daily-standup-prep
Generate a per-team standup report from recent Jira, GitLab, Confluence, and Git activity, written into the Obsidian vault. Use when the user says "daily standup prep", "standup report", "generate sta
SKILL.mdskills/daily-standup-prep/SKILL.md
GitHubDaily Standup Prep
Per-team standup report. Read-only across Jira, GitLab, Confluence, and Git. Writes one markdown file per team to the Obsidian vault.
This skill is a Claude Code port of D:\powershell-scripting\src\bin\Get-Standup-Report.ps1 — same inputs, same template placeholders, same output path layout, same identity-matching rules. See REFERENCE.md for the algorithm and per-section formatter contracts. See EXAMPLES.md for invocation parity with the PS script.
Quick start
/daily-standup-prep → Pyrite, last 2 days, INC 25 / Sprint 1
/daily-standup-prep Onyx → single team
/daily-standup-prep Pyrite,Onyx --days 3 --inc 26 → multi-team, custom window
/daily-standup-prep --no-gitlab --no-confluence → toggle sections off
Parameters
| Param | Default | Maps to PS |
|---|---|---|
Teams |
["Pyrite"] |
-Teams (string[]) |
DaysToLookBack |
2 |
-DaysToLookBack |
JQL |
auto-built | -JQL |
GitRepoPath |
D:\paas\ |
-GitRepoPath |
Inc |
25 |
-Inc |
Sprint |
1 |
-Sprint |
IncludeJiraIssues / IncludeGitCommits / IncludeGitLabActivity / IncludeConfluenceActivity / IncludeKanbanDiagram / IncludeStandupOrder |
all true |
same |
IncludeSprintPulse |
true |
new — runs Phase 4.5 (daily snapshot + council burndown/forecast/suggestions). Toggle off with --no-sprint-pulse |
JIRA base, Confluence base, GitLab base + group, vault root, and template path are all resolved from memory — never accept them as args. See Config resolution below.
Config resolution (run once at the start of every invocation)
| Placeholder | Memory file | Default content / how to populate |
|---|---|---|
{{vault_root}} |
reference_obsidian_vault.md (**Vault root:** line) |
C:\Users\chris.adams\dev\gd-pkms\ — already populated |
{{jira_project}} |
reference_jira_default_project.md (**Default Jira project key:** line) |
PRJ — already populated |
{{confluence_space}} |
reference_confluence_default_space.md |
PP — already populated |
{{template_path}} |
derived | {{vault_root}}\🗃Templates\Standup.md (no separate memory) |
{{rosters_dir}} |
derived | {{vault_root}}\Scrum Teams\_rosters\ |
{{output_root}} |
derived | {{vault_root}}\Scrum Teams\ (per-team subfolder appended later) |
If a memory file is missing or its value is <unset>, prompt once via AskUserQuestion with the PS-script default as the recommended option, then write the answer back so the next run skips the prompt. Same pattern as daily-briefing.
Roster bootstrap (first-run only)
The PS script reads D:\powershell-scripting\src\bin\Teams\<Team>.csv. We move ownership of those CSVs into the vault so the user maintains them in one place going forward.
For each team in the run:
- Look for
{{rosters_dir}}\<Team>.csv(case-insensitive). If present, use it. - If missing and
D:\powershell-scripting\src\bin\Teams\<Team>.csvexists, prompt viaAskUserQuestion: "No roster found at{{rosters_dir}}\<Team>.csv. Copy from the powershell-scripting repo?" — options:Copy (recommended),Use repo path this run only,Skip team. - On
Copy:mkdir -p "{{rosters_dir}}", then copy the file. Tell the user the new path is the canonical location going forward. - If neither path exists, skip the team and warn — do not invent a roster.
CSV format is the same headerless layout the PS Team-Helpers module uses: FullName,Email,Alias,Role,Flags. Lines starting with // and blanks are ignored. Alias and Flags accept ;-separated values. See REFERENCE.md.
Workflow (per team)
Loop the steps below for each team in Teams. Each iteration is independent — one file per team, matching the PS script's $teamsToProcess | ForEach-Object block.
