log · diff
git:20260610.be2a852 to git:20260610.c15894f
181 added, 4 removed. Audit A to A.
---
name: log
description: "Capture events — links, calls, meetings, ideas — into an append-only log. Bulk mode for WhatsApp dumps. Use when something happened."
effort: low
keywords: [logging, events, capture, meetings, links, observations, whatsapp, bulk]
task_strategies: [feature, spike]
stream_affinity: [roadmap, docs]
argument-hint: "[event text or bulk]"
group: capture
model: haiku
allowed-tools:
- Read
- Write
- Edit
- Glob
- Grep
- Bash
- AskUserQuestion
status: stable
growth_stage: evergreen
---
+ # Log — Event Capture
- <!-- PROCEDURE_FILE: procedures/log.md -->
- This skill's full procedure is in a separate file for startup performance (ADR-034).
- Read and execute `../../procedures/log.md` resolved against this skill's base directory (the path announced when the skill loads) — i.e. `{base-dir}/../../procedures/log.md`. This form is valid in both the repo layout and the deployed-plugin layout.
- If the path doesn't resolve, use Glob to find `**/procedures/log.md`.
+ Append-only event log. The lowest-friction entry point into brana — no need to know tasks.json schema, pipeline stages, or memory conventions. Just `/brana:log "something happened"`.
+
+ ## When to use
+
+ - Something happened (call, meeting, idea, observation) and you want to record it
+ - You received links or references to capture for later
+ - Pasting a WhatsApp or chat dump to triage
+ - Anything that doesn't fit `/brana:backlog add` (not a commitment yet) or `/brana:pipeline` (not a qualified lead yet)
+
+ ## Commands
+
+ ```
+ /brana:log "text with optional #tags" — quick append
+ /brana:log bulk — paste multi-line content, parse + deduplicate
+ ```
+
+ ## File
+
+ Per-project log with global fallback.
+
+ ### Resolution order
+
+ 1. **CWD project**: `git rev-parse --show-toplevel` → basename → find matching CC project memory dir → `{dir}/event-log.md`
+ 2. **Tag routing**: if entry has `#projectslug` matching a registered project in `tasks-portfolio.json`, route to that project's log
+ 3. **Global fallback**: `~/.claude/memory/event-log.md` — when no project context is detected
+
+ ### Finding the CC project memory dir
+
+ ```
+ ~/.claude/projects/-{sanitized-path}/memory/event-log.md
+ ```
+
+ Where `sanitized-path` is the absolute project path with `/` replaced by `-` and leading `-`. Example: project at `/home/user/enter_thebrana/thebrana` → `~/.claude/projects/-home-user-enter-thebrana-thebrana/memory/event-log.md`.
+
+ Create the file on first use if it doesn't exist.
+
+ ---
+
+ ## /brana:log "text"
+
+ Quick append — the default mode.
+
+ ### Steps
+
+ 1. **Parse input.** Extract the quoted text from the argument.
+
+ 2. **Extract tags.** Find all `#word` tokens in the text. Remove the `#` prefix and collect as tags. Leave the `#tag` inline in the entry text — tags are visible, not hidden metadata.
+
+ 3. **Resolve the log file** using the resolution order above (CWD project → tag routing → global fallback). Read it.
+ - If it doesn't exist, create it with a header:
+ ```markdown
+ # Event Log
+ ```
+
+ 4. **Find or create today's section.** Look for a `## YYYY-MM-DD` heading matching today's date.
+ - If it exists, append after the last entry under that heading.
+ - If it doesn't exist, append a new section at the bottom of the file:
+ ```markdown
+
+ ## YYYY-MM-DD
+ ```
+
+ 5. **Format the entry.** Use this template:
+ ```
+ - HH:MM — {text}
+ ```
+ Timestamp is current time (24h format). The text is verbatim from the user's input (including inline #tags).
+
+ 6. **URL detection.** If the text contains one or more `https://` URLs:
+ - Read the current project's `.claude/tasks.json` (if it exists) and the portfolio's tasks files
+ - Check if any URL already exists in a research task's description, context, or notes
+ - For each **new** URL (not already in any task):
+ - Use AskUserQuestion: "Found {N} new URL(s). Create research tasks?"
+ - Options: "Yes — create tasks (Recommended)", "No — just log"
+ - If yes: for each URL, run `/brana:backlog add` with stream=research, the URL in context, and tags from the log entry
+ - For URLs that already exist in tasks: note "(already tracked as {task-id})" silently in the log entry
+
+ 7. **Write the entry** using the Edit tool (append to the day section).
+
+ 8. **Archival check.** Count total lines in the file.
