Session Report

Update the session log for today's work. This is the report-writing half of the session lifecycle — it records what happened. No build gating and no worktree finalization. /session-end invokes this skill as one of its phases; you can also run it standalone to checkpoint mid-session.

Project configuration

The manifest is this skill's configuration. If $(git rev-parse --show-toplevel)/.claude/lifecycle-manifest.md does not exist, the project has not adopted the kit: say so, offer to create it, and stop. Never infer the keys and run anyway. The kit ships the template as lifecycle-manifest.template.md; locate the installed copy with find ~/.claude/plugins -path '*lifecycle-kit*' -name lifecycle-manifest.template.md | head -1. The marketplace's ADOPTING.md carries a repo-inspecting setup prompt.

Read .claude/lifecycle-manifest.md first (resolve via $(git rev-parse --show-toplevel)/.claude/lifecycle-manifest.md). Where a step names a manifest key in code font (log_dir, log_pattern, branch_pattern, docs_log_branch, build_commands, test_commands), use that key's value. The suffix examples (_audio, _api) are illustrative.

1. Determine the session number N and the log filename from the branch, never by minting a new number. /session-start is the sole authority that mints N (it bakes it into the branch name at branch birth); this skill only consumes it.
   • Normal case, on a branch_pattern branch (feature/session-<N>-<topic> for Untype): N is that number and <suffix> is the topic slug (or a _audio/_api-style workstream suffix when several agents log in parallel). Parse both from git branch --show-current.
   • Log-only exception, on a docs_log_branch (docs/session-<N>-log) branch: N is that number, no suffix.
   • Standalone on main with no session branch (rare): do not invent a number from filenames. If today already has a session log, update it; otherwise ask the caller for N.
   • The log file follows log_pattern under log_dir (sessions/YYYY-MM-DD_session_<N>[_<suffix>].md for Untype) for the current date. The identity is the pair (N, suffix): a suffixed log and a bare-N log are distinct files, so worktree agents each get their own log and never collide. If it exists, update it (step 5); if not, create it from the pattern in the most recent existing log.
2. Read the current session file and the previous one for format reference.
3. Update the session file with:
   • ## Context — 1-paragraph framing
   • ## What happened — numbered sections with details
   • ## Files changed — table: file | change
   • ## Build status — the build_commands/test_commands result. If the caller (e.g. /session-end) passed pre-computed output via $ARGUMENTS, fold it in verbatim rather than re-running the commands. Otherwise mark as "not re-run this session" — do not run build/test from this skill.
   • ## Decisions worth remembering — non-obvious calls made
   • ## What's next — immediate + deferred
   • ## Commits — this session's commits as <short-sha> <title>, newest last; /session-start reads the last entry as its "shipped since" anchor, so keep it current
   • The header block at the top, in the format /session-start parses: an H1 # Session <N>: <theme> title, then **Date**:, **Branch**: (the branch_pattern branch, feature/session-<N>-<topic> for Untype, with worktree path if any), and **Type**: lines
4. Keep the format consistent with existing session logs in log_dir.
5. Do not remove existing content — append or update sections as needed.
6. No emojis.

Arguments

$ARGUMENTS is optional. If the caller (typically /session-end) passes pre-computed build/test output, fold it into ## Build status verbatim.