context-degradation

github:muratcankoylan/Agent-Skills-for-Context-Engineering · first committed 2025-12-21 · source · badge

This skill should be used when the user asks to "diagnose context problems", "fix lost-in-middle issues", "debug agent failures", "understand context poisoning", or mentions context degradation, attention patterns, context clash, context confusion, or agent performance degradation. Provides patterns for recognizing and mitigating context failures.

Skill Health: not yet scored

Listed, not yet scored. This skill has no public model test yet. It is in the scoring queue; submitted skills go first, then the most-installed.

Version history

13 commits to this file on GitHub (newest first).

2026-06-29 · commit 8fc51c7 · chore: polish repo-wide production readiness
2026-06-29 · commit 9749e5c · fix: certify production Agent Skills compatibility across Cursor, Claude Code, and Codex (#96)
2026-05-19 · commit cbc2c97 · Harden full skill corpus for v2.3.0 release
2026-05-15 · commit 0d7595d · Release v2.2.0: researcher operating system + continuous loop
2026-03-18 · commit 9cd3073 · Fix review issues from post-rewrite audit of all 13 skills
2026-03-17 · commit c847b20 · Rewrite all 13 skills from textbook to hybrid instructional voice
2025-12-31 · commit cfe1870 · fix: remove version field from SKILL.md frontmatter
2025-12-31 · commit 0b2be51 · fix: restructure plugin to match official Claude Code format
2025-12-21 · commit bf5e138 · Update AI model references to latest versions
2025-12-21 · commit cc35bb7 · Validate and enhance skills with research findings
2025-12-21 · commit def956a · Fix internal hyperlinks to point to SKILL.md files
2025-12-21 · commit 773c4a4 · Fix validation issues and improve documentation across skills collection
2025-12-21 · commit 969441a · Reorganize: move contents to root level

Archived snapshots