emmy-planning-ingest

name	emmy-planning-ingest
description	Ingest Emmy Goals Space principles and goals from pasted text or local files into Confluence with a human approval gate. Use only when explicitly asked to import, ingest, add, or update Emmy planning principles, goals, outcomes, or "principals"; never use for ambient planning lookup, Jira ticket creation, or Knowledge Store ingestion.

Emmy Planning Ingest

Use this skill to turn human-provided planning material into normalized Emmy Goals Space drafts, then write the approved plan to Confluence only after explicit human authorization.

Treat principals as principles when the request clearly refers to planning principles.

Source Of Truth

• Goals Space root: 1398768351
• Strategic Principles index: 1398768358
• Goals Registry index: 1398342919
• Outcomes Index: 1399162264
• Planning Log: 1398768357
• Confluence space: SFIV
• MCP server: cms-atlassian-confluence
• Required user environment variable: CONFLUENCE_PERSONAL_TOKEN
• The shared plugin MCP config may expose Jira URL, token, and tool settings for collision-safe overlap with Jira plugins. This skill uses Confluence only.

This skill writes only to the Emmy Goals Space. Do not write to the Emmy Agent Knowledge Store. Do not call Jira tools from this skill, even though the shared MCP config may expose them.

References

• Read references/write-safety.md before any approved Confluence write.
• Read references/dry-run-fixtures.md when testing the workflow, doing a draft-only run, or comparing an ingest review packet to expected output.

Supported Inputs

Accept only these v1 source types:

• Pasted text in the conversation.
• Local files the user names or attaches, such as Markdown, text, PDF, Word, or other readable planning documents.

Do not ingest URLs, external Confluence pages, repository checkouts, Jira issues, or Knowledge Store pages in v1. If the user provides an unsupported source, stop and explain that v1 supports pasted text and local files only.

Use this input shape when the caller provides structured values:

{
  "source_text": "optional pasted planning text",
  "source_files": ["optional/path.md"],
  "ingest_types": "principles | goals | both",
  "requested_status": "Active | Deferred",
  "human_authorization_required": true
}

Default ingest_types to both when the source clearly contains both principles and goals. Default requested_status to Active only when the source or human says the item should be active; otherwise ask.

Read First

Before drafting or writing:

1. Read Goals Space root 1398768351 and verify it is initialized.
2. Read Strategic Principles index 1398768358.
3. Read Goals Registry index 1398342919.
4. Read Outcomes Index 1399162264.
5. Read existing principle and goal child pages as needed for duplicate detection and ID allocation.

If a required page is missing or unreadable, stop and report the page ID. Do not rely on cached content or memory.

Source Handling

• For pasted text, cite the source as human-provided text with the ingest date.
• For local files, inspect the file locally, extract relevant text, and record filename, file type, and date reviewed.
• Do not persist local absolute paths to Confluence.
• Do not upload local files or attachments unless the user explicitly asks.
• Stop before writing if the source appears to contain secrets, credentials, private personal data, unsupported binary content, or speculative content that is not suitable for the Goals Space.
• Do not normalize requirement-like material into principles or goals. If the source includes obligations, compliance gates, blockers, due dates, control statements, approval restrictions, or evidence expectations, flag REQUIREMENT_CANDIDATE and route that material to emmy-requirements-ingest. Requirement ingestion uses emmy-requirements-interpret as its interpretation step.

Draft Objects

Normalize source material into these draft objects.

PrincipleDraft

{
  "id": "PRIN-001",
  "name": "Principle name",
  "status": "Active",
  "owner": "Owner or team",
  "grounded_in": "Policy, legislation, CMS standard, or program mandate",
  "last_reviewed": "YYYY-MM-DD",
  "principle_statement": "One to three sentences",
  "why_this_matters": "Strategic context",
  "practice_implications": [
    {
      "area": "Area",
      "implication": "Concrete implication"
    }
  ],
  "evaluation_criteria": ["Falsifiable question?"],
  "known_tensions": ["Documented tension"],
  "related_sources": ["Source or link"]
}

GoalDraft

{
  "id": "GOAL-001",
  "name": "Goal name",
  "statement": "Outcome-oriented goal statement",
  "status": "Active",
  "owner": "Owner or team",
  "horizon": "Date or quarter",
  "last_reviewed": "YYYY-MM-DD",
  "why_this_matters": "Strategic context",
  "outcomes": [
    {
      "id": "GOAL-001-O1",
      "statement": "Outcome statement",
      "success_criteria": "Falsifiable success condition",
      "status": "Not Started",
      "jira_epic": null
    }
  ],
  "open_blockers": [],
  "knowledge_store_links": []
}

Use - for optional text fields only when the source and human approval make clear the value is intentionally empty. Otherwise ask for the missing value.

