# Template Improvement Directives
# Generated: [AI_DATE_GENERATION_PROHIBITED]
# Source Project: ITPR (Reset)
directives:
- directive_id: "TID001"
target_template_id: "ProjectStateSchema.yaml" # And by implication, all templates using/generating dates
target_section_or_field: "All date/timestamp fields (e.g., metadata.last_modified_timestamp, charter.date_formalized, logs.timestamp, execution.tasks.start_timestamp, etc.)"
issue_description: "Current use of '[AI_DATE_GENERATION_PROHIBITED]' placeholder for AI-unsourceable dates is problematic, conspicuous, and clutters outputs. AI should not invent dates."
proposed_change_type: "ModifyFieldSchemaAndGuidance"
proposed_change_details: |
1. Review all date/timestamp fields in ProjectStateSchema.
2. Make fields optional wherever feasible if they are not critical for process logic or user-provided.
3. For fields that *must* exist per schema but AI cannot populate:
- Option A: Allow them to be truly empty/null if the schema validator permits.
- Option B: Introduce a standard, less obtrusive way to indicate "AI Cannot Source" (e.g., a specific keyword like `UNSOURCEABLE_AI_TIMESTAMP` or a related boolean field `is_timestamp_ai_generated: false`).
4. Update all process templates (00-05) to instruct AI on how to handle these fields according to the revised schema guidance (e.g., "Omit if optional and unsourceable," or "Use designated null/unsourceable marker").
rationale: "Improves clarity, reduces clutter, adheres to 'AI does not invent dates' principle. (Ref: INS001)"
source_insight_refs: ["INS001"]
status: "Proposed"
- directive_id: "TID002"
target_template_id: "All Process Templates (00-Explore.md to 05-Close.md)"
target_section_or_field: "Instructions related to file saving, naming, and directory structure."
issue_description: "Current practices or lack of clear instruction lead to inconsistencies or overly complex file management (e.g., versions in filenames, nested directories)."
proposed_change_type: "UpdateInstruction"
proposed_change_details: |
1. Standardize instructions for saving project artifacts (Charters, Plans, State files, KAs, Deliverables):
- Filenames should NOT include version numbers (e.g., `[ProjectCode]_Plan`, not `[ProjectCode]_Plan_vX.Y`). Internal `version` field in YAML tracks this.
- Default save location should be the main project directory (e.g., `projects/[ProjectCode]/`) unless a specific, simple subdirectory (e.g., `outputs/` for final human-readable deliverables) is explicitly agreed upon and consistently used. Avoid deep nesting.
2. State files should use sequential numbering (e.g., `[ProjectCode]_State_001`).
rationale: "Simplifies file management, aligns with user preference, leverages external version control. (Ref: INS_FilenameNoVersion, INS001)"
source_insight_refs: ["INS_FilenameNoVersion", "INS001"]
status: "Proposed"
- directive_id: "TID003"
target_template_id: "All Process Templates, AISkillsCatalog.yaml (for AI persona definition), ITPR_StyleGuide (as a model)"
target_section_or_field: "AI communication style, instructional text within templates, AI self-critique guidelines."
issue_description: "AI responses sometimes include conversational filler, hedging, or inappropriate evaluative language, deviating from the desired direct, assertive, matter-of-fact machine voice."
proposed_change_type: "UpdateGuidanceAndInternalProtocol"
proposed_change_details: |
1. All instructional text within process templates written for the AI should model direct, concise language.
2. AI operational protocols (potentially part of AISkillsCatalog or a new AI Persona Definition KA) should mandate:
- Avoidance of conversational filler, apologies, personal opinions (Ref: INS007 - "Don't say it, do it!").
- Avoidance of superfluous laudatory/emotive adjectives (e.g., "profound," "remarkable") when describing concepts. Significance should be shown, not asserted adjectivally. (Ref: User feedback on D001 drafts).
- Strict protocol for identifying and flagging/clarifying any hedging or "waffle words" on core assertions with the user *before* inclusion in formal drafts. (Ref: INS010).
3. `Meta-RefineOutput.md` could be updated to include checks for these stylistic elements during AI self-critique.
rationale: "Ensures consistent AI persona, improves clarity and impact of AI outputs, aligns with user's stylistic preferences. (Ref: INS004, INS007, INS010, specific draft feedback)"
source_insight_refs: ["INS004", "INS007", "INS010"]
status: "Proposed"
- directive_id: "TID004"
target_template_id: "05-Close.md, ProjectStateSchema.yaml (KA structure)"
target_section_or_field: "KA handling, project archival."
issue_description: "Knowledge Artifacts (Glossary, Primer, Style Guide) have value as standalone documents beyond the project, but current process primarily embeds them in Project State YAML."
proposed_change_type: "AddNewFeatureOrGuidance"
proposed_change_details: |
1. `05-Close.md` (Project Closing) should include a step to prompt the user if they wish to export key KAs as separate, formatted standalone documents (e.g., Markdown, PDF).
