factual-summarizer
Skill0 starsChanged 2 months agoMITView on GitHub
Summarizes supplied material strictly without adding external claims, assumptions, or hallucinations.
In this repository49 other files it gives its agents.
- bug-reproducerSkill
- claim-checkerSkill
- command-failure-diagnoserSkill
- config-risk-reviewerSkill
- csv-validatorSkill
- deployment-checklist-makerSkill
- diff-reviewerSkill
- field-extractorSkill
- api-contract-testerSkill
- database-migration-helperSkill
- security-vulnerability-fixerSkill
- git-commit-stagerSkill
- git-conflict-resolverSkill
- github-actions-debuggerSkill
- github-issue-triagerSkill
- github-pr-creatorSkill
- github-pr-reviewerSkill
- go-code-qualitySkill
- go-lint-and-vetSkill
- go-testingSkill
- json-schema-converterSkill
- lint-and-typecheckSkill
- meaning-preserving-editorSkill
- meeting-notes-to-actionsSkill
- minimal-bug-fixSkill
- option-comparatorSkill
- py-code-qualitySkill
- py-lint-and-typecheckSkill
- py-testingSkill
- readme-writerSkill
- regression-test-writerSkill
- release-notes-drafterSkill
- repo-inspectorSkill
- research-briefSkill
- reversible-cleanup-plannerSkill
- rust-clippy-checkSkill
- rust-code-qualitySkill
- rust-testingSkill
- small-refactorSkill
- status-updateSkill
- step-by-step-tutorialSkill
- step-decomposerSkill
- tabular-analyzerSkill
- task-briefSkill
- test-runner-debuggerSkill
- ts-code-qualitySkill
- ts-lint-and-typecheckSkill
- ts-testingSkill
- skill-templateSkill
The file
We split this skill into the rules it gives an agent
12 rules. About 0 are common, 8 are distinctive, and 4 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈8
Distinctive · estimate
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 14
Use this skill when:
- Very specific: line 20
Do not use to synthesize multi-source research briefs with hypothesis evaluation.
- Very specific: line 21
Do not use to verify specific disputed claims.
- Very specific: line 22
Do not use to rephrase or polish text.
- Very specific: line 29
If source_material is missing: stop and ask: "Please provide the text you would like summarized."
- Distinctive: line 30
If the source material does not mention a topic: state "The provided text does not mention this topic." Do not fill gaps from world …
- Distinctive: line 33
Follow these numbered actions:
- Distinctive: line 39
Write short, clear sentences.
- Distinctive: line 42
Check each bullet against source text: if it cannot be pointed to, remove it.
- Distinctive: line 47
If the text contains contradictory statements: report both contradictory statements verbatim.
- Distinctive: line 48
If the text mentions probabilities or estimates: preserve qualifiers ("approximately", "alleged", "estimated").
- Distinctive: line 80
Stop and ask the user when:
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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
Factual Summarizer
Produces strict, grounded summaries of provided text. Every bullet point is backed by exact source sentences with zero external extrapolation.
1. Purpose
Condense long documents, incident reports, or articles into concise facts. "Grounded" means no information outside the supplied text is introduced.
2. When to Use
Use this skill when:
- Summarizing technical papers, logs, vendor documents, or legal notices.
- User explicitly requests a factual or neutral summary.
- Preparing concise input for another model with tight context limits.
3. When Not to Use
- Do not use to synthesize multi-source research briefs with hypothesis evaluation; use
research-brief. - Do not use to verify specific disputed claims; use
claim-checker. - Do not use to rephrase or polish text; use
meaning-preserving-editor.
4. Required Inputs
source_material: The complete text to summarize (string).max_bullets: Optional maximum number of bullet points (default: 5).
5. Missing-Input Behavior
- If
source_materialis missing: stop and ask: "Please provide the text you would like summarized." - If the source material does not mention a topic: state "The provided text does not mention this topic." Do not fill gaps from world knowledge.
6. Procedure
Follow these numbered actions:
- Action 1: Parse Key Statements
- Extract primary claims, numbers, dates, and conclusions directly stated in the text.
- Action 2: Filter Extraneous Commentary
- Discard rhetorical flourishes, greetings, or off-topic tangents.
- Action 3: Formulate Concise Bullets
- Write short, clear sentences.
- Restrict count to
max_bullets.
- Action 4: Verify Grounding
- Check each bullet against source text: if it cannot be pointed to, remove it.
- Action 5: Emit Summary
- Format using Section 8.
7. Decision Rules
- If the text contains contradictory statements: report both contradictory statements verbatim; do not reconcile them.
- If the text mentions probabilities or estimates: preserve qualifiers ("approximately", "alleged", "estimated"); do not convert to certainties.
8. Output Format
Emit the summary using this template:
### Factual Summary
- **Source Word Count**: [Estimated count]
- **Bullet Count**: [Count]
#### Key Factual Findings
- [Fact 1 with specific metrics or entities]
- [Fact 2]
- [Fact 3]
#### Unstated / Excluded Topics
- [Note if a commonly expected detail like dates or cause was omitted in the source]9. Verification Checklist
- Every bullet point maps directly to a sentence in the source.
- No unstated dates, names, or metrics were introduced.
- Qualifiers (e.g. "likely", "estimated") are preserved.
- Bullet count does not exceed
max_bullets.
10. Common Failures and Recovery
| Failure Symptom | Cause | Recovery Action |
| Hallucinated background | Model drawing on pretraining data | Delete any sentence not verifiable in provided source. |
| Over-generalizing | Replacing specific numbers with "several" | Keep exact numbers and metrics from text. |
11. Stop or Escalation Conditions
Stop and ask the user when:
- Source material is in an unreadable format or foreign language not requested.
12. Examples
Example 1: Successful Execution
Input:
Source: The Apollo 11 lunar module Eagle landed on the Moon on July 20, 1969, at 20:17 UTC. Neil Armstrong became the first person to step onto the lunar surface six hours and 39 minutes later on July 21 at 02:56 UTC. Buzz Aldrin joined him 19 minutes later. They spent two hours and 31 minutes exploring the site.
Max bullets: 3Output:
### Factual Summary
- **Source Word Count**: 62 words
- **Bullet Count**: 3
#### Key Factual Findings
- Apollo 11's lunar module Eagle landed on the Moon on July 20, 1969, at 20:17 UTC.
- Neil Armstrong stepped onto the lunar surface on July 21 at 02:56 UTC, followed 19 minutes later by Buzz Aldrin.
- Armstrong and Aldrin explored the landing site for two hours and 31 minutes.
#### Unstated / Excluded Topics
- Source does not mention the third astronaut (Michael Collins) or launch details.Example 2: Ambiguous or Missing Input
Input:
Source: "The meeting occurred."Output:
### Factual Summary
- **Source Word Count**: 3 words
- **Bullet Count**: 1
#### Key Factual Findings
- A meeting occurred.
#### Unstated / Excluded Topics
- Date, time, participants, location, and agenda are all omitted in the source text.Example 3: Expected Final Output
Output:
### Factual Summary
- **Source Word Count**: 85 words
- **Bullet Count**: 2
#### Key Factual Findings
- System outage began at 14:02 UTC following a configuration reload in the edge gateway.
- Traffic was rerouted to secondary region at 14:18 UTC, restoring availability to 100%.
#### Unstated / Excluded Topics
- Root cause of the edge gateway reload failure is not stated.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.