+ - If >500 lines: identify entries older than 90 days
+ - Move them to `~/.claude/memory/event-log-archive-YYYY.md` (year of the oldest moved entry)
+ - Use AskUserQuestion to confirm before archiving: "Log has {N} lines. Archive {M} entries older than 90 days?"
+
+ 9. **Report.** Confirm the entry was logged:
+ ```
+ Logged: "text" [#tag1, #tag2] at HH:MM
+ ```
+
+ ### Example
+
+ User: `/brana:log "Call with Juan from Kapso — interested in automation for their onboarding flow #somos #call"`
+
+ Result in `~/.claude/memory/event-log.md`:
+ ```markdown
+ ## 2026-03-07
+
+ - 14:32 — Call with Juan from Kapso — interested in automation for their onboarding flow #somos #call
+ ```
+
+ Report: `Logged: "Call with Juan from Kapso..." [#somos, #call] at 14:32`
+
+ ---
+
+ ## /brana:log bulk
+
+ Paste and parse multiple entries at once — designed for WhatsApp dumps, meeting notes, or batched captures.
+
+ ### Steps
+
+ 1. **Prompt for content.** Use AskUserQuestion:
+ ```
+ question: "Paste the content to log (WhatsApp dump, meeting notes, etc.)"
+ options: ["Ready — I'll paste below"]
+ allowFreeText: true
+ ```
+
+ The user pastes multi-line text.
+
+ 2. **Parse entries.** Split on line boundaries. For each line:
+ - Strip WhatsApp metadata (timestamps like `[DD/MM/YY, HH:MM:SS]`, sender names before `:`)
+ - Strip empty lines and media placeholders (`<Media omitted>`, `image omitted`, etc.)
+ - Keep the content text
+ - Extract any inline `#tags`
+ - Detect URLs
+
+ 3. **Deduplicate.** Read the existing log file. For each parsed entry:
+ - Check if the same text (fuzzy: ignore whitespace, case) already exists in the log
+ - Mark duplicates for exclusion
+
+ 4. **URL cross-reference.** For entries containing URLs:
+ - Read tasks.json files (current project + portfolio)
+ - Check if each URL already exists in a research task
+ - Mark as "already tracked" or "new"
+
+ 5. **Present for confirmation.** Use AskUserQuestion with a summary:
+ ```
+ question: "Parsed {total} entries from your paste. {new} new, {dupes} duplicates (skipped), {urls_new} new URLs."
+ options: ["Confirm — log all new entries", "Edit — let me adjust", "Cancel"]
+ ```
+
+ If "Edit": ask which entries to exclude or modify, then re-present.
+
+ 6. **Write entries.** For each confirmed new entry:
+ - Format as `- HH:MM — {text}` under today's date section
+ - Use the original timestamp from the paste if available, otherwise use current time
+
+ 7. **URL task creation.** If new URLs were found:
+ - Use AskUserQuestion: "Create research tasks for {N} new URLs?"
+ - Options: list each URL with a checkbox-style selection
+ - For confirmed URLs: run `/brana:backlog add` with stream=research
+
+ 8. **Report.**
+ ```
+ Bulk log complete:
+ {N} entries logged
+ {M} duplicates skipped
+ {K} research tasks created
+ ```
+
+ ---
+
+ ## Rules
+
+ - **Append-only.** Never edit or delete existing entries. The log is a historical record.
+ - **CWD first, tags second.** Route by project (CWD), fall back to `#tag` matching, then global. Works from anywhere — global log catches entries with no project context.
+ - **No auto-detection.** Don't try to classify entries as "call", "meeting", "idea" automatically. The user adds `#call`, `#meeting`, `#idea` if they want to categorize.
+ - **Confirm URLs.** Never silently create tasks from URLs. Always ask first via AskUserQuestion.
+ - **Chronological within each day.** New entries go at the bottom of the day's section.
+ - **Per-project files.** Each project gets its own `event-log.md` in its CC memory dir. Tags still provide cross-project filtering (grep across all logs).
+ - **Archival is conservative.** Only archive entries >90 days old, only when the file exceeds 500 lines, and only after user confirms.
+
+ ## What /brana:log is NOT
+
+ - Not a replacement for `/brana:backlog add` — tasks are commitments, log entries are observations
+ - Not a replacement for MEMORY.md — memory stores patterns, log stores events
+ - Not a replacement for `/brana:pipeline` — pipeline tracks deals, log captures first contact
+ - Not a calendar, reminder system, or analytics tool
+
+ The log is an **inbox**. Other commands are the **outbox**. Capture fast, route later.