agentleFS
jcesarperezjcesarperez/claude-em/.claude/skills/write-update/SKILL.md

write-update

Skill94 starsChanged 5 months agoMITView on GitHub

Write a periodic update (weekly, monthly, etc.) for a team. Reads data from Jira, Slack, and configured URLs for the last N days and generates the update in the chosen template format.

  • Reads credentials

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 3 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Check the user's prompt for:

    line 14
  2. Very specific:

    Update source URLs (under ## Update Sources)

    line 22
  3. Distinctive:

    Wait for all subagents to complete.

    line 135
  4. Very specific:

    Do not invent data. If a template section has no supporting facts, write No data available this week.

    line 148
  5. Very specific:

    Read the chosen template file from this skill's assets/ folder.

    line 154
  6. Distinctive:

    If any source failed, append after the update:

    line 164

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Skill: Write Update

You are helping an Engineering Manager write a periodic update for their team.


Step 1 — Determine inputs

Check the user's prompt for:

  • Team (required — e.g. "events"). If missing, ask.
  • Format (optional — template name). If missing, list the .md files in this skill's assets/ folder and ask the user to choose.
  • Date range (optional — defaults to last 7 days inclusive from today). Accept natural language like "last week" or "from Monday to Sunday". Convert to absolute dates before passing to subagents.

Once you have the team, read data/team_{name}.md and extract:

  • Jira project key, board ID, Jira skill name, and conventions
  • Slack channels (under ## Slack)
  • Update source URLs (under ## Update Sources)

Then read the chosen template file. If it contains an ## Instructions section, follow those instructions now — before deciding which sources to query. The instructions may:

  • Tell you to ask the user which project this update is for
  • Restrict which Jira epics, Slack channels, and URLs to use
  • Tell you to filter shared sources (e.g. meeting notes) by project name or epic keys

Resolve all source scoping decisions before launching subagents in Step 2.


Step 2 — Collect data via parallel subagents

Launch one subagent per data source using the Agent tool, all in parallel (single message, multiple tool calls). Do not wait for one before launching the others.

Pass each subagent only what it needs. Each subagent must return a structured summary — never raw content.

Subagent: Jira

Prompt template:

You are collecting Jira data for a team update.

Team: {team} Date range: {date_from} to {date_to} Jira project key: {project_key} Jira skill: {jira_skill} Jira conventions: {conventions}

Use the {jira_skill} skill to query:

  • Epics updated or closed in this date range
  • Issues updated or closed in this date range
  • Apply all team-specific filters

Return a structured summary with:

  • Completed: list of done items (title, Jira URL, epic they belong to, any metrics)
  • In progress / updated: notable items that moved but aren't done
  • Incidents / bugs: any incidents or customer cases
  • Allocation: counts by Strategy / Scale / KTLO if available

If data is unavailable, return {"error": "reason"}.

Subagent: Slack (one per channel)

Launch one subagent per Slack channel. Prompt template:

You are extracting signal from a Slack channel for a team update.

Channel: {channel_name} Date range: {date_from} to {date_to}

Important — use mcp__claude_ai_Slack__slack_read_channel as the primary tool. The local mcp__slack__* tools (slack_list_channels, slack_get_channel_history, etc.) have broken authentication in this workspace and must NOT be used.

Steps:

  1. Compute Unix timestamps from the human-readable dates using Python or date CLI before calling Slack:
    python3 -c "import datetime; print(int(datetime.datetime(YYYY, MM, DD, tzinfo=datetime.timezone.utc).timestamp()))"
    Do NOT pre-calculate or hard-code Unix timestamps — always derive them from the date strings at runtime to avoid year-offset bugs.
  2. Call mcp__claude_ai_Slack__slack_read_channel with the channel name (without the #).
    • Pass the computed oldest and latest Unix timestamps to filter messages server-side.
    • If the channel is private and the name doesn't resolve, try with the channel ID if known.
  3. For individual threads with relevant signal, call mcp__claude_ai_Slack__slack_read_thread to fetch replies.
  4. DM channels (IDs starting with D) are not accessible — return {"error": "DM channel not accessible"}.

Extract only:

  • Shipped features or completed work mentioned
  • Decisions made
  • Risks, incidents, or blockers raised
  • Customer feedback quotes (verbatim if possible)

Ignore: bot messages, standup check-ins, scheduling, reactions-only threads.

Return a structured bullet list of findings, or {"error": "reason"} if the channel is unreachable.

Subagent: URL (one per Update Source)

Launch one subagent per URL. Prompt template:

You are extracting relevant information from a page for a team update.

URL: {url} Date range: {date_from} to {date_to}

For Atlas URLs (home.atlassian.com — project or goal update pages): Atlas is a separate Atlassian product; the MCP only covers Jira and Confluence, and WebFetch cannot authenticate. Use the bash script data/atlassian/scripts/get_atlas_project_updates.sh, which queries the Atlassian GraphQL API using credentials from .env.local:

# Credentials are loaded automatically from .env.local by the script
data/atlassian/scripts/get_atlas_project_updates.sh {project_key} {site_id} {date_from} {date_to}

Extract the project_key from the URL (e.g. MEWS-4494 from .../project/MEWS-4494/updates). The Mews site ID is 9cb80e14-6d83-4419-a743-3b8c78b0bf7f.

For Confluence URLs (mews.atlassian.net/wiki):

  1. First try mcp__claude_ai_Atlassian__fetch with the full URL — this handles personal spaces and folders that the REST API cannot address by page ID.
  2. If that fails, extract the space key from the URL path (e.g. ~71202020c7dda3e907495697c548dd58e509ff) and search with CQL: space = "{space_key}" AND created >= "{date_from}" AND created <= "{date_to}" using mcp__claude_ai_Atlassian__searchConfluenceUsingCql.
  3. Do NOT try to call the REST API directly using the numeric folder/page ID from the URL — personal space folders are not addressable this way and will return 404.

For other URLs: fetch with WebFetch.

Extract only content dated within the range. If dates are ambiguous, include the content and flag it.

Return:

  • Decisions / outcomes from meetings or notes
  • Risks or blockers surfaced
  • Notable context relevant to a weekly team update

Return {"error": "reason"} if the URL fails to load.


Step 3 — Collect results and note gaps

Wait for all subagents to complete. Collect their outputs. For any subagent that returned an error, record it for the Data Gaps section — do not stop or retry.


Step 4 — Synthesise

From all subagent outputs, build an internal fact list:

  • Achievements with metrics
  • Delays, incidents, risks
  • Customer feedback quotes
  • Pipeline / upcoming items
  • Allocation data

Do not invent data. If a template section has no supporting facts, write No data available this week.


Step 5 — Write the update

Read the chosen template file from this skill's assets/ folder. Follow it exactly:

  • Same section headings, order, and emoji
  • Tone as described in the template
  • Link every initiative/feature to its Jira or Atlas URL
  • Quantify achievements wherever data supports it

Step 6 — Report gaps

If any source failed, append after the update:

---
**Data gaps:**
- Slack #rnd-events-team: could not connect
- URL https://...: failed to load

Omit entirely if all sources returned data.


Output

The full update, ready to review. No meta-commentary before or after. The EM will review, edit, and publish manually.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.