agentleFS
Sign inStart free
Liberty91LTDLiberty91LTD/cti-skills/skills/feedback-loops/SKILL.md

feedback-loops

Skill35 starsChanged 45 days agoMITView on GitHub

Feedback loop implementation for continuous CTI improvement. Consumer feedback, analyst retrospectives, source quality tracking.

2 other projects use a version of this.

Adopted by

Most recently taken up by

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
  1. Distinctive:

    After every significant product delivery

    line 19
  2. Distinctive:

    Not useful → Review PIR alignment, stakeholder needs, product format

    line 35
  3. Distinctive:

    After significant investigations or assessments:

    line 53
  4. Distinctive:

    Update SOPs with lessons learned

    line 63
  5. Distinctive:

    Update knowledge cells with validated/invalidated assumptions

    line 66
  6. Distinctive:

    No hits → Verify rule logic, check data source availability

    line 79
  7. Distinctive:

    All feedback feeds back into Phase 1 (Planning & Direction):

    line 83

≈ “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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Feedback Loops

Feedback is Phase 6 of the CTI Hyperloop. Without it, the cycle is open-ended and quality degrades over time.

Types of Feedback

1. Consumer Feedback

Intelligence consumers (stakeholders) assess the value of products they receive.

Collection points:

  • After every significant product delivery
  • During quarterly PIR reviews
  • During stakeholder briefings

Questions:

QuestionResponse Options
Was this intelligence useful for your decision-making?Very useful / Somewhat useful / Not useful
Was it delivered in time to act on?Yes / Partially / No
Was the format appropriate for your needs?Yes / Too technical / Too high-level / Wrong format
Was the confidence assessment helpful?Yes / Unclear / Not included
What should we prioritise next?Free text

Actions:

  • Useful + timely → Continue current approach, reinforce collection sources
  • Useful + late → Improve detection/triage speed, adjust collection frequency
  • Not useful → Review PIR alignment, stakeholder needs, product format
  • Format wrong → Adjust per stakeholder-management skill

2. Source Quality Tracking

Track the reliability of intelligence sources over time.

For each source used in products:

  • Was the information confirmed, partly confirmed, or disproven?
  • Did the source's Admiralty rating need adjustment?
  • Were there timeliness issues?

Actions:

  • Consistently confirmed → Upgrade reliability rating (e.g., C → B)
  • Inconsistent → Maintain or downgrade rating
  • Repeatedly disproven → Downgrade to D or E
  • Feed updates into source-assessment guidance

3. Analyst Retrospectives

After significant investigations or assessments:

Questions:

  • What analytical techniques were applied? Were they effective?
  • Were key assumptions validated or invalidated?
  • Were there intelligence gaps that could have been filled?
  • Were alternative hypotheses adequately considered?
  • What would you do differently?

Actions:

  • Update SOPs with lessons learned
  • Adjust technique selection guidance
  • Fill identified collection gaps
  • Update knowledge cells with validated/invalidated assumptions

4. Detection Effectiveness

For detection rules produced by the platform:

Metrics:

  • True positive rate (did the rule catch real threats?)
  • False positive rate (did it generate noise?)
  • Coverage (did it miss known instances?)

Actions:

  • High FP → Refine rule, add exclusions
  • Missed detections → Expand rule, add variants
  • No hits → Verify rule logic, check data source availability

Feedback Integration

All feedback feeds back into Phase 1 (Planning & Direction):

Consumer says "not useful" → Review and adjust PIRs
Source proves unreliable → Update Admiralty ratings, adjust collection
Analyst retrospective finds gap → Create new collection tasking
Detection rule has high FP → Refine and redeploy

Tracking

The orchestrator maintains feedback state:

  • Consumer feedback logged in stakeholder register
  • Source quality tracked in knowledge cell Sources & References
  • Retrospective outcomes logged in investigation workspace
  • Detection effectiveness tracked in detection rule metadata

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.

Check your setup, free