Per-team progress:
- [ ] Phase 1: Resolve config + load roster
- [ ] Phase 2: Compute window + auto-build JQL
- [ ] Phase 3: Gather data (Jira, Git, GitLab, Confluence) — parallel
- [ ] Phase 4: Match activity to team members + flag HasActivity
- [ ] Phase 4.5: Sprint pulse — daily snapshot + council (burndown / forecast / suggestions)
- [ ] Phase 5: Render section markdown + Mermaid kanban + talking order
- [ ] Phase 6: Substitute template placeholders + write to vault
- [ ] Phase 7: Console summary
Phase 1 — Config + roster
Resolve every {{placeholder}} per the table above. Load the roster CSV into a list of {FullName, FirstName, LastName, JiraName, Email, Alias[], DisplayFirstName, Role, Flags[]} records — derive JiraName = "{Last}, {First}" and DisplayFirstName = Alias[0] ?? FirstName. Tag each member with HasActivity = false and empty Issues / Commits / GitLabActivities / ConfluenceActivities lists.
Phase 2 — Window + JQL
activity_since= midnight (Pittsburgh local)DaysToLookBackdays ago.since_iso=YYYY-MM-DDTHH:mm:ssZfor GitLab/Confluence cutoffs.since_date=YYYY-MM-DDforgit log --since.- If
JQLarg is empty: buildproject = {{jira_project}} AND updated >= startOfDay(-{N}d)whereN = DaysToLookBack.
State the chosen window in one short sentence: "Gathering Pyrite standup for window 2026-05-11 → 2026-05-13 (2-day lookback)."
Phase 3 — Gather data (parallel)
Issue these MCP/Bash calls in a single response when section toggles allow:
- Jira issues (
IncludeJiraIssuesorIncludeKanbanDiagram):jira_search(jql=<jql>, fields="summary,status,assignee,issuetype,updated")with pagination to a sane cap. For each result, follow up withjira_get_changelogandjira_get_commentsonly when the issue'supdatedis within the window — skip the fetch otherwise to save tokens. - Jira kanban (
IncludeKanbanDiagramonly): secondjira_searchwithproject = {{jira_project}} AND sprint in openSprints() AND labels in ({lowercase team names}). Capturekey, summary, status, assignee.displayName, issuetype.name. - GitLab MRs + discussions (
IncludeGitLabActivity):list_group_projects(group_id="{{gitlab.XXXX}}")to get all project IDs in the group (subgroups included).- For each project:
list_merge_requests(project_id, updated_after=<since_iso>, per_page=100). - For each MR:
mr_discussions(project_id, merge_request_iid)capped at 100 notes.
- GitLab pipelines (
IncludeGitLabActivity): for each project,list_pipelines(project_id, updated_after=<since_iso>, per_page=100). For each,get_pipeline(project_id, pipeline_id)to read the triggering user. - Confluence pages (
IncludeConfluenceActivity):confluence_search(cql="lastModified >= '{since_date}' AND space = {{confluence_space}}"). For each,confluence_get_page(page_id, expand="history.lastUpdated,version,space,body.view"). - Git commits (
IncludeGitCommits): Bash sequence — confirm{{git_repo}}\.gitexists;git -C "{{git_repo}}" fetch --all;git -C "{{git_repo}}" log --all --since="{since_date}" --pretty=format:"%H|%an|%ae|%aI|%s". Parse pipe-delimited lines.
If any external system fails to respond, log a warning and continue with empty results for that section — never fabricate. If IncludeKanbanDiagram is on but Jira fails, render _No kanban data — Jira unavailable._ rather than skipping the placeholder.
Phase 4 — Match activity to team members
Use the cascading 8-strategy matcher described in REFERENCE.md. For every activity record, attempt to match by (email, displayName, username) — Jira gives email + displayName; GitLab gives email + name + username; Confluence gives email + displayName; Git gives email + author name only.
