shelf-feedback

shelf-feedback — Read and Process Feedback

Surface any feedback notes left in the Obsidian project dashboard's ## Feedback section. Display each item with a suggested action, then archive processed items to ## Feedback Log with a timestamp.

User Input

$ARGUMENTS

No arguments required.

Step 1: Resolve Project Identity (FR-005, FR-006)

Determine the project slug and base path. Priority order: explicit argument > .shelf-config > git remote defaults.

1. If .shelf-config exists in the repo root: a. Parse it: skip lines starting with # (comments) and blank lines; split each remaining line on the first = to get key and value; trim whitespace from both b. Extract base_path and slug values c. If both are present and non-empty: use them as $BASE_PATH and $SLUG — do NOT derive from git remote or prompt the user (FR-006). Skip to substep 4 d. If either is missing or empty: warn ".shelf-config is malformed — missing {key}. Falling back to defaults." and continue to substep 2
2. If no valid .shelf-config: a. If the user provided a project name as an argument: use it as $SLUG b. Otherwise: run git remote get-url origin and extract the repo name (last path segment, strip .git suffix) as $SLUG c. Set $BASE_PATH = "projects" (default)
3. All vault paths use: {$BASE_PATH}/{$SLUG}/...

Step 2: Read Project Dashboard (FR-022)

mcp__obsidian-projects__read_file({ path: "{base_path}/{slug}/{slug}.md" })

• If not found: suggest "No project found — run /shelf-create first" and STOP (FR-022)
• If MCP fails: warn "MCP server unavailable — cannot read feedback" and STOP (NFR-004)

Store the full dashboard content for later update.

Step 3: Extract Feedback Section (FR-019)

Parse the dashboard content and extract everything between ## Feedback and the next ## heading (which should be ## Feedback Log).

• Each feedback item is a line starting with - (a markdown list item)
• Trim whitespace, ignore empty lines

Step 4: Check for Empty Feedback (FR-023)

If no feedback items found:

• Report: "No feedback items found in the Obsidian dashboard."
• STOP — no further action needed

Step 5: Display Feedback and Suggest Actions (FR-020)

For each feedback item, categorize it and suggest an action:

Pattern	Category	Suggested Action
Contains "fix", "bug", "broken", "error"	Fix request	"Run /fix {issue} or /shelf-sync to track as an issue"
Contains "add", "feature", "want", "need"	Feature request	"Consider adding to backlog via /report-issue"
Contains "scope", "plan", "rethink", "consider"	Planning note	"Note for next /specify or /plan session"
Contains "urgent", "asap", "blocker"	Urgent	"Priority item — address before continuing"
Other	General note	"Informational — noted for this session"

Display format:

## Feedback from Obsidian

1. **{category}**: {feedback text}
   Suggested action: {action}

2. **{category}**: {feedback text}
   Suggested action: {action}

...

Step 6: Archive to Feedback Log (FR-021, FR-005)

After displaying all items, move them from ## Feedback to ## Feedback Log:

1. Get the current timestamp: [YYYY-MM-DD HH:MM]
2. For each feedback item, prepend the timestamp: - [YYYY-MM-DD HH:MM] {item text}
3. Rebuild the dashboard content:
   • Clear the ## Feedback section (leave the heading, remove all items)
   • Append the timestamped items to the ## Feedback Log section
   • Preserve all other sections unchanged (frontmatter, Human Needed, etc.)
   • Preserve the project: "[[{slug}]]" backlink and tags: in frontmatter (FR-005) — do not remove these fields when rewriting

mcp__obsidian-projects__update_file({
  path: "{base_path}/{slug}/{slug}.md",
  content: "{rebuilt dashboard with feedback archived}"
})

If MCP fails: warn "Could not archive feedback — items will reappear next time" (NFR-004)

Step 7: Report Results

Processed {N} feedback item(s).

  Archived to Feedback Log with timestamp [{YYYY-MM-DD HH:MM}].
  {if any urgent items: "URGENT items noted above — address before continuing."}

Rules

• All writes go through MCP — never write directly to the filesystem (NFR-001)
• No hardcoded vault paths — always use the resolved base path (NFR-002, NFR-003)
• Graceful degradation — if MCP is unavailable, warn and exit cleanly (NFR-004)
• No secrets in notes — do not include API keys, tokens, or credentials (NFR-005)
• Read-then-write — always read the full dashboard before updating to avoid clobbering
• Preserve sections — only modify Feedback and Feedback Log; leave everything else intact