append-planning-log

name	append-planning-log
description	Append an approved Emmy planning-session entry to the Goals Space Planning Log in Confluence. Use only when the user explicitly authorizes recording a planning session after planning context was read; never use for ambient lookup, ticket creation, or editing goals, outcomes, or principles.

Append Planning Log

Use this skill to append one authorized planning-session audit entry to the Emmy Goals Space Planning Log. This is the write boundary for the planning context plugin.

Source Of Truth

• Planning Log page ID: 1398768357
• Planning Log URL: https://confluenceent.cms.gov/spaces/SFIV/pages/1398768357
• Confluence space: SFIV
• MCP server: cms-atlassian-confluence
• Required user environment variable: CONFLUENCE_PERSONAL_TOKEN

Preconditions

Before writing, confirm all of these are true:

1. The user explicitly authorized appending a Planning Log entry.
2. The caller provides a complete PlanningLogEntry draft.
3. The entry describes a real planning session that read the Goals Space.
4. The write target is page 1398768357.

If any precondition is missing, stop and ask for the missing information. Do not invent log-entry fields.

PlanningLogEntry Shape

Require this input object:

{
  "date": "YYYY-MM-DD",
  "session_type": "Session type",
  "goal_or_principle_ids": ["GOAL-001"],
  "agent_read": ["1398768351", "1398342919"],
  "trigger": "What prompted the planning session",
  "outcomes_reviewed": ["GOAL-001-O1"],
  "requirements_reviewed": ["REQ-ATO-001"],
  "tickets_proposed": "0 - held",
  "tickets_created": ["FFS-123"],
  "tickets_held": ["Held ticket summary and reason"],
  "human_decisions": ["Approved logging the session"],
  "notes": "Additional context"
}

Use none or an empty array only when the human-approved draft explicitly says there were no values.

requirements_reviewed is optional for backward compatibility. If provided and the live Planning Log template does not have a dedicated field, include the requirement IDs in agent_read or notes rather than changing the page template.

Append Sequence

1. Re-read Planning Log page 1398768357 immediately before writing.
   • Include metadata so the current page version is known.
   • Use storage format when needed to preserve existing structure.
2. Verify the page title is Planning Log.
3. Find the Log Entries section and the final append-only notice.
4. Format the new entry exactly like the page template:

## YYYY-MM-DD - [Session Type] - [Goal or Principle ID(s)]

**Agent read:** [Pages consulted: KS sections, Goal pages, Jira epics]
**Trigger:** [What prompted this planning session]
**Outcomes reviewed:** [Outcome IDs considered]
**Tickets proposed:** [N] - [created / held / deferred]
**Tickets created:** [Jira ticket keys, or "none"]
**Tickets held:** [Summary of held tickets and reason]
**Human decisions:** [Any approvals, rejections, modifications, or deferrals]
**Notes:** [Anything else relevant to future planning sessions]

When requirement IDs shaped the session, preserve them in the formatted entry. Use an existing field such as Agent read or Notes unless the live page already supports a Requirements reviewed field.

After formatting the entry:

1. Append the new entry after existing log entries and before the final append-only notice when that notice is present.
2. Preserve all prior entries and unrelated page content exactly.
3. Update the page with the smallest possible body change.
4. Re-read the page or fetch the resulting diff.
5. Confirm the new heading and core fields are present.
6. Report the changed page title, page ID, URL, and validation result.

Safety Rules

• No silent writes. User authorization must be explicit in the current task.
• Do not edit goals, principles, outcomes, or any page other than Planning Log.
• Do not edit, reorder, summarize, or delete previous log entries.
• Do not append if the page structure is ambiguous or the append location cannot be identified.
• Do not append if required entry fields are missing.
• Do not include secrets, credentials, private personal data, local scratch paths, or speculative claims.
• If the page changed after the first read and before update, compare the latest content. Merge only when the change is clearly independent; otherwise stop and report the conflict.
• Do not call Jira tools or create tickets from this skill, even though the shared MCP config may expose them.
• Do not update the Outcomes Index from this skill.

MCP Tool Use

Use Confluence MCP tools such as confluence_get_page, confluence_get_page_history, confluence_get_page_diff, and confluence_update_page for the append workflow.

If Confluence authentication fails, stop and tell the user that this plugin needs a valid CONFLUENCE_PERSONAL_TOKEN in the agent environment. Do not ask the user to paste tokens into chat.