Filter rules:
- Jira issue inclusion — keep an issue iff its assignee matches a team member OR any comment/changelog entry within the window was authored by a team member. (Direct port of the PS
$isAssignedToTeamMember -or $hasRecentActivityfilter.) - Git/GitLab/Confluence inclusion — keep an activity iff its author matches a team member.
- Set
member.HasActivity = truefor any matched member, and append the activity to the member'sIssues / Commits / GitLabActivities / ConfluenceActivitieslists for the talking-order/per-member sections.
Phase 4.5 — Sprint pulse (daily snapshot + council)
Skip if IncludeSprintPulse is off (substitute empty strings for the three pulse placeholders).
Otherwise follow PULSE.md: run sprint-snapshot --phase daily, read the sprint's
_snapshots.jsonl, and have clarity-council (infographics-expert + statistics-expert + scrum-master)
produce {{burndown_chart}} (burn-up chart per the contract there), {{forecast_note}}, and
{{scrum_suggestions}}. Never fabricate trend data; on snapshot failure render the unavailable note.
Phase 5 — Render sections
Each placeholder in the template gets a string built from the matched data. Formatter contracts (exact markdown shape) live in REFERENCE.md:
{{talking_order}}— randomized active members (✅) first, inactive (🧽) last. Each line- [ ] [[@First Last]] - ✅(or🧽). Multi-team runs append(TeamName)after the wikilink. IfIncludeStandupOrderis off, leave the placeholder empty.{{jira_state}}— Mermaidflowchartwith five status columns (Blocked / Ready / In Progress / In Review / Done). Spec in REFERENCE.md. IfIncludeKanbanDiagramis off, empty.{{jira_issues}}— per-issue blocks with### KEY - Summary, type/status/assignee line, then a bulleted activity log of(date) WHO / WHAT / WHYplustextcode fences for details.{{git_updates}}— per-member blocks### [[@First Last]]then- (date)hash7- messagelines, sorted desc by date.{{gitlab_updates}}— three subsections (## Merge Request Activity,## Comment Activity,## Pipeline Activity) only if non-empty; otherwise the placeholder line_No GitLab activity found in the specified time period._.{{confluence_updates}}— per-page blocks with title + space tag, type/last-modified/author line, link, and a content snippet (first 200 chars of stripped HTML in atextcode fence).
Phase 6 — Substitute and write
Read{{template_path}}. The template is the user's master Standup template — never modify it.- Replace placeholders in this exact set (matches the PS
ConvertTo-TemplateMarkdownfunction plus the three Phase 4.5 additions):{{date}},{{team}},{{sprint}},{{increment}},{{talking_order}},{{jira_state}},{{jira_issues}},{{git_updates}},{{confluence_updates}},{{gitlab_updates}},{{burndown_chart}},{{forecast_note}},{{scrum_suggestions}}. Date formatYYYY-MM-DD(Pittsburgh local). Team name written TitleCase (first char upper, rest lower) to match the PS$formattedTeamline. The three pulse placeholders are no-ops if the user's template doesn't reference them — the substitution still runs cleanly. Note this once when a template lacks them so the user knows to add them if they want the pulse rendered (e.g. under a## Sprint Pulseheading containing{{burndown_chart}}, then### Forecast{{forecast_note}}, then### Suggestions{{scrum_suggestions}}). - Compute output path:
{{vault_root}}\Scrum Teams\<TeamTitleCase>\Scrum 📅\INC {Inc}\Sprint {Sprint}\YYYY-MM-DD.md— note the📅emoji is preserved literally; forward slashes work fine in bash on Windows. mkdir -pthe parent directory.- Never overwrite silently. If the file exists,
Readit, thenAskUserQuestion:Overwrite/Append as ## Re-run HH:MM section/Skip save. Writethe file. Print the saved path back as a clickable markdown link.
Sibling-skill delegation in Phase 4.5 / Phase 6:
The daily snapshot is always delegated to
sprint-snapshotviaSkillwith--phase "daily". Never reimplement Jira sprint fetching inline —sprint-snapshotowns the JQL, identity matching, capacity math, and the JSONL trend schema. If its phase derivation logic doesn't recognizedaily, pass it through as an explicit--phaseoverride (the skill already supports arbitrary phase strings via that flag).The pulse council is always delegated to
clarity-councilviaSkillwith personas pinned toinfographics-expert + statistics-expert + scrum-master. Do not author burndown SVG/Mermaid, forecast paragraphs, or suggestions inline — let each persona produce its own artifact under its own constraints.If extending the template with new Obsidian Flavored Markdown constructs (callouts, embeds, dataview, block IDs, frontmatter properties beyond
Increment/Sprint/Date), invokeobsidian-markdownviaSkillrather than authoring the syntax inline — it's the canonical reference and stays in sync with vault conventions.For wikilink target verification (the
[[@First Last]]rule in talking_order / git_updates), prefer delegating the lookup toobsidian-vaultif the run will need to verify many person notes — its search is more efficient than repeatedGlobcalls. For a single team run (≤15 lookups) the inlineGlobis fine.If the user later wants a dashboard
.basefile that aggregates team standups (e.g. "all standups for Pyrite this sprint"), invokeobsidian-basesrather than hand-authoring the YAML.For CLI-driven vault inspection (e.g. checking note frontmatter, running a vault-wide property query before deciding whether to overwrite), invoke
obsidian-cli— it has direct property/task access thatRead+Globcannot replicate cheaply.
Phase 7 — Console summary
Print a brief block per team (matches the PS === REPORT GENERATION SUMMARY === block):
=== <Team> standup report ===
Team members loaded: N
Jira issues included: N (of M scanned)
Git commits matched: N
GitLab activities matched: N
Confluence activities matched: N
Sprint pulse: snapshot=<ok|failed|skipped> · trend rows=N · council=<ok|failed|skipped>
Saved: <markdown link>
Rules
- Read-only across Jira, GitLab, Confluence. Never transition issues, post comments, approve MRs, edit pages, or modify any external state.
- Read-only on the Git repo.
fetchandlogonly — never checkout, pull, or merge. - No vault writes outside
Scrum Teams/. The template at🗃Templates/Standup.mdis read-only too. - Never overwrite an existing standup file silently. Always confirm via
AskUserQuestion. - Confirm before persisting to memory. When populating an
<unset>value (GitLab base URL, etc.), use the PS-script default as the recommended option but let the user override before writing. - Wikilinks for team members only when the vault person note exists. Use
Glob("{{vault_root}}/🤼 Team/**/@*.md")to verify before writing[[@First Last]]. If the vault has no canonical note for a roster member, render plain text — don't pollute the graph with broken links. - TitleCase team names in output.
pyrite→Pyrite(first char upper, rest lower) — matches the PS$formattedTeamline so vault paths and frontmatter stay consistent. - Pittsburgh local time. All
Datecolumns and the filename use America/New_York. Thegit log%aIand ISO timestamps from Jira/GitLab/Confluence are converted before display. - Section toggle semantics match the PS script. Skipping a section (e.g.,
--no-gitlab) means: don't gather, don't fetch, and substitute an empty string for the placeholder — the template still resolves cleanly. - No fabricated data. If an MCP fails, surface the failure in the console summary and put a
_<system> unavailable_note in the corresponding section. Do not invent. - Sprint pulse is delegated, never inlined. The
dailysnapshot must come fromsprint-snapshotand the burndown/forecast/suggestions must come fromclarity-council's three pinned personas. Do not author burndown charts, forecasts, or scrum-master suggestions inline — that bypasses the persona constraints (infographics-expert's chartjunk rules, statistics-expert's confidence-band requirement, etc.) that make these artifacts trustworthy. - Burndown trend data is read-only and append-only. Phase 4.5 reads
_snapshots.jsonlto feed the chart but never edits past rows. Pastdaily.canvas/daily.mdare allowed to be overwritten — the trend record lives in the JSONL.
Edge cases
- No team members loaded — warn and skip the team. Don't render an empty report (matches PS
if ($teamMembers.Count -eq 0)guard). - Roster CSV missing for one of N teams — process the others; warn for the missing one.
