AGENT SKILL

pre-mortem

Guide prospective hindsight analysis to identify project risks before failure occurs. Teams imagine the project has failed spectacularly, then work backward to identify causes. Increases risk identification by 30% compared to traditional planning.

License 许可
MIT
所属仓库
loriensleafs/brain
最近更新
2026-03-11
Stars
★ 0
Forks
0

01 前言 Front Matter

name
pre-mortem
model
claude-sonnet-4-6
agents
[ "analyst", "critic", "planner" ]
license
MIT
metadata
{ "version": "1.0.0", "timelessness": "9/10", "academic_source": "HBR 2007 - Performing a Project Premortem by Gary Klein" }
description
Guide prospective hindsight analysis to identify project risks before failure occurs. Teams imagine the project has failed spectacularly, then work backward to identify causes. Increases risk identification by 30% compared to traditional planning.

02 资源文件 2

03 SKILL.md 正文

Pre-Mortem Risk Identification

When this skill activates, you become a pre-mortem facilitator. Your role is to guide users through prospective hindsight analysis, helping them identify project risks by imagining failure has already occurred.

Triggers

Activate when the user:

  • Run a pre-mortem on...
  • What could cause this project to fail?
  • Identify project risks for...
  • What could go wrong with...
  • Pre-mortem analysis

Quick Reference

Phase Duration Output
1. Brief 2-3 min Project context documented
2. Failure Announcement 30 sec Mindset shift established
3. Independent Analysis 3-5 min Individual failure reasons
4. Round-Robin Collection 5-10 min Consolidated failure list
5. Review and Mitigate 10-15 min Risk inventory with mitigations

When to Use

Use this skill when:

  • Starting a new project or feature and want to identify risks upfront
  • Team is overly optimistic and potential downsides need surfacing
  • Stakeholders need to understand risk before committing resources
  • After major scope changes to reassess risk landscape

Use threat-modeling instead when:

  • Analyzing security threats specifically (STRIDE, attack trees)
  • Need structured vulnerability assessment, not general project risk

Use decision-critic instead when:

  • Challenging a specific decision, not brainstorming broad risks

Why Pre-Mortem Works

Research by Gary Klein demonstrates that prospective hindsight (imagining an event has already occurred) increases the ability to identify reasons for outcomes by 30%.

Key psychological benefits:

  1. Reduces "damn-the-torpedoes" attitude - Acknowledging failure is possible
  2. Creates psychological safety - Dissenters can voice concerns as analysis, not criticism
  3. Surfaces hidden knowledge - People share what they know but fear saying
  4. Catches blind spots early - Before commitment and sunk costs

Process

Phase 1: Project Brief (2-3 minutes)

Input Required:

  • Project name and objective
  • Timeline and key milestones
  • Team composition (if team context)
  • Success criteria

Actions:

  1. Document the project scope clearly
  2. Identify key stakeholders
  3. Note any constraints or dependencies
  4. Confirm understanding with user

Output: Project context document

Phase 2: Failure Announcement (30 seconds)

The critical mindset shift.

Announce to the user (or team):

"The project has failed. It is [timeline endpoint]. The project did not just miss its goals, it failed spectacularly. Your task now is to identify all the reasons why it failed."

Important: Use past tense. The failure has already happened. This is not speculation about what "might" happen; it "did" happen.

Phase 3: Independent Analysis (3-5 minutes)

For individual context:

  • User writes down 5-10 reasons why the project failed
  • No filtering or evaluation yet
  • Include both obvious and "unlikely" causes
  • Consider technical, human, organizational, and external factors

For team context:

  • Each person writes independently
  • No discussion or sharing yet
  • Silent reflection maximizes diversity of thought

Prompt the user:

"Write down every reason you can think of for why this project failed. Include causes you consider unlikely. Do not filter yourself. You have 3-5 minutes."

Output: Raw list of failure reasons

Phase 4: Round-Robin Collection (5-10 minutes)

For individual context:

  • Review each failure reason
  • Group by category (see Categories below)
  • No evaluation yet, just collection

For team context:

  • Go around the room, each person shares ONE reason
  • Continue rounds until all reasons exhausted
  • Record without discussion or debate
  • No "that won't happen" or "we already thought of that"

Categories for grouping:

  1. Technical - Technology, architecture, implementation
  2. People - Skills, availability, communication
  3. Process - Methodology, workflow, coordination
  4. Organizational - Politics, priorities, resources
  5. External - Market, competitors, regulations, dependencies
  6. Unknown Unknowns - Black swan events, assumptions

Output: Categorized failure reasons

Phase 5: Review and Mitigate (10-15 minutes)

For each failure reason:

  1. Assess likelihood (1-5 scale)

    • 1: Very unlikely
    • 2: Unlikely
    • 3: Possible
    • 4: Likely
    • 5: Very likely
  2. Assess impact (1-5 scale)

    • 1: Minor inconvenience
    • 2: Delays
    • 3: Significant setback
    • 4: Major failure
    • 5: Project-killing
  3. Calculate risk score: Likelihood x Impact

  4. Identify mitigation strategies:

    • Prevention: How to stop it from happening
    • Detection: How to know early if it is happening
    • Response: What to do if it happens
  5. Assign ownership (if team context)

Prioritization:

  • Risk score 15-25: Critical, address immediately
  • Risk score 8-14: High, plan mitigation
  • Risk score 4-7: Medium, monitor
  • Risk score 1-3: Low, accept or defer

Output: Complete risk inventory (see template)

Output Template

Generate the risk inventory using this structure:

# Pre-Mortem Risk Inventory

**Project:** [Name]
**Date:** [Date]
**Participants:** [Names or "Individual Analysis"]

## Project Context

- **Objective:** [What success looks like]
- **Timeline:** [Start to end]
- **Key Dependencies:** [Critical external factors]

## Risk Summary

| Priority | Count | Top Risk |
|----------|-------|----------|
| Critical | N | [Highest scoring risk] |
| High | N | [Second highest category risk] |
| Medium | N | |
| Low | N | |

## Critical Risks (Score 15-25)

### R1: [Risk Name]

- **Category:** [Technical/People/Process/Organizational/External]
- **Failure Scenario:** [How this causes project failure]
- **Likelihood:** [1-5] | **Impact:** [1-5] | **Score:** [L x I]
- **Root Cause:** [Why this might happen]
- **Mitigation:**
  - Prevention: [How to avoid]
  - Detection: [Early warning signs]
  - Response: [Contingency plan]
- **Owner:** [Person responsible]
- **Status:** [Open/Mitigating/Accepted]

[Repeat for each critical risk]

## High Risks (Score 8-14)

[Same structure as Critical]

## Medium Risks (Score 4-7)

[Abbreviated: Name, Category, Score, One-line mitigation]

## Low Risks (Score 1-3)

[List only: Name and score]

## Action Items

| Action | Owner | Due Date | Risk Addressed |
|--------|-------|----------|----------------|
| [Specific action] | [Name] | [Date] | R1, R3 |

## Review Schedule

- **Next review:** [Date]
- **Review frequency:** [Weekly/Bi-weekly/Monthly]

Scripts

pre-mortem.ts

Validates risk inventory completeness and calculates aggregate statistics.

bun run ${CLAUDE_SKILL_DIR}/scripts/pre-mortem.ts \
  --inventory-path <path-to-risk-inventory.md> \
  --validate

Exit Codes:

  • 0: Valid inventory with all required fields
  • 1: Invalid arguments
  • 10: Validation failed (missing required sections)

Anti-Patterns

Avoid Why Instead
Skipping Phase 2 Mindset shift is critical for effectiveness Always announce failure explicitly
Allowing discussion during Phase 3 Reduces diversity of thought Enforce silent writing
Filtering "unlikely" causes Miss black swan events Include all causes, prioritize later
Stopping at identification Risk without mitigation is incomplete Always complete Phase 5
One-time exercise Projects evolve, new risks emerge Schedule periodic reviews

Verification

After completing a pre-mortem:

  • [ ] All 5 phases completed (Brief, Failure Announcement, Analysis, Collection, Mitigation)
  • [ ] Risk inventory generated with risk scores (Likelihood x Impact)
  • [ ] All Critical/High risks have mitigation strategies
  • [ ] Each mitigation has Prevention, Detection, and Response
  • [ ] Action items assigned with owners and due dates
  • [ ] Review schedule established

Facilitation Tips

For team settings:

  • Emphasize no blame or criticism
  • Senior members speak last in round-robin
  • Celebrate thorough risk identification
  • Treat skeptics as valuable, not negative

For individual use:

  • Role-play multiple perspectives (developer, user, manager)
  • Challenge your own assumptions explicitly
  • Consider "what would my critics say?"

Extension Points

  1. Custom Categories: Add domain-specific risk categories
  2. Risk Templates: Pre-built risks for common project types
  3. Integration: Link to issue trackers for action item follow-up
  4. Quantification: Add financial impact estimation

Related Skills

Skill Relationship
decision-critic Post-decision validation (pre-mortem is pre-decision)
planner Use pre-mortem output to inform planning
architect Technical risks feed into ADR considerations

References

  • Klein, G. (2007). "Performing a Project Premortem." Harvard Business Review.
  • Mitchell, D. J., et al. (1989). "Back to the Future: Temporal Perspective in the Explanation of Events." Journal of Behavioral Decision Making.

04 同仓库更多 Skills

loriensleafs
loriensleafs/brain

validation-authority

Treat upstream validators as authoritative. Align local config to them. Use when validation fails unexpectedly, before modifying validator behavior, or when tempted to change upstream tool code.

★ 0
loriensleafs
loriensleafs/brain

using-brain-memory

Guidance for using Brain semantic memory effectively. Applies Zettelkasten atomic note principles with observations and relations. Use when deciding whether to query or create notes, structuring note content, or understanding the semantic knowledge graph.

★ 0
loriensleafs
loriensleafs/brain

threat-modeling

Structured security analysis using OWASP Four-Question Framework and STRIDE methodology. Generates threat matrices with risk ratings, mitigations, and prioritization. Use for attack surface analysis, security architecture review, or when asking what can go wrong.

★ 0
loriensleafs
loriensleafs/brain

taste-lints

Custom lints with agent-readable remediation instructions. Enforces taste invariants (file size, naming conventions, structured logging, complexity) and surfaces errors that agents can act on directly. Use when writing or reviewing code to catch style violations early.

Skill与智能体开发
★ 0
loriensleafs
loriensleafs/brain

style-enforcement

Validate code against style rules from .editorconfig, StyleCop.json, and Directory.Build.props. Detects line ending violations, naming convention issues, indentation problems, and charset mismatches across C#, Python, TypeScript, and JavaScript. Produces JSON reports for pre-commit hooks and CI pipelines.

★ 0
loriensleafs
loriensleafs/brain

steering-matcher

Match file paths against steering file glob patterns to determine applicable steering guidance. Use when orchestrator needs to inject context-aware guidance based on files being modified.

★ 0