narrative-verification
Skill8 starsChanged 11 days agoApache-2.0View on GitHub
Prevent fabrication when writing narrative blog posts from session logs or omnibus logs — adds a verification checkpoint layer on top of existing voice/template standards.
In this repository39 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- .cursor/rules/architecture-patterns.mdcCursor rule
- .cursor/rules/completion-discipline.mdCursor rule
- .cursor/rules/github-tracking.mdcCursor rule
- .cursor/rules/programmer-briefing.mdcCursor rule
- .cursor/rules/verification-first.mdcCursor rule
- ab-a-isolationSkill
- assign-sprint-safelySkill
- audit-cascadeSkill
- brief-coding-agentSkill
- check-mailboxSkill
- cleanup-dev-activeSkill
- close-issue-properlySkill
- cohort-attention-rollupSkill
- compost-reviewSkill
- continue-narrativeSkill
- create-omnibusSkill
- create-session-logSkill
- cut-releaseSkill
- delete-module-safelySkill
- deliver-mailSkill
- discovered-work-captureSkill
- doc-sync-sweepSkill
- draft-blog-postSkill
- draft-weekly-shipSkill
- duty-cycle-tickSkill
- piper-draft-issueSkill
- piper-draft-specSkill
- piper-sprint-planSkill
- piper-stakeholder-updateSkill
- piper-synthesize-feedbackSkill
- propose-featureSkill
- publish-to-blogSkill
- query-github-boardSkill
- template-auditSkill
- trust-checkSkill
- update-calendarSkill
- update-current-stateSkill
- update-piperSkill
The file
We split this skill into the rules it gives an agent
7 rules. About 0 are common, 7 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈7
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 24
When you encounter a gap in the narrative, the placeholder IS the correct content.
- Distinctive: line 30
Before writing ANY prose, create a verified facts list from the source logs.
- Distinctive: line 59
Only after this extraction is complete should you begin drafting.
- Distinctive: line 81
Use specific placeholders that explain what's needed:
- Distinctive: line 101
Stop and verify if you notice yourself:
- Distinctive: line 109
When you catch yourself doing any of these: STOP and create a placeholder instead.
- Distinctive: line 115
Before delivering a draft, review each specific claim:
≈ “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.
Files
Narrative Verification Skill
Purpose: Prevent fabrication when writing narrative blog posts from session logs or omnibus logs.
When to use: Any narrative piece ("The Drift We Didn't See," "The Calendar That Wasn't Mine") where the source material is logs describing what happened.
Does NOT replace: Voice & Tone Style Guide, blog post templates, existing formatting standards. This skill ADDS a verification layer.
Core Principle
Placeholders are safeguards, not clutter.
When you encounter a gap in the narrative, the placeholder IS the correct content. Filling it with plausible fabrication defeats its purpose. The PM will provide the missing detail — or decide it's not needed.
Pre-Draft: Facts Extraction (REQUIRED)
Before writing ANY prose, create a verified facts list from the source logs.
Extract and cite:
## Verified Facts from [Log Name]
### Trigger/Origin
- [What actually started this? Line/timestamp reference]
### Sequence of Events
- [First thing that happened - cite line]
- [Second thing - cite line]
- [etc.]
### Specific Numbers
- [Files changed: X - cite line]
- [Duration: X - cite line]
- [Issues created: X - cite line]
### Quotes/Exact Phrasing
- [Any direct quotes from log - cite line]
### Causality
- [What caused what? Only if explicitly stated in log]
### NOT IN LOGS (will need placeholders)
- [List what the narrative needs but logs don't provide]Only after this extraction is complete should you begin drafting.
Claims Requiring Verification
These categories are high-risk for fabrication. If you can't cite a source line, use a placeholder.
| Category | Example | Risk |
| Triggers | "It started with X" | Often invented to create narrative arc |
| Numbers | "73 columns," "three days" | Easy to misremember or round |
| Sequences | "X led to Y led to Z" | Causality often invented |
| Timestamps | "8:06 AM," "by afternoon" | Specific times need verification |
| Technical details | "migration hash," "function name" | Plausible but often wrong |
| Emotional moments | "the moment we realized" | Logs rarely capture feelings |
| Comparisons | "faster than before" | Need baseline evidence |
Placeholder Templates
Use specific placeholders that explain what's needed:
[PM PLACEHOLDER: What was the actual trigger for this investigation?
Logs show X but not what prompted looking at X initially.]
[PM PLACEHOLDER: The logs show 47 columns, not 73. Please verify
the correct number, or explain if 73 refers to something else.]
[PM PLACEHOLDER: How did you feel when this was discovered?
Logs don't capture the emotional moment.]
[PM PLACEHOLDER: Is this sequence correct? Logs show A and B
but don't explicitly state A caused B.]Red Flags: Common Fabrication Patterns
Stop and verify if you notice yourself:
- Inverting causality — Something discovered DURING work becomes the "trigger" for that work
- Smoothing details — Adding specifics (times, names, numbers) that make the story flow but aren't in logs
- Narrative completion — Filling gaps because the story "needs" something there
- Memory-feeling — Writing details that feel like memory but have no citation
- Plausible specifics — Technical details (hashes, function names) that could be true but aren't verified
When you catch yourself doing any of these: STOP and create a placeholder instead.
Verification Checkpoint
Before delivering a draft, review each specific claim:
- Trigger/origin — cited from log?
- All numbers — cited from log?
- Sequence/causality — explicitly stated in log, or marked as interpretation?
- Technical details — verified or placeholdered?
- Timeline — matches log timestamps?
- Quotes — actually in the log?
Any unchecked item → add placeholder or remove claim.
Integration with Existing Workflow
This skill fits INTO the existing process:
1. Receive logs from PM
2. Read logs thoroughly
3. ★ NEW: Create Verified Facts extraction ★
4. Pitch narrative ideas (based on verified facts)
5. PM approves direction
6. Draft narrative (using extraction as guardrails)
7. ★ NEW: Run Verification Checkpoint before delivery ★
8. Deliver draft with placeholders intact
9. PM reviews, fills placeholders, polishes
10. PublishSteps 3 and 7 are the additions. Everything else stays the same.
Why This Matters
Narrative pieces tell readers "here's what happened." Fabricated details — even small ones — erode trust and create false records. The PM has the actual memories; we have the logs. When logs are insufficient, we ask. We don't invent.
Version 1.0 — February 12, 2026 Created after Calendar and Drift fabrication incidents
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
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.

