| name | note |
| description | This skill should be used when the user asks to "note a tldr improvement", "log a tldr issue", "capture a tldr idea", "jot down a tldr problem", "/tldr:note", or wants to quickly record a bug, feature request, or improvement idea for the TLDR plugin without disrupting their current workflow. Supports an optional --ex flag to include an example of the issue, which auto-triggers evaluation via /tldr:feedback. |
note
Quickly capture improvement ideas, bugs, or feature requests for the TLDR plugin without disrupting the current workflow.
Purpose
Observations about what needs fixing or improving often arise mid-task. Provide a fire-and-forget capture mechanism so nothing is lost. All entries are stored in a single catalog file for later review via /tldr:review.
Path Resolution
All data lives inside the TLDR plugin directory. This skill is at skills/note/SKILL.md โ the plugin root is two directories up (the directory containing skills/, docs/, and .claude-plugin/). All file paths below are relative to the plugin root.
- Notes catalog:
docs/evaluation/notes.md
- Evaluation log:
docs/evaluation/evaluation-log.md (used when --ex triggers auto-evaluation)
- Sample files:
docs/evaluation/samples/ (used when --ex triggers auto-evaluation)
Input Format
/tldr:note [context of the improvement/issue/idea]
/tldr:note [context] --ex [copy-paste example of the issue]
Capture Workflow
Step 1: Parse Input
Extract two components from the user's input:
- Context โ The description of the improvement, issue, or idea (everything before
--ex if present, or the entire input)
- Example (optional) โ Content after the
--ex flag, typically a copy-paste of TLDR output demonstrating the issue
Step 2: Expand the Note
From the user's raw context, generate an expanded interpretation:
- Restate the observation in clear, specific terms
- Identify the category:
bug, improvement, feature, or observation
- Suggest the likely impact on TLDR quality
- Keep the expansion brief (2-4 sentences)
Step 3: Append to Catalog
Read docs/evaluation/notes.md and append a new entry. Determine the next note number by reading existing entries โ zero-pad to 3 digits (e.g., Note 007).
After appending, update the header counters in notes.md:
- Increment Total Notes by 1
- Increment Pending by 1
Entry format:
---
### Note NNN
**Date:** YYYY-MM-DD
**Category:** [bug | improvement | feature | observation]
**Status:** pending
**Raw:** [User's original context text]
**Expanded:** [Agent's intuited expansion of the context]
**Example:** [If --ex was provided, the example content. Otherwise omit this field.]
Step 4: Handle --ex Flag (When Present)
When the user includes --ex [example]:
- Log the note as above (Steps 1-3)
- Auto-evaluate the example by following the
/tldr:feedback workflow steps inline with --no-user-score mode (do not prompt the user or invoke a separate command):
- Treat the example as a TLDR sample
- Score it on the 4 criteria (Completeness, Conciseness, Actionability, Accuracy)
- Create a sample file in
docs/evaluation/samples/
- Update
docs/evaluation/evaluation-log.md
- The note's context provides the "what went wrong" framing for the evaluation
- Cross-reference: Add the sample ID to the note entry, and reference the note ID in the sample file
Step 5: Confirm and Return
Provide a brief confirmation (2-3 lines max):
- Note ID and category
- One-line summary of what was captured
- If
--ex was used: mention the evaluation was logged
Then immediately return control to the user. Do NOT engage in discussion or ask follow-up questions. This is fire-and-forget.
Important Constraints
- Zero disruption โ Confirm in 2-3 lines, then stop. No follow-up questions.
- Preserve user flow โ The user invoked this mid-task. Do not derail their current work.
- Expand thoughtfully โ The agent expansion should add clarity, not just restate. Consider what a future reviewer would need to understand the note.
- Idempotent numbering โ Always read the catalog to determine the next note number. Never assume.