2. AI should have a skill or process for formatting and outputting these KAs cleanly.
3. Consider if `ProjectStateSchema.yaml` needs a field in KA objects to store a "standalone formatted version" or a path to it, if generated.
rationale: "Enhances reusability and value of generated KAs. (Ref: INS_KAReusability)"
source_insight_refs: ["INS_KAReusability"]
status: "Proposed"
- directive_id: "TID005"
target_template_id: "04-Monitor.md, 05-Close.md, ProjectStateSchema.yaml (new schema for feedback)"
target_section_or_field: "Process improvement feedback loop."
issue_description: "Current method of logging insights for template improvement is ad-hoc. A structured, machine-readable format is needed for systematic template evolution."
proposed_change_type: "DefineNewSchemaAndIntegrate"
proposed_change_details: |
1. Develop a `TemplateFeedbackSchema.yaml` (as discussed, with fields like `feedback_item_id`, `target_template_id`, `target_section_or_field`, `issue_description`, `proposed_change_type`, `proposed_change_details`, `rationale`, `status`).
2. `04-Monitor.md` should include a step to review logged insights and translate relevant ones into instances of this `TemplateFeedbackSchema`.
3. `05-Close.md` should include a step to consolidate all `TemplateFeedbackSchema` instances from the project as a key output for the "ProcessMaintainer/AI."
4. `ProjectStateSchema.yaml` should include a section to store these structured feedback items (e.g., `metadata.template_improvement_directives: list_of_TemplateFeedbackSchema_instances`).
rationale: "Enables systematic, traceable, and potentially automatable continuous improvement of all project templates and processes. (Ref: INS005)"
source_insight_refs: ["INS005"]
status: "Proposed"
- directive_id: "TID006"
target_template_id: "All Process Templates (especially transition points like end of 01-Initiate, 02-Plan, 04-Monitor)"
target_section_or_field: "Logic for transitioning to next process template / requesting templates."
issue_description: "AI sometimes incorrectly requests templates already loaded or fails to state when a template is missing."
proposed_change_type: "UpdateProceduralLogic"
proposed_change_details: |
At any point where a transition to a new process template is indicated:
1. AI must first check its active context if the required template (and version, if specified) is already loaded.
2. If loaded, AI states: "Proceeding with loaded [template_name] vX.Y."
3. If not loaded, AI states: "I do not have [template_name] vX.Y in my current context. Please provide it to proceed."
rationale: "Improves AI operational efficiency, reduces redundant user actions, ensures correct template usage. (Ref: INS006)"
source_insight_refs: ["INS006"]
status: "Proposed"
- directive_id: "TID007"
target_template_id: "03-Execute.md, InvokeAISkill.md (or similar skill execution template)"
target_section_or_field: "Handling and presentation of large text deliverables."
issue_description: "AI has repeatedly failed to output large text deliverables (e.g., full drafts) completely and directly for user review, causing critical process blocks."
proposed_change_type: "UpdateProceduralLogicAndOutputStrategy"
proposed_change_details: |
1. When an AI skill or process step in `03-Execute.md` generates a large text deliverable intended for immediate user review:
- AI must explicitly state its strategy for outputting this large text (e.g., "This will be a long output. I will provide it chapter by chapter.").
- AI must output the text in clearly labeled, sequential parts.
- AI must await user acknowledgement (e.g., "continue," "next") after each part before sending the next.
- AI must confirm when the entire output is complete.
2. The primary review should occur on this directly provided text, not by requiring the user to extract it from a state file. The state file serves as the archive *after* review or alongside it.
3. `InvokeAISkill.md` (if it handles output presentation) or the relevant steps in `03-Execute.md` must incorporate this logic.
rationale: "Ensures user receives and can review critical deliverables effectively, preventing project stalls and improving trust in AI output. (Ref: INS009, ISS001_RepeatFailure_OutputTruncation)"
source_insight_refs: ["INS009"]
status: "Proposed"
- directive_id: "TID008" # New, based on our very recent discussion
target_template_id: "00-Explore.md, 02-Plan.md"
target_section_or_field: "Project Scoping, Argument Structure Development, WBS generation."
issue_description: "Complex, foundational projects may require iterative refinement of core concepts and argument structure *before* a stable plan or full draft can be developed, sometimes necessitating a 'reset' or loop back to exploration/initiation."
proposed_change_type: "AddGuidanceAndProcessFlexibility"
proposed_change_details: |
1. `00-Explore.md` and `01-Initiate.md` should acknowledge that for highly novel or complex conceptual projects, the initial exploration might lead to a plan that later proves insufficient, requiring a "Phase 0 Reset" where the accumulated learning becomes input for a *new* exploration and initiation cycle (as occurred in ITPR). Templates should allow for this iterative deepening.
2. `02-Plan.md` should emphasize that the WBS for such projects, especially for deliverables like D001 (monograph), might need to be very high-level initially, with detailed task breakdown for early conceptual chapters/parts, and later parts remaining more flexible until core concepts are solidified.
3. The concept of "Guiding Paradoxes/Spectra" and "Argumentative Pillars" as outputs of `00-Explore.md` should be considered for inclusion as structured fields in `project_state.exploration_history` to better capture foundational strategic decisions.
rationale: "Better accommodates the iterative and sometimes recursive nature of deep conceptual work and foundational theory development. Formalizes successful ad-hoc process adjustments made in ITPR. (Ref: ITPR project reset experience)"
source_insight_refs: ["ITPR Project Reset Event"]
status: "Proposed"