Requirement Candidate Handling

Requirements are planning constraints, not principles or goals. When source text contains requirement signals such as must, shall, required, cannot, approval required, ATO, CFACTS, PIA, POA&M, control, due, or expires, do all of the following:

1. Extract only the true principle or goal material into PrincipleDraft and GoalDraft records.
2. Add warning code REQUIREMENT_CANDIDATE with the relevant source excerpt or summary.
3. State that requirement ingestion belongs to emmy-requirements-ingest, with emmy-requirements-interpret used before any requirement draft is persisted.
4. Do not write requirement text into principle statements, goal statements, outcomes, or blockers unless the human explicitly rewrites it as planning intent.

ID And Duplicate Rules

• Allocate new principle IDs as PRIN-### from the current maximum.
• Allocate new goal IDs as GOAL-### from the current maximum.
• Allocate new outcome IDs as GOAL-###-O# within each goal.
• Honor user-provided IDs only when they do not conflict with existing pages, index rows, or draft objects.
• Treat near-duplicate principles, goals, or outcomes as review blockers. Present the overlap and wait for human direction.
• Check Outcomes Index before approving new outcomes. Exact duplicates must not be written as new outcomes.

Governance Rules

• Active principles and active goals require explicit human approval in the current conversation before any Confluence write.
• Do not infer leadership intent. If the source does not clearly identify the strategic reason, owner, horizon, status, or success criteria, draft the item with warnings and stop before writing.
• Deferred items may be drafted from source material, but never promote a deferred or ambiguous item to Active by inference.
• Ask the human to resolve conflicts between source material and existing Goals Space entries before writing.
• Treat approval to draft, normalize, or review as different from approval to write. Only approval to write authorizes Confluence changes.

Review Packet

Before any write, present a compact review packet containing:

• Proposed principle and goal drafts.
• Proposed child pages to create or update.
• Proposed Strategic Principles and Goals Registry index row changes.
• Proposed Outcomes Index additions or updates.
• Source provenance.
• Duplicate and quality warnings.
• Requirement candidate warnings and routing notes.
• Exact page IDs to be written.
• A Planning Log entry draft.

Ask for explicit approval of the whole write plan. Do not write if approval is partial, ambiguous, or missing.

Approved Write Sequence

After explicit approval:

Read references/write-safety.md, then:

1. Re-read every page that will be written, including current metadata/version.
2. Create approved principle child pages under Strategic Principles 1398768358, or narrowly update approved existing principle pages.
3. Create approved goal child pages under Goals Registry 1398342919, or narrowly update approved existing goal pages.
4. Update Strategic Principles and Goals Registry index tables narrowly.
   • Remove placeholder rows such as None yet or em dash rows only when adding the first real row.
   • Update existing rows only when the approved plan explicitly changes them.
5. Update Outcomes Index 1399162264 so every approved goal outcome appears in the flat index.
6. Preserve unrelated content and all existing rows not touched by the plan. Add or update Change Log rows on changed child pages.
7. Re-read changed pages or fetch diffs and validate the intended entries are present.
8. If the user approved the Planning Log entry, append it through the append-planning-log workflow.
9. Report created or updated page titles, page IDs, URLs, and validation results.

Stop rather than guessing if a page changed in a conflicting way, if the table shape is not recognized, or if the append/create/update location is ambiguous.

Page Templates

Use the live Confluence templates from the index pages as the structure for new child pages. Principle pages must include:

• Status, Owner, Grounded in, Last reviewed.
• Principle Statement.
• Why This Matters.
• What This Means in Practice table.
• Agent Evaluation Criteria.
• Known Tensions.
• Related Policies / Sources.
• Change Log.

Goal pages must include:

• Status, Owner, Horizon, Last reviewed.
• Goal Statement.
• Why This Matters.
• Outcomes table.
• Open Blockers.
• Knowledge Store Links.
• Change Log.

Constraints

• No silent writes.
• No Jira calls and no ticket creation.
• No Knowledge Store writes or reads.
• No unsupported source ingestion.
• No requirements ingestion. Requirement-like source material should be flagged and routed to emmy-requirements-ingest.
• No edits to unrelated Goals Space content.
• No uploads of local files unless explicitly requested.
• Always cite pages read and pages written in the final response.