borghei/Claude-Skills / project-management/discovery/pre-mortem
pre-mortem
VerifiedRuns a pre-mortem that sorts launch risks into evidenced tigers, paper tigers and unspoken elephants, each with an owner. Use when a big launch, build or migration is coming, or the team has a bad feeling it cannot name.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 8.0 KB |
| assets/pre_mortem_template.md | markdown | 2.9 KB |
| assets/sample_risks.json | config | 1.9 KB |
| assets/sample_risks_flawed.json | config | 1.8 KB |
| examples/shared-dashboards-v1-pre-mortem.md | markdown | 18.0 KB |
| references/methodology-and-tools.md | markdown | 10.3 KB |
| references/pre-mortem-guide.md | markdown | 10.1 KB |
| references/red-flags.md | markdown | 12.8 KB |
| scripts/risk_categorizer.py | script | 14.9 KB |
references/methodology-and-tools.md
View on GitHub# Pre-Mortem Methodology & Tooling
> Read this when running a session end-to-end: the thought experiment, full risk classification with examples, the 6-phase facilitation script, the `risk_categorizer.py` tool, output formats, troubleshooting, and success criteria.
## Core Concept
### The Thought Experiment
> "It is 14 days after launch. The product has failed. What went wrong?"
This framing exploits a cognitive bias: people are better at explaining past events than predicting future ones. By placing the failure in the "past" (even fictitiously), participants generate more specific and honest risk assessments.
## Risk Classification
### Tigers (Real Risks)
Tigers are real, evidence-backed risks that could cause serious harm if not addressed.
**Characteristics:**
- Supported by data, past experience, or observable trends.
- The team can describe a plausible failure scenario in concrete terms.
- Ignoring them would be negligent.
**Examples:**
- "Our authentication service went down 3 times last month. A launch-day outage is plausible."
- "We have zero customers in the enterprise segment. Our sales team has no enterprise relationships."
- "The EU regulation takes effect in 60 days. We have not started compliance work."
### Paper Tigers (Look Scary but Unlikely)
Paper Tigers are risks that sound alarming but, on closer inspection, are unlikely or have minimal real impact.
**Characteristics:**
- Based on hypothetical scenarios without supporting evidence.
- The probability is very low, or the impact would be manageable.
- Often raised because of general anxiety rather than specific knowledge.
**Examples:**
- "A competitor might copy our feature" -- possible, but their execution timeline is 6-12 months.
- "The server might not handle 100x traffic" -- but our realistic projection is 5x, with auto-scaling.
- "Users might hate the new UI" -- but usability testing with 8 users showed high satisfaction.
### Elephants (Unspoken Concerns)
Elephants are the risks everyone knows about but nobody talks about. They are the "elephant in the room."
**Characteristics:**
- The team avoids discussing them due to politics, hierarchy, or discomfort.
- Often involve people, process, or organizational issues rather than technical ones.
- Frequently the actual cause of failure when projects fail.
**Examples:**
- "The tech lead does not believe in this project and has been disengaged for weeks."
- "The CEO's pet feature is driving the roadmap, but customers have not asked for it."
- "We do not have a plan for what happens when the contractor's contract ends next month."
## Tiger Urgency Classification
Once a risk is classified as a Tiger, assign an urgency level:
| Urgency | Definition | Action Required |
|---------|-----------|----------------|
| **Launch-Blocking** | If unresolved, the launch should not proceed. | Concrete mitigation plan, assigned owner, decision date before launch. |
| **Fast-Follow** | Must be addressed within 2 weeks after launch. | Documented plan, assigned owner, scheduled for first post-launch sprint. |
| **Track** | Should be monitored and addressed if it escalates. | Added to risk register, reviewed at regular cadence. |
## Methodology
### Phase 1: Set the Scene (5 minutes)
The facilitator reads this prompt to the group:
> "Imagine it is 14 days after our launch. The product has failed. Users are not adopting it, key metrics are down, and leadership is asking what went wrong. Take 10 minutes to write down every reason you can think of for why we failed. Be specific. Be honest. Nothing is off limits."
