agentleFS
shinprshinpr/nautilus/.agents/skills/recipe-reflect/SKILL.md

recipe-reflect

Skill7 starsChanged 3 months agoMITView on GitHub

Orchestrates reflection by updating affected artifacts and distilling reusable knowledge from validation results.

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 5 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    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:

    Verify the hypothesis file has been updated with results (validation results, confidence scores, evidence)

    line 35
  2. Distinctive:

    Document a learning when the result changes the parent Opportunity or a later decision

    line 36
  3. Distinctive:

    Check if this result changes understanding of the parent Opportunity

    line 37
  4. Distinctive:

    Start from the Opportunity and load the hypothesis evidence needed to assess candidate learnings and contradictions

    line 40
  5. Distinctive:

    Write learnings to the Opportunity file's "Tier 2 Learnings" section

    line 68

≈ “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.

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

Context: Drive the feedback loop by reflecting on outcomes, updating target artifacts, and distilling learnings across the knowledge pyramid (see product-principles skill for Tier definitions).

Required Skills [LOAD BEFORE EXECUTION]

  1. [LOAD IF NOT ACTIVE] product-principles — Knowledge Pyramid and promotion criteria
  2. [LOAD IF NOT ACTIVE] hypothesis-discipline — validation results, lifecycle status, and confidence changes

Delegate Level 2 and Level 3 distillation to knowledge-distiller for independent evidence synthesis.

Execution Decision Flow

1. Scope Assessment

Input: Use the path or text supplied with the explicit skill invocation. If no input was supplied and the target cannot be inferred unambiguously, ask for it.

Determine the reflection level (see references/reflection-guide.md):

TriggerLevelTarget Files
Hypothesis concludedLevel 1: HypothesisThe hypothesis file
Multiple hypotheses concluded under an OpportunityLevel 2: OpportunityOpportunity file (Tier 2 Learnings section)
PRD delivered, quarterly review, strategic pivotLevel 3: Visiondocs/product/vision.md, docs/product/learnings.md

2. Result Recording

Level 1: Hypothesis Reflection

  1. Verify the hypothesis file has been updated with results (validation results, confidence scores, evidence)
  2. Document a learning when the result changes the parent Opportunity or a later decision
  3. Check if this result changes understanding of the parent Opportunity

Level 2: Opportunity Reflection

  1. Start from the Opportunity and load the hypothesis evidence needed to assess candidate learnings and contradictions
  2. Prepare context for knowledge-distiller (hypothesis summaries, results, confidence changes)

Level 3: Vision Reflection

  1. Gather the cross-Opportunity evidence needed for the outcome, NSM, or Tier 1 decision
  2. Review Product Outcomes — are targets still correct?
  3. Review NSM — still the right connecting metric?
  4. Prepare context for knowledge-distiller

3. Knowledge Distillation

Invoke knowledge-distiller for evidence distillation:

  • knowledge-distiller operates in a separate context to avoid individual hypothesis bias
  • It distills evidence into candidate learnings and contradictions
  • It proposes Tier 2 learnings (for Opportunity) or Tier 1 learnings (for Vision)
  • It enforces distillation quality criteria (per product-principles skill)

Present distillation results to the user for the knowledge-promotion decision:

  • Candidate learnings with proposed tiers, supporting evidence, and contexts
  • Contradictions found and the decisions they may affect

End the current turn with the distillation result as the workflow output. Promote knowledge only after the user confirms the proposal in a later turn.

4. Knowledge Promotion

After that confirmation:

Tier 3 → Tier 2

  • Write learnings to the Opportunity file's "Tier 2 Learnings" section
  • Include hypothesis references that support each learning

Tier 2 → Tier 1

  • Write to docs/product/learnings.md
  • Include supporting evidence references and their dates
  • State the contexts where the learning applies

Sub-agent Usage

AgentWhenWhy (context separation benefit)
knowledge-distillerLevel 2 and Level 3 reflectionIndependent evidence synthesis across hypotheses

Scope Boundaries

Included: Result recording, knowledge distillation, and Tier promotion Not included: Hypothesis validation (→ recipe-validate), new hypothesis generation (→ recipe-discover)

Completion

The workflow is complete when approved promotions are updated, or a no-change result is reported with the evidence and contradictions relevant to the decision.

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.