- Vault
🤼 Teamfolder doesn't have a@First Last.mdnote for a roster member — render plainFirst Lastinstead of a wikilink in{{talking_order}}and{{git_updates}}. - Git repo not present at
{{git_repo}}— skip git commit collection silently; render an empty{{git_updates}}section. - Jira sprint label missing for the team (kanban query returns 0) — render
_No <Team> items found in current sprint._for{{jira_state}}. - GitLab group returns 0 projects — surface a warning; the user may have lost group membership. Render
_No GitLab activity found in the specified time period._. - Re-run on the same day — same overwrite/append/skip prompt as
daily-briefing. - Multi-team run, partial failure — generate reports for the teams that succeeded; surface per-team status in the final summary block.
- Sprint pulse with no prior snapshots — first daily run of a sprint will only have one trend row after Step 1 writes it. Render the burndown as a single point with a one-line caption (e.g.
_First daily snapshot of Sprint N — trend will fill in over coming days._) and let the statistics-expert flag the data as too sparse to forecast. - Sprint config note missing —
sprint-snapshotwill prompt to create one. If the user declines, skip the entire pulse phase rather than half-rendering it.
Related skills
This skill writes into and reads from the user's Obsidian vault. When any vault-related extension or refinement is needed, delegate to the relevant obsidian-* skill via Skill rather than reimplementing — they are the canonical references for vault conventions and stay in sync with Obsidian feature changes.
| Skill | Use it for |
|---|---|
daily-briefing |
Personal (Outlook-driven) morning prep. Sibling skill — both can run on the same morning. Same {{vault_root}} resolution pattern. |
sprint-snapshot |
Owns the daily snapshot in Phase 4.5. Always invoke via Skill with --phase "daily"; never reimplement Jira sprint fetching, identity matching, or the JSONL trend schema inline. The _snapshots.jsonl it writes is the burndown's data source. |
clarity-council |
Owns the pulse council in Phase 4.5. Always invoke with the three pinned personas (infographics-expert, statistics-expert, scrum-master). Lets each persona enforce its own constraints on the burndown chart, forecast, and suggestions. |
issue-suggest-component |
Pattern reference for memory-driven default Jira project + bulk AskUserQuestion pacing if the multi-team loop ever grows beyond a handful of teams. |
issue-feature-breakdown |
Pattern reference for Jira+Confluence context-gathering depth. |
obsidian-markdown |
Canonical reference for Obsidian Flavored Markdown. Consult before extending the standup template with new constructs (callouts, embeds, dataview, block IDs, frontmatter properties). |
obsidian-vault |
Use for batch wikilink target verification, vault-wide note discovery, and managing the standup index notes. Prefer over repeated Glob calls when the run needs >15 lookups. |
obsidian-cli |
Use for property/task queries, frontmatter inspection, and any CLI-driven vault operation that Read + Glob cannot do cheaply. |
obsidian-bases |
Use when the user wants a dashboard .base file (e.g. "show me all standups for Pyrite this sprint", "which team members had no activity in the last 5 standups"). Don't hand-author .base YAML. |
obsidian-canvas |
Use if the user wants a visual board summarizing the standup (e.g. a sprint-overview canvas with linked standup notes as nodes). |
Quality Loop
Before returning, check the output against these criteria. If two or more fail, revise and re-check — at most two passes, then note what still falls short.
- Activity is mapped to the correct roster member from the CSV
- The Mermaid kanban renders and reflects the gathered Jira/GitLab state
- Sources cover the requested N-day window across Jira, GitLab, Confluence, and Git
- One file per team is written to the vault
<!-- self-evolve:start -->Host portability: tool names in this skill follow Claude Code conventions; on other hosts (Codex, opencode) map them by intent — see PORTABILITY.md.
Self-Evolve Loop
Journal ~/.ink-and-agency/learnings/daily-standup-prep.md (or workspace-local .ink-and-agency/ where the sandbox confines writes). Read it first; append what the run taught — SELF-EVOLVE.md.