**Ground rules:**
- Anonymous contributions (sticky notes or digital equivalent).
- No attribution, no blame, no judgment.
- Quantity over quality in the first round.
### Phase 2: Generate Risks (10 minutes)
Each participant independently writes down failure scenarios. One risk per sticky note. Aim for 5-10 per person.
**Prompts to stimulate thinking:**
- What technical system is most likely to break?
- What customer objection have we not addressed?
- What team dynamic could derail us?
- What external event could change our assumptions?
- What are we pretending is not a problem?
- What decision are we avoiding?
### Phase 3: Share and Cluster (15 minutes)
1. Read each risk aloud (without attribution).
2. Place on the board.
3. Group similar risks together.
4. Merge duplicates.
### Phase 4: Classify (15 minutes)
For each cluster, the group decides:
| Classification | Criteria |
|---------------|----------|
| **Tiger** | Supported by evidence. Plausible failure scenario. |
| **Paper Tiger** | Sounds scary but unlikely or low-impact on inspection. |
| **Elephant** | The room got quiet when this was read. People exchanged glances. |
For each Tiger, assign urgency: Launch-Blocking, Fast-Follow, or Track.
### Phase 5: Mitigation Plans (15 minutes)
For each **Launch-Blocking Tiger**, complete:
| Field | Description |
|-------|-------------|
| **Risk** | Clear description of the risk |
| **Evidence** | What data or experience supports this being a real risk? |
| **Mitigation** | Specific, concrete action to reduce the risk |
| **Owner** | Single person accountable |
| **Decision Date** | Date by which the mitigation must be complete or the launch decision revisited |
### Phase 6: Address Elephants (10 minutes)
Elephants require a different approach than Tigers:
1. **Acknowledge** -- Name the elephant explicitly. "The team is concerned that..."
2. **Assess** -- Is this actually a Tiger in disguise? If so, reclassify.
3. **Decide** -- Either address it (assign an owner) or consciously accept it (document the acceptance and rationale).
## Python Tool: risk_categorizer.py
Categorize and analyze risks using the CLI tool:
```bash
# Run with demo data
python3 scripts/risk_categorizer.py --demo
# Run with custom input
python3 scripts/risk_categorizer.py input.json
# Output as JSON
python3 scripts/risk_categorizer.py input.json --format json
```
### Input Format
```json
{
"risks": [
{
"description": "Authentication service has had 3 outages in the last month",
"category": "tiger",
"evidence": "Incident reports from last 30 days",
"urgency": "launch_blocking"
}
]
}
```
### Output
Risk distribution summary, action plans for launch-blocking tigers, and flags for elephants that may need investigation.
See `scripts/risk_categorizer.py` for full documentation.
## Output Format
### Pre-Mortem Summary
```
Pre-Mortem Analysis: [Product/Feature Name]
Date: YYYY-MM-DD
Participants: [list]
Total risks identified: N
- Tigers: X (Launch-Blocking: A, Fast-Follow: B, Track: C)
- Paper Tigers: Y
- Elephants: Z
```
### Risk Registry
| # | Risk | Category | Urgency | Evidence | Mitigation | Owner | Decision Date |
|---|------|----------|---------|----------|-----------|-------|--------------|
| 1 | ... | Tiger | Launch-Blocking | ... | ... | ... | ... |
| 2 | ... | Tiger | Fast-Follow | ... | ... | ... | ... |
| 3 | ... | Paper Tiger | -- | ... | -- | -- | -- |
| 4 | ... | Elephant | TBD | ... | ... | ... | ... |
Use `assets/pre_mortem_template.md` for the full document template.
