CtrlK
BlogDocsLog inGet started
Tessl Logo

reflexion

Cross-session learning through guided dialogue. Extracts session insights and integrates into persistent memory when session knowledge should be preserved. Alias: Reflexion.

98

2.19x
Quality

Pending

Does it follow best practices?

Impact

92%

2.19x

Average score across 3 eval scenarios

SecuritybySnyk

Risky

Do not use without reviewing

SKILL.md
Quality
Evals
Security

Evaluation results

84%

74%

Automating Session Reflection for a Development Team

Phase tracking and parallel agent invocation

Criteria
Without context
With context

TodoWrite initialization

0%

90%

Phase transition updates

0%

87%

Glob session discovery

25%

100%

Memory mode detection

25%

100%

Handoff directory creation

0%

90%

Three parallel plugin agents

0%

91%

Blocking completion wait

30%

90%

Fallback to general-purpose

12%

12%

Large file chunking

25%

62%

Cleanup step

0%

87%

Correct output filenames

0%

100%

100%

36%

Writing Memory Files from a Completed Reflection Session

Memory integration: Tier A/B rules and frontmatter

Criteria
Without context
With context

Tier A path derivation

0%

100%

Tier A no YAML frontmatter

100%

100%

Tier A English only

100%

100%

Tier A topic header format

100%

100%

Tier B required frontmatter fields

100%

100%

Tier B session as array

100%

100%

Merge: date updated

0%

100%

Merge: session array appended

100%

100%

Merge: tags/keywords union

100%

100%

Domain insight placement

0%

100%

User-level Tier A unavailable

0%

100%

92%

40%

Designing the Interactive Selection Dialogue for a Reflection Session

Guided dialogue design and Recognition over Recall

Criteria
Without context
With context

AskUserQuestion for all exchanges

70%

90%

Q1 options enumerated

37%

100%

Q2 multiSelect enabled

100%

100%

Q2 includes skip option

100%

100%

Q3 inferred options

70%

60%

Q4 merge decision present

30%

100%

Q5 project-level: Tier A and Tier B

66%

100%

Q5 user-level: Tier B only

0%

100%

Q5b scope options

20%

100%

Tier guidance included

40%

70%

Repository
jongwony/epistemic-protocols
Evaluated
Agent
Claude Code
Model
Claude Sonnet 4.6

Table of Contents

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.