # Session Handoff: [Topic] **Date:** [YYYY-MM-DD] **Session Duration:** [approximate] **Session Focus:** [one sentence] **Context Usage at Handoff:** [estimated percentage if known] ## What Was Accomplished 1. [task completed] → output at `[file path]` 2. [task completed] → output at `[file path]` ## Exact State of Work in Progress - [work item]: completed through [specific point], next step is [specific action] - [work item]: blocked on [specific issue] ## Decisions Made This Session - DR-[number]: [decision] (see `./docs/summaries/decision-[file]`) - [Ad-hoc decision]: [what] BECAUSE [why] — STATUS: [confirmed/provisional] ## Key Numbers Generated or Discovered This Session - [metric]: [value] — [context for where/how this was derived] ## Conditional Logic Established - IF [condition] THEN [approach] BECAUSE [rationale] ## Files Created or Modified | File Path | Action | Description | |-----------|--------|-------------| | `[path]` | Created | [what it contains] | | `[path]` | Modified | [what changed and why] | ## What the NEXT Session Should Do 1. **First**: [specific action with file paths] 2. **Then**: [specific action] 3. **Then**: [specific action] ## Open Questions Requiring User Input - [ ] [question] — impacts [what downstream deliverable] - [ ] [question] ## Assumptions That Need Validation - ASSUMED: [assumption] — validate by [method/person] ## What NOT to Re-Read - `[file path]` — already summarized in `[summary file path]` ## Files to Load Next Session - `[file path]` — needed for [reason] - `[file path]` — needed for [reason] ``` --- ## Template 5: Project Brief (Initial Setup) **Use when:** Creating the 00-project-brief.md at project start **Write to:** `./docs/summaries/00-project-brief.md` ```markdown # Project Brief: [Project Name] **Created:** [YYYY-MM-DD] **Last Updated:** [YYYY-MM-DD] ## Client - **Name:** [client name] - **Industry:** [industry] - **Size:** [employee count / revenue if known] - **Relationship:** [through AutomatonsX / Lagrange Data / direct / other] - **Key Contacts:** [names and roles if known] ## Engagement - **Type:** [proposal / workshop / competitive analysis / agent development / hybrid] - **Scope:** [one paragraph description] - **Target Deliverable:** [specific output expected] - **Timeline:** [deadline if known] - **Budget Context:** [if known — exact figures] ## Input Documents | Document | Path | Processed? | Summary At | |----------|------|-----------|------------| | [name] | `[path]` | Yes/No | `[summary path]` | ## Success Criteria - [criterion 1] - [criterion 2] ## Known Constraints - [constraint 1] - [constraint 2] ## Project Phase Tracker | Phase | Status | Summary File | Date | |-------|--------|-------------|------| | Discovery | Not Started / In Progress / Complete | `[path]` | | | Strategy | Not Started / In Progress / Complete | `[path]` | | | Deliverable Draft | Not Started / In Progress / Complete | `[path]` | | | Review & Polish | Not Started / In Progress / Complete | `[path]` | | ``` --- ## Template 6: Task Definition **Use when:** Defining a discrete unit of work before starting execution ```markdown ## Task: [name] **Date:** [YYYY-MM-DD] **Client:** [if applicable] **Work Type:** [proposal / workshop / analysis / content / agent development] ### Context Files to Load - `[file path]` — [why needed] ### Action [What to produce. Be specific about format, length, and scope.] ### Verify - [ ] Numbers match source data exactly - [ ] Open questions marked OPEN - [ ] Output matches what was requested, not what was assumed - [ ] Claims backed by specific data - [ ] Consistent with stored decisions in docs/context/ ### Done When - [ ] Output file exists at `[specific path]` - [ ] Summary written to `docs/summaries/[specific file]` ``` --- ## Subagent Output Contracts **CRITICAL: When subagents return results to the main agent, unstructured prose causes information loss. These output contracts define the EXACT format subagents must return.** ### Contract for Document Analysis Subagent ``` === DOCUMENT ANALYSIS OUTPUT === SOURCE: [file path] TYPE: [document type] CONFIDENCE: [high/medium/low] NUMBERS: - [metric]: [exact value] [repeat for all numbers found] REQUIREMENTS: - REQ: [requirement] | CONDITION: [if any] | PRIORITY: [level] | CONSTRAINT: [if any] [repeat] DECISIONS_REFERENCED: - DEC: [what] | WHY: [rationale] | BY: [who] [repeat] CONTRADICTIONS: - [this document says X] CONTRADICTS [other known fact Y] [repeat or NONE] OPEN: - [unresolved item] | NEEDS: [who/what to resolve] [repeat or NONE] QUOTES: - "[verbatim]" — [speaker], [context] [repeat, max 5] === END OUTPUT === ``` ### Contract for Research/Analysis Subagent ``` === RESEARCH OUTPUT === QUERY: [what was researched] SOURCES: [list] CONFIDENCE: [high/medium/low] BECAUSE [reason] CORE_FINDING: [one sentence] EVIDENCE: - [data point]: [exact value] | SOURCE: [where] | DATE: [when] [repeat] CONCLUSIONS: - IF [condition] THEN [conclusion] | EVIDENCE: [reference] [repeat] GAPS: - [what was not found or not covered] [repeat or NONE] NEXT_STEPS: - [recommended action] | PRIORITY: [level] [repeat] === END OUTPUT === ``` ### Contract for Review/QA Subagent ``` === REVIEW OUTPUT === REVIEWED: [file path or deliverable name] AGAINST: [what standard — spec, requirements, style guide] PASS: [yes/no/partial] ISSUES: - SEVERITY: [critical/major/minor] | ITEM: [description] | LOCATION: [where in document] | FIX: [suggested resolution] [repeat] MISSING: - [expected content/section not found] | REQUIRED_BY: [which requirement] [repeat or NONE] INCONSISTENCIES: - [item A says X] BUT [item B says Y] | RESOLUTION: [suggested] [repeat or NONE] STRENGTHS: - [what works well — for positive reinforcement in iteration] [max 3] === END OUTPUT === ``` --- ## Phase-Based Workflow Templates ### Template A: Enterprise Sales Deliverable ``` Phase 1: Discovery & Input Processing ├── Process all client documents → Source Document Summaries ├── Identify gaps in information → flag as OPEN items ├── Create Decision Records for any choices made ├── Write: ./docs/summaries/01-discovery-complete.md (Handoff Template) ├── → Suggest new session for Phase 2 Phase 2: Strategy & Positioning ├── Read summaries only (NOT source documents) ├── Competitive positioning analysis → Analysis Summary ├── Value proposition development ├── ROI framework construction with EXACT numbers ├── Write: ./docs/summaries/02-strategy-complete.md (Handoff Template) ├── → Suggest new session for Phase 3 Phase 3: Deliverable Creation ├── Read strategy summary + project brief only ├── Draft deliverable (proposal / deck / workshop plan) ├── Output to: ./output/deliverables/ ├── Write: ./docs/summaries/03-deliverable-draft.md (Handoff Template) ├── → Suggest new session for Phase 4 Phase 4: Review & Polish ├── Read draft deliverable + strategy summary ├── Quality review using Review/QA Output Contract ├── Final edits and formatting ├── Output final version to: ./output/deliverables/ ``` ### Template B: Agent/Application Development ``` Phase 1: Requirements → Spec ├── Process all input documents → Source Document Summaries ├── Generate structured specification ├── Output: ./output/SPEC.md ├── Write: ./docs/summaries/01-spec-complete.md (Handoff Template) ├── → Suggest new session for Phase 2 Phase 2: Architecture → Schema ├── Read SPEC.md + summaries only ├── Design data model ├── Define agent behaviors and workflows ├── Output: ./output/schemas/data-model.yaml ├── Output: ./output/schemas/agent-definitions.yaml ├── Write: ./docs/summaries/02-architecture-complete.md (Handoff Template) ├── → Suggest new session for Phase 3 Phase 3: Prompts → Integration ├── Read schemas + spec only ├── Write system prompts for each agent ├── Map API integrations and data flows ├── Output: ./output/prompts/[agent-name].md (one per agent) ├── Output: ./output/schemas/integration-map.yaml ├── Write: ./docs/summaries/03-prompts-complete.md (Handoff Template) ├── → Suggest new session for Phase 4 Phase 4: Assembly → Package ├── Read all output files ├── Assemble complete application package ├── Generate deployment/setup instructions ├── Output: ./output/deliverables/[project]-complete-package/ ├── QA check against original spec using Review/QA Output Contract ``` ### Template C: Hybrid (Sales + Agent Development) ``` Phase 1: Client Discovery → Summaries Phase 2: Solution Design → Architecture + Schema Phase 3a: Client-Facing Deliverable (proposal/deck) Phase 3b: Internal Technical Package (schemas/prompts) Phase 4: Review both tracks against each other for consistency ``` --- ## End of Templates **Return to your task after reading the template(s) you need. Do not keep this file in active context.**