## Troubleshooting
| Symptom | Likely Cause | Resolution |
|---------|-------------|------------|
| Team generates mostly paper tigers | Risk aversion or surface-level thinking; team not fully immersing in the failure scenario | Re-read the thought experiment prompt slowly; extend silent writing time from 10 to 15 minutes; use specific prompts |
| No elephants surfaced | Psychological safety too low, or facilitator is a manager creating power dynamics | Use anonymous contribution (sticky notes or digital tools); consider an external facilitator; separate session from performance reviews |
| All risks classified as tigers | Team lacks calibration on what constitutes real evidence vs. anxiety | Require concrete evidence for each tiger; if evidence is hypothetical, reclassify as paper tiger |
| Elephant escalation check flags too many false positives | ESCALATION_KEYWORDS list is broad, matching common words in non-critical contexts | Review matched keywords in output; refine escalation thresholds; use the recommendation as a prompt, not a verdict |
| Launch-blocking tigers have no owners assigned | Pre-mortem session ended without Phase 5 mitigation planning | Always reserve 15 minutes for mitigation plans; do not skip Phase 5 even if the session runs long |
| Validation errors on input JSON | Missing required fields (`description`, `evidence`, `category`) or invalid urgency for tigers | Check that every tiger has `urgency` set to one of: `launch_blocking`, `fast_follow`, `track` |
## Success Criteria
- Pre-mortem conducted before every major launch, migration, or significant resource commitment
- At least 5-10 risks generated per person during the silent writing phase
- Risk distribution includes all three categories (not all tigers, not all paper tigers)
- Every launch-blocking tiger has an assigned owner, concrete mitigation plan, and decision date
- Elephants are explicitly named and either addressed or consciously accepted with documented rationale
- Pre-mortem findings are reviewed against actual outcomes post-launch to calibrate future sessions
- Session duration stays within 60-90 minutes total across all 6 phases
## Tool Reference
### risk_categorizer.py
Categorizes pre-mortem risks as Tigers, Paper Tigers, or Elephants. Generates action plans for launch-blocking tigers and flags elephants with escalation signals.
| Flag | Type | Default | Description |
|------|------|---------|-------------|
| `input_file` | positional | (optional) | Path to JSON file with risks array |
| `--demo` | flag | off | Run with built-in sample data (7 risks across all categories) |
| `--format` | choice | `text` | Output format: `text` or `json` |
**Input fields per risk:**
- `description` (required): Clear description of the risk scenario
- `category` (required): One of `tiger`, `paper_tiger`, `elephant`
- `evidence` (required): Supporting data or observations
- `urgency` (required for tigers): One of `launch_blocking`, `fast_follow`, `track`
## Bibliography
- Gary Klein, "Performing a Project Pre-Mortem," *Harvard Business Review* (2007)
- Daniel Kahneman, *Thinking, Fast and Slow* (2011) -- prospective hindsight
- Chip Heath & Dan Heath, *Decisive* (2013) -- decision-making under uncertainty
- Amy Edmondson, *The Fearless Organization* (2018) -- psychological safety for surfacing elephants
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| add-product-e2e-evalpaperclipai/paperclipAdd or extend a Paperclip full-stack runner E2E workflow, fixture, matcher, or report evidence path for local or Daytona execution. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |
| add-runner-evalpaperclipai/paperclipAdd or extend a Paperclip Runner protocol evaluation definition, roster, assertion, or report fixture with provenance and narrow validation. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |
| diagnose-why-work-stoppedpaperclipai/paperclipCompatibility entry point for failed, stalled, looping, or over-recovered Paperclip work. Use fix-failed-paperclip-run for the full diagnosis, regression, repair, real verification, and green-PR workflow; honor an explicitly requested diagnosis-only scope. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |
| fix-failed-paperclip-runpaperclipai/paperclipDiagnose and fix failed, stuck, slow, or incorrectly blocked Paperclip tasks and runs. Use with a task/run URL or failure report to trace the root cause, prove a regression red then green, check second-order effects, verify the real workflow, and deliver a PR with passing checks. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |