More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Postmortem Forge: Incident Postmortem & RCA

    1

    Build a source-linked incident timeline from Slack exports, PagerDuty JSON and logs. Review the milestones, calculate TTD/TTM, then lint the postmortem for missing owners, dates and verification.

    $5

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Postmortem Forge: Incident Postmortem & RCA

    Postmortem Forge: Incident Postmortem & RCA

    Example session with this skill installed

    Use the included synthetic inventory incident exports to draft a blameless postmortem. Show the source-linked timeline, explain which milestone IDs were reviewed, and calculate TTD and TTM without guessing missing facts.

    • Read your context and instructions
    • Compiled the postmortem forge

    Reproducible synthetic example, not a production case study. This is the complete captured JSON output from the packaged timeline script with examples/milestones.json. Milestone choices and the finished report are included in examples/README.md and examples/postmortem.md.

    {
      "events": [
        {
          "id": "E001",
          "time": "2026-10-17T14:02:08.000Z",
          "text": "2026-10-17 14:02:08 UTC deploy: inventory-api release b7c41a2 (PR #681)",
          "kind": "change",
          "sources": [
            "examples/raw/deploy.log:1"
          ]
        },
        {
          "id": "E002",
          "time": "2026-10-17T14:03:00.000Z",
          "text": "2026-10-17 14:03:00 UTC synthetic trace: first inventory reservation timeout",
          "kind": "note",
          "sources": [
            "examples/raw/deploy.log:2"
          ]
        },
        {
          "id": "E003",
          "time": "2026-10-17T14:10:12.000Z",
          "text": "triggered_at: inventory-api p99 latency above 800ms",
          "kind": "detection",
          "sources": [
            "examples/raw/pagerduty.json#/alerts/0/triggered_at"
          ]
        },
        {
          "id": "E004",
          "time": "2026-10-17T14:11:30.000Z",
          "text": "acknowledged_at: inventory-api p99 latency above 800ms",
          "kind": "note",
          "sources": [
            "examples/raw/pagerduty.json#/alerts/0/acknowledged_at"
          ]
        },
        {
          "id": "E005",
          "time": "2026-10-17T14:15:00.000Z",
          "text": "[2:15 PM] Eli: maybe it is Redis? checking connection counts",
          "kind": "hypothesis",
          "sources": [
            "examples/raw/slack-export.txt:2"
          ]
        },
        {
          "id": "E006",
          "time": "2026-10-17T14:20:00.000Z",
          "text": "[2:20 PM] Noor: Redis healthy; reject the Redis hypothesis",
          "kind": "hypothesis",
          "sources": [
            "examples/raw/slack-export.txt:5"
          ]
        },
        {
          "id": "E007",
          "time": "2026-10-17T14:25:00.000Z",
          "text": "[2:25 PM] Noor: flipped inventory_cache_v2 off; checking effect",
          "kind": "note",
          "sources": [
            "examples/raw/slack-export.txt:4"
          ]
        },
        {
          "id": "E008",
          "time": "2026-10-17T14:31:00.000Z",
          "text": "[2:31 PM] Eli: mitigation verified: inventory latency under 500ms for 6 minutes",
          "kind": "note",
          "sources": [
            "examples/raw/slack-export.txt:3"
          ]
        },
        {
          "id": "E009",
          "time": "2026-10-17T14:47:00.000Z",
          "text": "[2:47 PM] Noor: confirmed cache TTL was zero after PR #681; schema allowed zero",
          "kind": "diagnosis",
          "sources": [
            "examples/raw/slack-export.txt:1"
          ]
        },
        {
          "id": "E010",
          "time": "2026-10-17T15:03:22.000Z",
          "text": "2026-10-17 15:03:22 UTC deploy: inventory-api reverted PR #681",
          "kind": "mitigation",
          "sources": [
            "examples/raw/deploy.log:3"
          ]
        },
        {
          "id": "E011",
          "time": "2026-10-17T15:19:40.000Z",
          "text": "resolved_at: inventory-api p99 latency above 800ms",
          "kind": "note",
          "sources": [
            "examples/raw/pagerduty.json#/alerts/0/resolved_at"
          ]
        }
      ],
      "milestones": {
        "started": {
          "id": "E002",
          "time": "2026-10-17T14:03:00.000Z",
          "sources": [
            "examples/raw/deploy.log:2"
          ]
        },
        "detected": {
          "id": "E003",
          "time": "2026-10-17T14:10:12.000Z",
          "sources": [
            "examples/raw/pagerduty.json#/alerts/0/triggered_at"
          ]
        },
        "diagnosed": {
          "id": "E009",
          "time": "2026-10-17T14:47:00.000Z",
          "sources": [
            "examples/raw/slack-export.txt:1"
          ]
        },
        "mitigated": {
          "id": "E008",
          "time": "2026-10-17T14:31:00.000Z",
          "sources": [
            "examples/raw/slack-export.txt:3"
          ]
        },
        "resolved": {
          "id": "E011",
          "time": "2026-10-17T15:19:40.000Z",
          "sources": [
            "examples/raw/pagerduty.json#/alerts/0/resolved_at"
          ]
        }
      },
      "metrics": {
        "ttd": {
          "seconds": 432,
          "from": "E002",
          "to": "E003"
        },
        "ttcd": {
          "seconds": 2208,
          "from": "E003",
          "to": "E009"
        },
        "ttm": {
          "seconds": 1248,
          "from": "E003",
          "to": "E008"
        },
        "duration": {
          "seconds": 4600,
          "from": "E002",
          "to": "E011"
        }
      },
      "warnings": []
    }
    

    Connects securely to your tools. The creator never sees your data.

    About this skill

    An incident report needs evidence you can check. Postmortem Forge reads local exports, keeps the source beside each event, and calculates durations after you review the milestones.

    The important boundary: a deploy does not prove when impact started. A rollback attempt does not prove recovery. Suggested tags stay suggestions, and missing milestones return unknown.

    What you get:

    • A UTC timeline builder for timestamped text, Slack JSON exports, PagerDuty lifecycle records and deploy logs. It preserves dates and seconds, handles explicit timezone offsets, sorts events and merges exact duplicates without dropping their sources.
    • A Markdown report linter. Missing sections, vague tasks, and action items without an owner, a valid ISO due date or a verification step fail the check. Blame language, timeline gaps, metric presence and Prevent/Detect/Mitigate coverage produce review advisories.
    • A contributing-factors playbook, report template and runnable synthetic example. The example includes the raw evidence, reviewed milestone IDs and a finished report. You can also lint a report you already wrote.

    Use it for a blameless incident postmortem, root cause analysis (RCA), outage report, post-incident review (PIR), or correction-of-error (COE) draft.

    The included example contains 11 events. TTD is 7m12s, measured from the first observed timeout to the alert. The flag is switched off before mitigation is verified. Those are separate events on purpose.

    Requires Node.js 18+ and an agent that can run local scripts. No package installation or API keys. Scripts make no network requests; your agent's usual provider policy still applies to any evidence or output it receives.

    Limits: the parser supports documented export shapes, not every vendor format. Clock-only messages require a date and UTC offset. Named timezones and daylight-saving transitions need explicit offsets. Optional masking catches common secrets, emails and IPv4 patterns, but it cannot guarantee safe sharing. Human review remains necessary.

    Supported export shapes only. Clock-only input needs a date and UTC offset; named timezones and DST are not inferred. Suggested tags never confirm causality. Metrics require reviewed milestone IDs. Masking is incomplete, including IPv6 and unknown secret formats. The linter checks report structure and language, not truth. Your agent/provider data policy applies to evidence it receives.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 1 month ago

    Requires

    Node.js 18+

    Permissions

    Terminal / Shell
    Read Files
    Write Files

    Terminal runs the bundled Node scripts. Read Files reads approved exports and drafts. Write Files covers temporary fixtures created by the optional tests, plus reports or milestone files only when you approve saving them. The timeline and linter print to stdout and make no network requests.

    Listed1 month ago

    What's inside

    Frequently Asked Questions

    No. It reads local exported files. You do not connect an account or supply an API key. Supported JSON shapes and text formats are documented in the package.

    You review the timeline and select event IDs for impact start, detection, confirmed diagnosis, effective mitigation and recovery. TTD is impact start to detection; TTM is detection to effective mitigation. Unselected milestones return unknown. It does not guess the correct diagnosis from keywords.

    Fully dated timestamps with explicit offsets can cross days. Clock-only messages need one date and UTC offset per run. Named timezones and daylight-saving changes are not inferred.

    Missing required sections, empty action lists, vague tasks, and action items missing owners, real ISO due dates or verification fail with exit 1. Blame language, metric presence, timeline gaps and control-plane coverage are advisories. The linter does not verify factual truth.

    The Node scripts make no network requests. Your AI agent may send evidence and script output to its provider under its usual policy. Optional regex masking is incomplete and needs review before sharing.

    Yes. The linter reads Markdown drafts, including bullet and table action items. You can run it manually or use its exit code in a documentation CI job. The included tests and synthetic example are runnable locally.

    New in Technical Documentation